text
stringlengths 87
880k
| pmid
stringlengths 1
8
| accession_id
stringlengths 9
10
| license
stringclasses 2
values | last_updated
stringlengths 19
19
| retracted
stringclasses 2
values | citation
stringlengths 22
94
| decoded_as
stringclasses 2
values | journal
stringlengths 3
48
| year
int32 1.95k
2.02k
| doi
stringlengths 3
61
| oa_subset
stringclasses 1
value |
|---|---|---|---|---|---|---|---|---|---|---|---|
==== Front
Chiropr OsteopatChiropractic & Osteopathy1746-1340BioMed Central London 1746-1340-13-121608078710.1186/1746-1340-13-12ReviewAnatomic and functional leg-length inequality: A review and recommendation for clinical decision-making. Part II, the functional or unloaded leg-length asymmetry Knutson Gary A [email protected] 840 W. 17th, Suite 5, Bloomington, IN, 47404, USA2005 20 7 2005 13 12 12 31 5 2005 20 7 2005 Copyright © 2005 Knutson; licensee BioMed Central Ltd.2005Knutson; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Part II of this review examines the functional "short leg" or unloaded leg length alignment asymmetry, including the relationship between an anatomic and functional leg-length inequality. Based on the reviewed evidence, an outline for clinical decision making regarding functional and anatomic leg-length inequality will be provided.
Methods
Online databases: Medline, CINAHL and Mantis. Plus library searches for the time frame of 1970–2005 were done using the term "leg-length inequality".
Results and Discussion
The evidence suggests that an unloaded leg-length asymmetry is a different phenomenon than an anatomic leg-length inequality, and may be due to suprapelvic muscle hypertonicity. Anatomic leg-length inequality and unloaded functional or leg-length alignment asymmetry may interact in a loaded (standing) posture, but not in an unloaded (prone/supine) posture.
Conclusion
The unloaded, functional leg-length alignment asymmetry is a likely phenomenon, although more research regarding reliability of the measurement procedure and validity relative to spinal dysfunction is needed. Functional leg-length alignment asymmetry should be eliminated before any necessary treatment of anatomic LLI.
Leg-length inequalityfunctionallow back pain
==== Body
Review
In Part I of this review, the literature regarding the prevalence, magnitude, effects and clinical significance of anatomic leg-length inequality (LLI) was examined. Using data on leg-length inequality obtained by accurate and reliable x-ray methods, the prevalence of anatomic inequality was found to be 90%; the mean was 5.2 mm (SD 4.1). The evidence suggested that, for most people, anatomic leg-length inequality is not clinically significant until the magnitude reaches ~20 mm (~3/4"). The phenomenon of the functional "short leg" will be considered in Part II of this review. The objective is to define functional "short leg", how it differs from anatomic LLI and explore any association with neuromuscular dysfunction. In addition we will review the apparent efficacy of heel lifts in some cases of mild anatomic LLI, plus muscular reactions to, and causes of, pelvic torsion.
The functional short leg, or unloaded leg-length alignment asymmetry
The functional short leg, or unloaded leg-length alignment asymmetry (hereafter abbreviated as LLAA) is itself a phenomenon much discussed and little understood. Essentially, when a subject lies prone or supine, unloading the pelvis, the feet are examined, most often at the welt (heel-sole interface), for the presence of a "short leg" or alignment asymmetry. Some hold the opinion that anatomic LLI can be measured in this way [1]. The examination for unloaded leg-length alignment asymmetry as a sign of "neuromuscular dysfunction" is a clinical test commonly used by chiropractors [2,3]. Given the frequent use of this test as an indicator of a functional problem, it is important to know whether the unloaded leg check test is an indicator of an anatomic short leg, or whether the test is reliable and valid as an instrument to measure functional "short leg" and whether LLAA findings are contaminated by anatomic LLI.
Anatomic LLI is caused by a natural developmental asymmetry or a variety of other factors, including fracture, disease, and complications of hip replacement surgery. Given the long-term loading, the lumbopelvic structure may be expected to adapt via Heuter Volkmanns' law [4] and soft tissue changes [4,5], establishing the compensated structural changes as "normal". This adaptive response is seen in the change of lumbosacral facet angles noted by Giles [6]. A case study followed the effect of anatomic LLI caused by hip replacement surgery on subjective symptoms, unloaded LLAA checks and pelvic unleveling, reporting that adaptive changes occurred over a period of several months [7].
Using a device to measure standing pelvic crest unleveling, Petrone et al found excellent intra and inter-examiner reliability, and validity (ICC, 0.89–0.90) relative to anatomic leg length inequality determined by x-ray measurement in asymptomatic subjects [8]. However, the correlation between the pelvic level and femoral head heights was "substantially lower" in a low back pain group. This indicates that some sort of functional pelvic tilt or torsion was present in the low back pain population that was unrelated to their anatomic LLI. While the decreased correlation between pelvic tilt and LLI in the back pain group was not examined relative to a functional short leg, the connection between back pain and the biomechanically unusual pelvic torsion stands out.
Lumbar lateral flexion was studied in a group of subjects 10 years after LLI caused by femoral fracture that occurred after they were skeletally mature [9]. Despite the compensatory lumbar scoliosis, these subjects had symmetrical lumbar lateral flexion, prompting the authors to comment that the "...acquired leg-length discrepancy produced little permanent structural abnormality in the lumbar spine..." [9]. Significant anatomic LLI acquired after skeletal maturity does not result in adaptive structural changes within a 10-year period.
However, another study from the same orthopedic center looked at the effects of significant (mean 3 cm) LLI acquired prior to skeletal maturity [10] in now mature subjects (17–38 years old, mean 28). In this group, there was considerable asymmetry of lumbar lateral flexion after placing a lift under the short leg to level the pelvis. This indicates that the body had permanently compensated to the structural changes in the spine/pelvis.
This type of permanent compensation to pre skeletal maturity LLI was also found in subjects with pelvic unleveling. Young et al [11] found that placing a lift under the foot of a subject with no pelvic unleveling resulted in greater lumbar lateral flexion towards the now high iliac crest side. In subjects with pelvic unleveling, when the lift was put under the foot on the side of the low iliac crest in order to level the crest, lateral flexion was increased towards the formerly low crest side. If the body remodels and adapts to the pelvic unleveling/torsion caused by anatomic LLI, then by putting a lift under the side of the "low" iliac crest, one is actually raising what the body has adapted to as level. In other words, the unlevel pelvis of those with anatomic LLI has been adapted to and is now "normal", and putting a lift under the low side has the same effect as putting a lift under the leg of an even pelvis (Figure 1).
Figure 1 Effects of a lift in level and unlevel compensated pelvis.
These two studies [10,11] provide evidence that in pre-skeletal maturity subjects, LLI and pelvic torsion – which describe the vast majority of LLI – adaptive changes take place in the muscles, ligaments, joints and bones to compensate for the imposed asymmetry. Because these adaptive compensations to the LLI have become anatomic, they are not likely to change as the body moves from a loaded (standing) to an unloaded (supine, prone) position. The nervous system also appears to compensate as demonstrated in the study by Murrell et al [12] in which there was no loss of stability in subjects with LLI, prompting them to point to "long-term adaptation by the neuromuscular system".
The persistence of pelvic torsion in subjects with anatomic LLI is supported by Klein [13] who found that such distortion remained in both standing and sitting positions. That pelvic torsion persists with the subjects' weight off the femoral heads indicates such torsion has been incorporated into the joints as the normal position. Rhodes et al demonstrated that the side and magnitude of prone and especially supine "short legs" were not significantly correlated with radiographic anatomic LLI, indicating they are separate phenomena [14].
The studies noted above provide indirect evidence that the pelvic torsion associated with childhood-onset anatomic leg-length inequality is adapted for and incorporated as normal. It follows then, that when an average person with an anatomic LLI and structurally compensatory pelvic torsion moves from a loaded (standing) to an unloaded (prone/supine) position, the torsion of the pelvis remains intact and the leg length at the feet/shoes would appear "even" on a visual check. The pelvis – joints, ligaments and muscles – have adapted to the anatomic LLI, making any torsion structural. It is this putative biomechanical adaptation that makes unloaded leg-length alignment asymmetry tests – the functional "short leg" tests – unreliable as a measure of anatomic LLI [14].
Unloaded LLAA is suspected to result from hypertonicity of suprapelvic muscles [15-17]. In a study of subjects with and without supine LLAA, Knutson & Owens found those with LLAA had significantly decreased endurance times for the erector (Biering-Sorensen test) and quadratus lumborum muscles [18]. Further, the side of LLAA significantly correlated with the side of the QL muscle quickest to fatigue. One of the causes of increased susceptibility of muscles to fatigue is hypertonicity. These results stand in contrast to Mincer et al [19] who suspected altered muscle fatigue profiles with anatomic leg-length inequality, but did not find such, providing further evidence that LLAA is a pathological process distinct from LLI.
When standing, the actions of the QL depend on whether the spine or the pelvis is stabilized. If the pelvis is stabilized, QL contraction laterally flexes and extends the spine [1,20,21]. With the spine stable, QL contraction pulls cephalically through its attachment to the posterior aspect of the hemipelvis [1,21]. This load on the posterior aspect of the iliac crest could act to rotate the ipsilateral anterior hemipelvis lower – an AS ilium – causing the pelvis to torque and having the opposite effect on the contralateral hemipelvis – a PI ilium. The degree of torsion (if any) would be dependent on the tension in the QL and the freedom of movement of the pelvis, and any pre-existing pelvic torsion due to anatomic LLI. However, if the subject now adopts an unloaded posture – supine or prone – QL hypertonicity is freed from the load of the body and able to lift the ipsilateral hemipelvis, hip and leg in the cephalic direction, producing leg-length alignment asymmetry at the feet. This model is in agreement with Travell and Simons who write, "In recumbancy, active TrPs [trigger points] shorten the [quadratus lumborum] muscle and can thus distort pelvic alignment, elevating the pelvis on the side of the tense muscle" [1].
Clinical considerations
Now we can return to the dilemma of how lifts may have a positive effect on back pain and muscle activity given that most anatomic LLI is not clinically significant. Torsion of the pelvis as an adaptive structural compensation in anatomic LLI has been shown to be limited. If a person has pelvic torsion due to anatomic LLI near the limits of the body's ability to adapt, and QL hypertonicity with its ability to cause pelvic torsion is superimposed, muscular bracing reactions and pain could be the result. Indahl et al [22] found that stimulation of the sacroiliac joint capsule (in pigs) caused reflexive muscular responses, depending on what area of the joint (dorsal/ventral) was stimulated. They note that, "Irritation of low threshold nerve endings in the sacroiliac joint tissue may trigger a reflex activation of the gluteal and paraspinal muscles that become painful over time". Interestingly, stimulation of the ventral area of the SI joint produced reflexive contraction of the quadratus lumborum. It may be that a positive feedback loop could be established where QL hypertonicity leads to lumbar curvature and pelvic torsion which stimulates the SI joint leading to more QL hypertonicity, more lumbar curvature and pelvic torsion. It will be interesting to see if a similar muscular reflex to SI stimulation is found in humans.
Based on their research, Allum et al [23] proposed that rotation of the trunk excites joint receptors in the lumbar spine triggering muscular contractions – paraspinal muscles – for balance correction. While these receptors likely have adapted to any pelvic/lumbar rotation caused by anatomic LLI, further pelvic torsion caused by QL hypertonicity may stimulate the balance receptors causing reflexive muscular contraction. A lift would reduce the pelvic torsion and lower the proprioceptive balance triggers below threshold, eliminating chronic, painful muscular contraction.
In a case of additive effects of anatomic LLI and QL/suprapelvic hypertonicity on pelvic torsion, a lift used to level the pelvis would take the strain off the sacroiliac and associated joints and ligaments and decrease potentially painful muscular bracing. Thus, lifts can work to decrease back pain in people with what seem to be clinically insignificant amounts of anatomic leg-length inequality. Of course, it would be important for the clinician to explore reasons for any quadratus lumborum and other suprapelvic muscle hypertonicity and eliminate them to provide a complete correction. On the other hand, pure anatomic LLI in the range of and above 20 mm – the upward limit for adaptive compensation – may stimulate sacroiliac and/or lumbar proprioceptors causing reflexive and ultimately painful muscular contractions that will only be relieved by a lift to level the pelvis.
Reliable detection of LLI and LLAA is difficult, but not impossible. Research has shown the examination procedures for putative LLAA both prone [24] and supine [25] to have intra- and inter-examiner reliability. In a controlled setting, Cooperstein et al investigated the accuracy of a compressive prone leg check in subjects with proscribed amounts of artificial LLI [26]. They found the procedure to be highly accurate – able to detect a difference in leg length magnitudes as little as +/- 1.87 mm, and noted that, "...compressive leg checking would be expected to identify the short or shortened leg side, irrespective of magnitude, 95.4% of the time". In this authors opinion, while it is necessary to be able to detect a functional asymmetry above a baseline amount, the LLAA is more of a go/no-go test relative to a clinical decision. As such, accuracy in magnitude is not critically important past that lower limit amount. In other words, as an example, clinicians would only have to agree that an asymmetry above 1/8" exists, and not whether the asymmetry is 1/2" versus 3/16". Studies designed to examine intra- and inter-examiner reliability should keep this in mind.
In addition to reliability, the leg check procedure outlined by Cooperstein et al demonstrated concurrent validity as assessed against artificial LLI [26]. However, as noted by the authors, the clinical relevance of the procedure is unknown. Other studies in a clinical environment have demonstrated validity of the supine procedure by correlating LLAA to increased rated pain (VAS) intensity and recurrent back pain [27], lower SF-12 general health scores [28] and altered supra-pelvic muscle function [18].
Once any suprapelvic muscle hypertonicity has been relieved – and the causes may be multiple, including upper cervical joint dysfunction [18,29-33] – the effect of anatomic LLI can be investigated. This treatment sequence – removal of suprapelvic muscle hypertonicity causing LLAA prior to investigating anatomic LLI – is also recommended by others [1,34]. Patient history (activities that involve prolonged, repetitive loading) and symptomatic presentation should arouse suspicion regarding a clinically significant anatomic LLI. The most accurate method to determine anatomic LLI is the A-P lumbopelvic x-ray with the central ray at the height of the femoral heads. If x-ray is undesirable, tape measure from the ASIS to the medial malleolus, while unreliable for LLI in amounts less than 10 mm [35], may be accurate enough with larger asymmetries if the average of two determinations are calculated [36]. Using a succession of blocks of known thickness under the leg ipsilateral to the low iliac crest in order to level the pelvis also may aid in determining the amount of lift necessary [37,38]. Both of the non-radiographic methods are questionable regarding accuracy and reliability; however, anatomic LLI is not likely to become clinically significant at much less than 20 mm (~3/4"), and this level of asymmetry may be found with greater reliability. If anatomic LLI is determined to be clinically significant, a lift may be indicated. Danbert [39] reviews the proper application of lifts, should they be necessary.
Conclusion
Anatomic leg-length inequality under 20 mm and leg-length alignment asymmetry caused by supra-pelvic muscle hypertonicity may interact in a loaded (standing) posture, but not in an unloaded (prone/supine) posture. Any leg-length alignment asymmetry due to suprapelvic muscular hypertonicity should be eliminated before any necessary treatment of anatomic leg-length inequality. By using this information, which is open to change based on new studies, the clinician may better understand the diverse and sometimes confusing findings relative to anatomic leg-length inequality and functional or unloaded leg-length alignment asymmetry, and be better able to make treatment recommendations.
Competing interests
The author(s) declare that they have no competing interests.
==== Refs
Travell JG Simons DG Chapter 4, Quadratus Lumborum Muscle Myofascial Pain and Dysfunction The Trigger Point Manual The Lower Extremities 1999 2 2 Williams & Wilkens: Baltimore 42,35,35,42,107
Mannello DM Leg Length Inequality J Manipulative Physiol Ther 1992 15 576 590 1469342
Walker BF Buchbinder R Most commonly used methods of detecting subluxation and the preferred term for its description: a survey of chiropractors in Victoria, Australia J Manipulative Physiol Ther 1997 20 583 588 9436143
White A Panjabi M Clinical Biomechanics of the Spine 1987 Philadelphia: J.B. Lippincott 96,352
Gossman MR Sahrmann SA Rose SJ Review of length-associated changes in muscle Physical Therapy 1982 62 1799 1807 6755499
Giles LGF Taylor JR Lumbar spine structural changes associated with leg length inequality Spine 1982 7 159 162 7089692
Knutson G The supine leg check as a determinant of physiological/postural leg length inequality; a case study and analysis Chiropr Res J 2000 VII 8 13
Petrone MR Guinn J Reddin A Sutlive TG Flynn TW Garber WP The accuracy of the palpation meter (PALM) for measuring pelvic crest height difference and leg length discrepancy J Orthop Sports Phys Ther 2003 33 319 25 12839206
Gibson PH Papaioannou T Kenwright J The influence on the spine of leg-length discrepancy after femoral fracture J Bone Joint Surg (Br) 1983 65 584 7 6643562
Papaioannou T Stokes I Kenwright J Scoliosis associated with limb-length inequality J Bone Joint Surg 1982 64-A 59 62 7054204
Young RS Andrew PD Cummings GS Effect of simulating leg length inequality on pelvic torsion and trunk mobility Gait Posture 2000 11 217 23 10802434 10.1016/S0966-6362(00)00048-5
Murrell P Cornwall MW Doucet SK Leg-length discrepancy: effect on the amplitude of postural sway Arch Phys Med Rehabil 1991 72 646 8 1859258
Klein KK Redler I Lowman CL Asymmetries of growth in the pelvis and legs of children: A clinical and statistical study 1964–1967 J Am Osteopath Assoc 1968 68 105 108
Rhodes DW Mansfield ER Bishop PA Smith JF Comparison of leg length inequality measurement methods as estimators of the femur head height difference in standing x-ray J Manipulative Physiol Ther 1995 18 448 452 8568426
Cooperstein R Lisi A Pelvic torsion: anatomic considerations, construct validity, and chiropractic examination procedures Top Clin Chiro 2000 7 38 49
Gossman MR Sahrmann SA Rose SJ Review of length-associated changes in muscle Physical Therapy 1982 62 1799 1807 6755499
Grostic JD Dentate ligament – cord distortion hypothesis Chiropr Res J 1988 1 47 55
Knutson G Owens E Erector and quadratus lumborum muscle endurance tests and supine leg-length alignment asymmetry: An observational study (accepted for publication, JMPT 12-03).
Mincer AE Cummings GS Andrew PD Rau JL Effect of leg length discrepancy on trunk muscle fatigue and unintended trunk movement J Phys Ther Sci 1997 9 1 6 10.1589/jpts.9.1
McGill SM Childs A Liebenson C Endurance times for low back stabilization exercises: clinical targets for testing and training from a normal database Arch Phys Med Rehabil 1999 80 941 944 10453772 10.1016/S0003-9993(99)90087-4
Andersson EA Oddsson LI Grundstrom H Nilsson J Thorstensson A EMG activities of the quadratus lumborum and erector spinae muscles during flexion-relaxation and other motor tasks Clin Biomech 1996 11 392 400 10.1016/0268-0033(96)00033-2
Indahl A Kaigle A Reikeras O Holm S Sacroiliac joint involvement in activation of the porcine spinal and gluteal musculature J Spinal Disord 1999 12 325 30 10451049
Allum JHJ Honnegger F Interactions between vestibular and proprioceptive inputs triggering and modulating human balance-correcting responses differ across muscles Exp Brain Res 1998 121 478 494 9746156 10.1007/s002210050484
Nguyen HT Resnick DN Caldwell SG Elston EW Bishop BB Steinhouser JB Gimmillaro TJ Keating JC Inter-examiner reliability of Activator methods relative to leg length evaluation in the prone, extended position J Manipulative Physiol Ther 1999 22 565 9 10626698
Hinson R Brown SH Supine leg length differential estimation: an inter- and intra-examiner reliability study Chiropr Res J 1998 5 17 22
Cooperstein R Morschhauser E Lisi A Nick TG Validity of compressive leg checking in measuring artificial leg-length inequality J Manipulative Physiol Ther 2003 26 557 66 14673405 10.1016/j.jmpt.2003.08.002
Knutson G Incidence of foot rotation, pelvic crest unleveling, and supine leg length alignment asymmetry, and their relationship to self-reported back pain J Manipulative Physiol Ther 2002 24 e1 10.1067/mmt.2002.121414
Knutson G Owens E Leg length Alignment Asymmetry in a Non-clinical Population and its Correlation to a Decrease in General Health as Measured by the SF-12: A Pilot Study Journal of Vertebral Subluxation Research 2004 1 1 5
Pollard H Ward G The effect of upper cervical or sacroiliac manipulation on hip flexion range of motion J Manipulative Physiol Ther 1998 21 611 616 9868632
Nansel DD Waldorf T Cooperstein R Effect of cervical spinal adjustments on lumbar paraspinal muscle tone: Evidence for facilitation of intersegmental tonic neck reflexes J Manipulative Physiol Ther 1993 16 91 95 8445359
Seemann DE Bilateral weight differential and functional short leg: an analysis of pre and post data after reduction of atlas subluxation Chiropr Res J 1993 2 33 8
Seemann DE Anatometer measurements: a field study intra- and inter-examiner reliability and pre to post changes following an atlas adjustment Chiropr Res J 1999 VI 7 9
Kondziella W Clinical and functional diagnosis and treatment of low-back pain from pelvic malposition Schmerz 1996 10 204 10 (article in German). 12799854 10.1007/s004820050041
McCaw ST Leg Length Inequality. Implications for running injury prevention Sports Medicine 1992 14 422 429 1470794
Friberg O Clinical symptoms and biomechanics of lumbar spine and hip joint in leg length inequality Spine 1983 8 643 651 6228021
Beattie P Isaacson K Riddle DL Rothstein JM Validity of derived measurements of leg-length differences obtained by use of a tape measure Phys Ther 1990 70 150 7 2304973
Hanada E Kirby RL Mitchell M Swuste JM Measuring leg-length discrepancy by the "iliac crest palpation and book correction" method: Reliability and validity Arch Phys Med Rehabil 2001 82 938 42 11441382 10.1053/apmr.2001.22622
Aspegren DD Cox JM Trier KK Short leg correction: A clinical trial of radiographic vs non-radiographic procedures J Manipulative Physiol Ther 1987 10 232 238 3694061
Danbert RJ Clinical assessment and treatment of leg length inequalities J Manipulative Physiol Ther 1988 11 290 295 3049890
|
16080787
|
PMC1198238
|
CC BY
|
2021-01-04 16:38:23
|
no
|
Chiropr Osteopat. 2005 Jul 20; 13:12
|
utf-8
|
Chiropr Osteopat
| 2,005 |
10.1186/1746-1340-13-12
|
oa_comm
|
==== Front
Chiropr OsteopatChiropractic & Osteopathy1746-1340BioMed Central London 1746-1340-13-161609113410.1186/1746-1340-13-16ReviewReflex control of the spine and posture: a review of the literature from a chiropractic perspective Morningstar Mark W [email protected] Burl R [email protected] Heidi [email protected] Mark [email protected] Trevor V [email protected] Director of Research; The Pettibon Institute, 3416-A 57 St Ct NW Gig Harbor, WA 98335, USA; Private practice of chiropractic, 10683 S Saginaw St, Suite B, Grand Blanc, MI 48439, USA2 Executive Director; The Pettibon Institute, 3416-A 57 St Ct NW Gig Harbor, WA 98335, USA3 Doctor of Chiropractic Candidate; Palmer College of Chiropractic. 1000 Brady St Davenport, IA 52803, USA4 Board of Trustees; Palmer College of Chiropractic. 1000 Brady St Davenport, IA 52803, USA2005 9 8 2005 13 16 16 28 4 2005 9 8 2005 Copyright © 2005 Morningstar et al; licensee BioMed Central Ltd.2005Morningstar et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Objective
This review details the anatomy and interactions of the postural and somatosensory reflexes. We attempt to identify the important role the nervous system plays in maintaining reflex control of the spine and posture. We also review, illustrate, and discuss how the human vertebral column develops, functions, and adapts to Earth's gravity in an upright position. We identify functional characteristics of the postural reflexes by reporting previous observations of subjects during periods of microgravity or weightlessness.
Background
Historically, chiropractic has centered around the concept that the nervous system controls and regulates all other bodily systems; and that disruption to normal nervous system function can contribute to a wide variety of common ailments. Surprisingly, the chiropractic literature has paid relatively little attention to the importance of neurological regulation of static upright human posture. With so much information available on how posture may affect health and function, we felt it important to review the neuroanatomical structures and pathways responsible for maintaining the spine and posture. Maintenance of static upright posture is regulated by the nervous system through the various postural reflexes. Hence, from a chiropractic standpoint, it is clinically beneficial to understand how the individual postural reflexes work, as it may explain some of the clinical presentations seen in chiropractic practice.
Method
We performed a manual search for available relevant textbooks, and a computer search of the MEDLINE, MANTIS, and Index to Chiropractic Literature databases from 1970 to present, using the following key words and phrases: "posture," "ocular," "vestibular," "cervical facet joint," "afferent," "vestibulocollic," "cervicocollic," "postural reflexes," "spaceflight," "microgravity," "weightlessness," "gravity," "posture," and "postural." Studies were selected if they specifically tested any or all of the postural reflexes either in Earth's gravity or in microgravitational environments. Studies testing the function of each postural component, as well as those discussing postural reflex interactions, were also included in this review.
Discussion
It is quite apparent from the indexed literature we searched that posture is largely maintained by reflexive, involuntary control. While reflexive components for postural control are found in skin and joint receptors, somatic graviceptors, and baroreceptors throughout the body, much of the reflexive postural control mechanisms are housed, or occur, within the head and neck region primarily. We suggest that the postural reflexes may function in a hierarchical fashion. This hierarchy may well be based on the gravity-dependent or gravity-independent nature of each postural reflex. Some or all of these postural reflexes may contribute to the development of a postural body scheme, a conceptual internal representation of the external environment under normal gravity. This model may be the framework through which the postural reflexes anticipate and adapt to new gravitational environments.
Conclusion
Visual and vestibular input, as well as joint and soft tissue mechanoreceptors, are major players in the regulation of static upright posture. Each of these input sources detects and responds to specific types of postural stimulus and perturbations, and each region has specific pathways by which it communicates with other postural reflexes, as well as higher central nervous system structures. This review of the postural reflex structures and mechanisms adds to the growing body of posture rehabilitation literature relating specifically to chiropractic treatment. Chiropractic interest in these reflexes may enhance the ability of chiropractic physicians to treat and correct global spine and posture disorders. With the knowledge and understanding of these postural reflexes, chiropractors can evaluate spinal configurations not only from a segmental perspective, but can also determine how spinal dysfunction may be the ultimate consequence of maintaining an upright posture in the presence of other postural deficits. These perspectives need to be explored in more detail.
Cervical spinePostureReflex
==== Body
Background
Historically, chiropractic has centered around the concept that the nervous system controls and coordinates all other systems within the human body [1,2]. Recent evidence has provided insight into the mechanisms responsible for this neurological governance of other body systems [3-9]. Perhaps the most important relationship from a chiropractic perspective, however, is that between the nervous and musculoskeletal systems. Specifically, many chiropractors believe that "subluxations" of the vertebral column somehow compromise the integrity and function of the nervous system, which may ultimately affect health and vitality [10]. However, to date, research attempting to identify the exact parameters of the chiropractic subluxation remains tenuous [11,12].
More recently, certain authors [13,14] have discussed an alternative concept of neurological dysfunction. Two virtually synonymous concepts, dysafferentation [13] and the wind-up phenomenon [14], are based on the premise that neurological dysfunction is caused by a constant barrage of afferent input into the nervous system, causing a hypersensitive state within the neuronal receptor pool. These receptor pools, made largely of interneurons, allow sensory input to be conveyed to higher spinal and cortical centers, while simultaneously providing the means for spinal reflexive control of various functions [13,15]. Neurologic dysfunction caused by afferent stimulation may be related to certain types of headache [14], joint dysfunction, and muscular restriction [13].
The chiropractic interest in static global spinal structure and its correction is growing [16-27]. Most of this research has only surfaced within the last 10 years. Much of this research is focused upon the inherent biomechanics of the vertebral column. Research in the areas of spinal modeling [16,17,24,26,27] and posture analysis [21] have attempted to provide a clinically valid outcome measure for the treatment of posture-related symptoms and pathologies. For example, Wiegand et al [27] demonstrated a correlation between certain cervical spinal configurations and the presence of pathology. Harrison et al [17,24,26] reported average ranges of the sagittal spine curves for 3 sets of asymptomatic populations. This type of biomechanical modeling is important for developing parameters by which outcome assessments can be created and implemented. Unfortunately, spinal modeling cannot account for the host of mechanisms and precipitating factors that promote the divergence of the spine away from these established biomechanical models. However, these concepts and models do not account for, or acknowledge, the importance of the neurological, reflexive control of posture. Rather than simply identifying that a given patient does not fit into a normal spinal model, further investigation into why that particular patient does not fit is perhaps more important in terms of developing patient management strategies. This is important not only for understanding why abnormal spinal configurations occur, but to also discuss the potential to recruit these same neurological pathways to aid in the correction of spinal or postural abnormalities.
Postural reflexes can be subcategorized as the following: visual righting reflexes, labyrinthine righting reflexes, neck righting reflexes, body on head righting reflexes, and body on body righting reflexes [28]. Although some of the reflexes and neuroanatomy have been defined and illustrated separately, these collective reflexes and their interactions have not been examined from a chiropractic perspective. Since conservative postural treatment is becoming increasingly investigated, knowledge of the postural reflexes will only aid the practitioner in providing treatment consistent with foundational postural neurophysiology. In our review, we will illustrate the mechanisms by which the nervous system controls and coordinates posture, with special emphasis placed on how the nervous system adapts to specific external environmental factors. This review will detail the neurological control of posture, specifically the afferent regulation of posture. We will illustrate the neuroanatomy involved in afferent postural control, giving most attention to those reflexes associated with the cervical spine and special senses. We also discuss the interactions between the various afferent structures and their postural effects.
The primary purpose of the postural reflexes is to maintain a constant posture in relation to a dynamic external environment. This review will discuss the main external environmental parameter by which these reflexes maintain and adapt postural control: gravity. Because earth's gravitational field is a constant, the postural reflexes develop and react to this constant. From the moment an infant learns to first hold its head up through the time the child begins to walk upright, these postural reflexes are essentially supervising spinal structural and functional development in direct response to the constant force of gravity. To allow for a balance of strength and flexibility, the spine develops natural sagittal curves that provide functional lever arms for muscular attachment and efficient movement. Again, all of this is achieved using the constant of gravity as the main reference point, and the postural reflexes serve as the neuromotor impetus for this adaptive response.
This review will also detail the mechanisms that cause the reactive musculoskeletal changes in response to sudden changes in the external environment. Primarily, we will illustrate and compare the effects of gravitational changes upon the cervical spine postural reflexes and resultant postural adaptations. Specifically, details of postural adaptation, musculoskeletal morphological changes, and clinical symptoms in microgravitational environments will be outlined and discussed.
Methods
Starting from the year 1970, we searched the MEDLINE database using the following key words and phrases: "posture," "ocular," "vestibular," "cervical facet joint," "afferent," "vestibulocollic," "cervicocollic," and "postural reflexes," "spaceflight," "microgravity," "weightlessness," "gravity," and "postural." Searches of the MANTIS database and the Index to Chiropractic Literature using the same key word were also performed. Nearly all of the articles relating to our review were also found on MEDLINE. A hand search of our personal libraries was also conducted, retrieving textbooks pertaining to this topic. For purposes of this review, we included original research articles, review papers, case series, or textbook chapters outlining the anatomy, physiology, evaluation, or pathophysiology and interaction of vision, the vestibular system, the vertebral column, or a combination of these. This review was organized so that a brief review of each structure could be discussed both individually and collectively. Although these databases house a vast multitude of articles on posture, only those specifically pertaining to neurological or neuromuscular control were included.
Visual Input
The visual pathway consists of the following parts: the optic nerve, optic chiasm, and the optic tracts which project to three subcortical areas known as the pretectum, the superior colliculus, and the lateral geniculate body. Information relayed by this pathway ascends from the optic nerve ultimately to the lateral geniculate body, with axons projecting to the primary visual cortex [29]. The primary visual cortex is located on the medial surface of the occipital lobe in the walls of the calcarine sulcus. [29]
The visual field and pathway are important regulators of postural control. Visual input for postural control helps to fixate the position of the head and upper trunk in space, primarily so that the center of mass of the trunk maintains balance over the well-defined limits of foot support [24]. Many studies have shown the destabilizing effects on postural regulation when the visual field is altered due to injury, disease, or congenital abnormality [31-38]. Guerraz et al [34] studied 21 patients diagnosed with visual vertigo. They found that subjecting these patients to disorienting visual environments markedly reduced postural control. Catanzariti et al [31] identified a correlation between the severity of postural deformity in scoliosis patients who present with visual disorders.
It is well known that vision has a major role in the regulation of upright posture, particularly by maintaining head position in space. Alterations in head posture may develop secondarily to visual changes. For example, Havertape and Cruz [35] showed how the addition of eyeglasses changed the head position in 5 patients with a chin-down posture as a result of high hyperopia. Likewise, Willford et al [39] showed that people who wear prescription multifocal lenses tend to exaggerate a forward head posture to utilize the proper area of the lense, depending upon the functional needs of the moment. This has important implications for posture rehabilitation and will be discussed in detail in this review. In a study of 125 patients with congenital nystagmus, Stevens and Hertle [38] found that those patients who assumed a compensatory abnormal head posture achieved better visual acuity than those who failed to adapt to the presence of the nystagmus. In 5 patients with unilateral vision loss due to cyclotropia or monocular nystagmus, Nucci and Rosenbaum [36] found that a compensatory head tilt or rotation could be reduced by surgical correction of the ocular disorder. Pyykko et al [37] conducted a study on 10 patients with Usher's syndrome and 10 patients with blindness. All 20 patients displayed a statistically significantly higher postural sway than the control group. It is noteworthy to point out that visual information relayed to higher centers is based upon relative information. Although postural control is highly dependent upon visual status, higher cortical functions are necessary to differentiate between a fixed person within a moving environment, or a moving person within a fixed environment. Buchanan et al [30] demonstrated how the central nervous system might actively suppress visual information that is inconsistent with afferent postural control input from other sources, such as the somatosensory system.
While vision is an important part of postural control, the information it relays to higher cortical areas remains based on relative perception. Postural corrections initiated by the visual system are made in the direction of visual stimulus [40]. Afferent stimulus provided by the visual field can include either movement of the environment around the person, or movement of the person in the environment [33]. As Guerraz et al [33] and DiZio et al [41] have pointed out, small changes in the visual environment can alter visually based posture control, such as darkness or changes below the conscious threshold. However, visual control of posture in real time does not receive much contribution from higher-level processes [42]. As infants learn to assume a sitting position, much of this postural development relies upon input from the visual environment. As the child repeats a sitting task, a visuomotor coordination develops, and becomes extremely sensitive to visual variables. As the child learns to stand and walk, however, the visual input must now coordinate with other postural control mechanisms, such as joint mechanoreceptors of the hips, knees, and ankles [42].
Aside from the visual field itself providing an important source of postural control, proprioceptive information may also be relayed from the extraocular muscles themselves. Buttner-Ennever and Horn [43] describe a 'dual control' system where two distinct pathways are responsible for afferent input into the oculomotor nuclei. One pathway serves to generate eye rotations, while the second pathway provides sensory information regarding eye alignment and stabilization [44]. This is an important part of the visual postural control pathway, as this pathway may compensate for visual deprivation such as in darkness. This ocular proprioceptive pathway passes through the optic tract nucleus to the rostral portion of the superior colliculus [45,46].
The superior colliculus is known for its essential role in head and eye orientation and coordination [47,48]. It serves as an important integration center for the extraocular proprioceptive pathway as well as the spinal trigeminal nucleus. The superior colliculus also has an extensive reciprocal feedback pathway with the reticular formation, which may also play a role in extraocular proprioception [49]
To further summarize the importance of vision in postural control, Buchanan et al [30] concluded that fixing the head and trunk in space achieves three major functional tasks: 1) it stabilizes the visual field for gaze stabilization, 2) it stabilizes the center of mass of the head and trunk within feet support, and 3) it minimizes the external stress acting upon the head and trunk. Because Buchanan et al [30] showed how visual deprivation destabilizes head and trunk position, this provides evidence that control of the head and trunk is assumed in a top-down mode. This organization may have clinical value when designing treatments to correct abnormal posture.
Vestibular Input
The vestibular system is an integral component in many of the postural reflexes, especially those that are responsible for upright human posture. The primary function of the vestibular apparatus is to provide sensory input about sustained postural stimulation [50]. The vestibular apparatus is composed of the utricle, saccule, and semicircular canals. Each of these organs is designed to detect specific types of motion. The utricle and saccule detect linear accelerations of the head in space. Since gravity exerts a constant vertical acceleration on the head and body, the utricle and saccule provide postural input on head position relative to gravity [50]. The semicircular canals relay afferent input about angular acceleration, such as head rotation. Buttner-Ennever [51] detailed the many connections from the utricle and saccule to the brainstem and cerebellum. The utricle detects changes in head position relative to gravity, such as a simple tilting of the head. The saccule, on the other hand, contributes a partial role in maintaining head position relative to the visual field.
Afferent information is collected and transmitted to higher levels by the vestibular nerve. The vestibular nerve carries afferent input from both the utricle and saccule, where it is transmitted to the lateral vestibular nucleus. Vestibular nuclei receive sensory input from the vestibular nerve as well as information from the cerebellum and the optic tract. Axons from the vestibular nuclei project to the thalamus, superior colliculus, reticular formation, cerebellar flocculus, and lower vestibulospinal nuclei. Of the vestibular nuclei, the lateral vestibular nucleus, or Deiter's nucleus, is perhaps one of the most important nuclei related to postural reflexes, through its projections to the vestibulospinal tract. The vestibulospinal tract and reflex will be discussed later in this review.
Previous experiments have illustrated the effects of vestibular loss on overall postural control [52-54]. Horak et al [53] compared 6 subjects with bilateral vestibular loss to 6 age and sex-matched controls. After subjecting each group to various postural tasks, they found that the experimental group showed increased head and trunk displacements compared to matched controls. In a similar study by Creath et al [52], they found that subjects with bilateral vestibular loss demonstrated a higher center-of-mass variability. However, this variability was reduced with the addition of light-touch fingertip contact. This suggests that despite vestibular deficits, postural control can be maintained by other afferent postural input. Schweigart et al [55] described how subjects with vestibular degradation could compensate with neck proprioception in instances of static postural stance, although postural control is significantly altered when the subject is moving.
Visual and Vestibular Interactions
While the visual and vestibular systems are individually two of the most important postural reflexes, it's their constant interaction that makes the control of upright posture possible, especially when considering their combined role in the reflex modulation of muscular tone through various groups of postural muscles. The visual and vestibular systems interact primarily through a series of reflexes and tracts, namely the vestibulo-ocular reflex [56-62], the vestibulospinal tract [50,63], and the dorsal and ventral spinocerebellar tracts [64-69].
The vestibulo-ocular reflex serves to orient the visual field by creating certain eye movements that compensate for head rotations [59,62] or accelerations [61]. The vestibulo-ocular reflex may be subdivided into three major components: 1) the rotational vestibulo-ocular reflex, which detects head rotation through the semicircular canals, 2) the translational vestibulo-ocular reflex, which detects linear acceleration of the head via the utricle and saccule, and 3) the ocular counter-rolling response, or optokinetic reflex, which adapts eye position during head tilting and rotation [50]. Through detection of head orientation in space, the vestibular apparatus transmits this information to the vestibular nuclei, where connections with the visual field aid in the correction and coordination of head and body posture via the vestibulo-ocular reflex [56]. The cerebellar flocculus may ultimately be responsible for integrating and executing the efferent corrections of the vestibulo-ocular reflex. Previous research has shown that resection of the cerebellar flocculus permanently prevents vestibulo-ocular reflex response, providing evidence for its direct involvement [58].
The medial and lateral vestibulospinal tracts may be viewed as the efferent equivalents of the vestibulo-ocular reflex, modulating motor neuron activity regarding the axial and appendicular muscles respectively so that rapid postural adaptations can take place. The cerebellum, where afferent information is collected from the visual field, the vestibular nerve, and the cervical mechanoreceptors, and is interpreted for generation of reactive postural corrections, modulates these tracts. Originating in the lateral and medial vestibular nuclei [66], these tracts allow the trunk and extremities to compensate for changes in head position. Reflexive responses from the vestibulospinal tracts help correct sudden perturbations in static upright posture. While the visual input may be more important in constant postural adaptation, the vestibular apparatus, via the vestibulospinal tracts, is much quicker to respond to early or slight postural disruptions, allowing for a faster response from the skeletal postural muscles [50].
Normal visual-vestibular interaction also incorporates afferent input from the dorsal and ventral spinocerebellar tracts. These tracts transmit sensory signals to the cerebellum regarding position sense of the lower extremity [65], primarily through joint, skin, muscle spindle, and golgi tendon organ afferents [64]. These tracts not only provide information relating the position of each lower extremity, but also in coordinating both lower extremities for combined postural tasks such as locomotion [70,71]. The spinocerebellar tracts arise from spinal interneurons within the gray matter between the first thoracic and the second lumbar segments, known as Clarke's nucleus [66]. These interneurons, in turn, communicate with both the afferent and efferent pathways of lower extremity neural control, via spinal reflexes. The clinical importance of this will be discussed in greater detail.
Cervical Mechanoreceptors
The cervical spine is a virtual warehouse of postural afferent input and integration. Several anatomic structures in this region, including the facet joint and capsule [72-78], spinal ligaments [71], and proprioceptive input from the cervical musculature [70,79,80] are collectively responsible for maintaining an orthogonal head on neck position. In order to understand how these various structures participate in postural regulation, observation of postural control changes in the presence of functional deficits provides evidence of their individual contributions.
The cervical facet joint houses a variety of mechanoreceptors responsible for providing afferent postural input to higher neurological pathways, including connections with the trochlear, abducens, spinal trigeminal, central and lateral cervical, and vestibular nuclei [81-87], as well as the cerebellar flocculus and vermis [83,84,88]. Several types of cervical facet mechanoreceptors have been identified [85,86]. Cervical facet joint mechanoreceptors may be dominant over the vestibular apparatus in regards to the maintenance of static posture [89,90]. For example, when the cervical facet joints are experimentally immobilized in the presence of vestibular dysfunction, postural instability becomes apparent [91]. However, postural stability is restored when the facet joints are mobilized. The facet joint has been the focus of several recent studies regarding whiplash type injuries. Specifically, the facet joints and capsules have been identified as a probable cause in chronic whiplash symptoms in the absence of obvious radiographic injury. A significant number of free nerve endings and lamellated corpuscles were found within the facet joint capsules [75]. These structures are important in the rapid adaptation of changes in cervical spine position. In a study of 105 patients with chronic whiplash symptoms, Treleaven et al [76] found that whiplash patients could not consistently reproduce a natural resting head position when compared to matched controls. Incidentally, Rubin et al [92] report that people with whiplash symptoms have a higher likelihood of suffering from balance failures. Since cervical facet joints contribute to postural orientation, injury to these joints may produce postural symptoms like vertigo and dizziness [76].
In addition to the facet joints, the paraspinal ligaments, such as the posterior longitudinal ligament, also contribute an extensive amount of sensory input for postural control [71,93-98]. The sensory innervation of spinal ligaments is provided by Pacinian and Ruffini corpuscles, and free nerve endings [94,96-98]. Jiang et al [97] repeatedly stretched an intertransverse ligament of a young chicken. Tracing neuronal production of Fos protein through various sensory pathways, they identified afferent connections with the gracilis and cuneatus nuclei, the vestibular nuclei, and the thalamus. Yamada et al [96] identified a sympathetic innervation of the upper cervical posterior longitudinal ligament, from fibers projecting from the stellate ganglion. Interestingly, Sjolander et al [71] discuss how spinal ligamentous afferent information is at least partially responsible for mediating the reflex activity of its associated muscle spindles. They concluded that although muscle spindles may be dominant over ligament afferent input, maximal accuracy regarding joint position sense requires both sets of joint proprioception.
The cervical spine also contains an intricate muscular afferent network, given the numerous anterior and posterior cervical muscles. The upper cervical spine contains a higher density of muscle spindles than in any other spinal region [70]. Many authors have tested the function of cervical afferents by applying vibration to both normal subjects and those with specific neurological deficiencies [99-112]. For example, Ledin et al [106] found that vibratory stimulation of the calf muscle creates body sway in the sagittal plane, and this sway is significantly altered by flexion or extension, but not rotation, of the head. They suggest that either altered neck muscle position or utricle and saccule proprioceptive interaction may account for this functional deficit during vibratory stimulation. Sagittal postural sway was also observed when vibration was applied to the lower posterior cervical musculature [102]. Like the vestibular apparatus, Ivanenko et al [102] suggest that postural afferents from the cervical muscles are also processed within the parameters of the visual field. In another lower leg vibration study by Vuillerme et al [112], they found that vibration applied to the lower leg in upright humans also increased postural sway, as did muscular fatigue in the lower leg. However, when vibration was applied to a fatigued muscle, the postural sway did not increase as the authors had hypothesized. The authors suggest that the central nervous system effectively disregards the afferent information provided by a fatigued muscle, thus relying on other postural control mechanisms, such as the visual and vestibular systems, to provide this lost control [112]. When vibration is applied to upper cervical musculature, a greater degree of postural compensation occurs compared to that occurring from lower cervical vibration, suggesting that the upper cervical spine has an even greater role in posture regulation through visual orientation, than even the lower cervical spine. This observation is supported by previous works from Bogduk [113-117] showing how injury or pathology of the upper cervical spine produces a significant amount of noxious afferent input into the central nervous system, which may interfere with postural control. This is apparent in individuals with previous neck trauma and concurrent chronic neck pain [118-120].
Vibrational studies have also been conducted on individuals with certain postural deficiencies. In a study comparing normal subjects to those with labyrinthine deficiency, Popov et al [109] observed that vestibular-deficient subjects could not achieve the same ocular tracking of a fixated target image as matched controls. The authors conclude that this may result from changes in the cervico-ocular reflex, which will be discussed later in this review. Interestingly, a study by Karnath et al [103] demonstrated that the head tilt associated with spasmodic torticollis can be significantly reduced at least temporarily, when subjected to cervical vibration for 15 minutes. This finding led the authors to conclude that the muscular spasm associated with spasmodic torticollis may be the result of aberrant afferent input relaying head position to the central nervous system. Bove et al [100] demonstrated how asymmetrical vibration of the sternocleidomastoid affects locomotion. They found that subjects would rotate away from the side of vibration when applied during stepping. However, when the vibration was applied before stepping, compensatory rotation occurred opposite the initial rotation. The authors suggest that cervical input plays a major role during locomotion, and a lesser-coordinated role during static posture. Two other studies by Strupp et al [111] and Betts et al [99] also demonstrated the ability of the cervical afferent input to compensate for a decline in vestibular function.
Visual and Cervical Interactions
While the visual field and the vestibular apparatus have intimate connections for postural control, they also have well-known connections with the cervical spine. Arising from sensory receptors in the cervical spine are three well-known reflexes that aid in postural control: 1) the cervico-ocular reflex [121-124], 2) the cervicocollic reflex [124-128], and 3) the vestibulocollic reflex [128-142], which will be discussed later in this review.
The cervico-ocular reflex serves to orient eye movement to changes in neck and trunk position [143-148] Similarly to other postural reflexes, a basic understanding of the cervico-ocular reflex is achieved by studying patients with specific postural reflex deficiencies. For example, Chambers et al [143] tested 6 patients with bilateral vestibular loss, and 10 controls. They found that light pattern stimulation caused at least a marginal amount of increased cervico-ocular reflex response, which was compensatory in half of the subjects. The authors concluded that the cervico-ocular reflex may at least partially compensate for absent vestibular function and vestibulo-ocular reflex. In another study by Bronstein and Hood [144], the postural control role of the cervico-ocular reflex was also tested in 12 patients with absent vestibular function. They found that the cervico-ocular reflex in patients with absent vestibular function seems to take on the lost function of the vestibulo-ocular reflex during specific postural tasks, such as ocular tracking in the direction of a visual target. Heimbrand et al [145] also studied 5 patients with vestibular absence to identify the compensatory nature of the cervico-ocular reflex. Their findings demonstrate a high degree of plasticity in the cervico-ocular reflex. The authors found that the cervico-ocular reflex could be modified with the addition of optical lenses, where magnifying lenses increase cervico-ocular response. The use of reduced lenses decreased the response. They also found that afferent input from the trunk, cognitive interpretation, and both peripheral and foveal retinal information all contributed to the observed cervico-ocular reflex plasticity. This information seems to be important for cervico-ocular stabilization of the visual field in space and in relation to a stationary neck and movable trunk. In an earlier study by Bronstein et al [146], they found that when the absence of vestibular function was present concurrently with reduced optokinetic reflex or ocular rolling response, the plastic adaptation of the cervico-ocular reflex did not seem to compensate for the vestibular absence. This suggests a necessity of an intact optokinetic reflex for optimal cervico-ocular response.
While cervico-ocular responses have been repeatedly observed in vestibular deficient subjects, its importance in healthy human subjects is debatable [121,147]. Schubert et al [147], in a study of 3 patients with unilateral vestibular dysfunction, could not establish any evidence of a cervico-ocular response in any of the 7 controls or in 2 of the 3 patients. In the single patient with evidence of cervico-ocular response, a change in the reflex could only be obtained following 10 weeks of vestibular exercises. More specifically, however, the cervico-ocular reflex can be subdivided into a slow phase of the response and a quick phase [121]. Jurgerns and Mergner [121] found that while the slow phase of the cervico-ocular reflex has no functional significance in humans, the quick phase does contribute to ocular stabilization and orientation to changes in neck and trunk position, during certain postural tasks. The quick phase of the cervico-ocular reflex also appears to be significantly adaptable in a relatively short period of time. Rijkaart et al [148] tested 13 healthy adults by subjecting them to trunk rotation in a dark room, thus providing conflicting somatosensory and visual input to test the function of the cervico-ocular reflex. They found a significant amount of cervico-ocular adaptation could be achieved in as little as 10 minutes of constant visual and somatosensory input. This may have important clinical benefits, and will be discussed further in the second part of this review.
Perhaps more widely known for its role in postural control, the cervicocollic reflex serves to orient the position of the head and neck in relation to disturbed trunk posture [149]. This reflex, acting similarly to a stretch reflex [149], involves reflexive correction of cervical spine position through co-contraction of specific cervical muscles, including the biventer cervicis, splenius capitis and cervicis, rectus capitis posterior major, and the obliquus capitis inferior [127]. The cervicocollic reflex is activated in response to stimulation of muscle spindles located in these muscles. This reflex seems to modulate upright cervical posture in close communication with the vestibulocollic reflex [124,125], which will be discussed later. There also seems to be a significant amount of overlap in the pathways and functions of the cervicocollic and vestibulocollic reflexes, perhaps to readily compensate for injury or reduction in either of these two reflexes [126]. The vestibulocollic reflex seems more sensitive to changes in head position in the horizontal plane, while the cervicocollic reflex seems more sensitive to vertical plane positional changes [125]. Given the high density of muscle spindles in the cervical musculature, the cervicocollic reflex possesses a high degree of sensitivity to relatively small cervical stimuli. This suggests that this reflex may heavily rely upon muscle spindle afferents to provide postural information, so that immediate cervical postural corrections can be made [127]. Evidence of these immediate changes was illustrated by Keshner et al [125], where patients performed simultaneous postural and cognitive tasks with and without weight placed on top their heads. They found that adding weight to the head did not significantly change head or neck position, suggesting an immediate and compensatory response to the added weight.
Vestibular and Cervical Interaction
Perhaps one of the most well studied postural reflexes; the vestibulocollic reflex maintains postural stability by actively stabilizing the head relative to space. It does this by reflexively contracting cervical muscles opposite of the direction of cervical spine perturbation [115,139]. In order to evaluate the mechanisms and efferent pathways of this reflex, several studies targeting this reflex using EMG recordings of various cervical muscles have been conducted [126,133,134,136,138]. The vestibulocollic reflex, from input originating in the semicircular canals, utricle, and saccule, stabilizes the head in space in response to even the slightest of head perturbations occurring in the horizontal plane [128,134,139,141,142]. From this perspective, the vestibulocollic reflex also acts much like a stretch reflex. Muscles that have been studied in connection with this reflex include the sternocleidomastoid [130,131,133,136], biventer cervicis, splenius cervicis and capitis, and the longus capitis [111].
There is an important distinction to make when discussing the vestibulocollic reflex. It should be noted that this reflex is distinct and largely dissociated from the vestibulospinal reflex, which orients the extremities to the position of the head and neck. Welgampola and Colebatch [138] found that the vestibulocollic reflex is not significantly affected by stimulation of lower extremity afferents, such as when a subject is placed in an upright posture on a narrow base and deprived of vision and external support. Likewise, Allum et al [128] showed that activation of the vestibulocollic reflex is mainly dependent upon stimulation of cervical afferents directly.
Another important aspect of the vestibulocollic reflex is the neural contribution it receives from the reticular formation [140,141]. This reticulospinal contribution is important because it may allow a "globalization" of this reflex, meaning that connection to the reticular formation allows postural information carried by the reflex pathway to be interpreted by several other central nervous system pathways, perhaps allowing the CNS as a whole to adapt to postural changes. These reticular connections also facilitate quicker vestibulocollic responses, and help increase the sensitivity of the vestibulocollic reflex to other postural afferents in related but divergent pathways [140].
Neurological Development of Postural Control
Any discussion pertaining to the mechanisms through which postural adaptations are made must include information on the development of these postural adaptive mechanisms. As already suggested, the visual field may be the most heavily favored of the postural reflexes. As many authors have pointed out, an infant's orientation to the extrauterine environment is dictated almost exclusively by the visual field [150-153]. As Precht [152] discussed, a human newborn is poorly adapted to the gravitational environment, given poor muscle power and weak or absent reflex control of the head and trunk. Infants at 2 months of age begin to consistently rely upon visual cues to orient the head and body. At 4 to 6 months, as infants begin to crawl, other postural reflexes begin to play important roles, such as joint mechanoreceptors and the vestibular system [154]. Pope [153] showed that as infants begin to crawl, reliance on visual feedback is reduced. Perhaps not coincidentally, however, certain stages of upright postural progression may be characterized by periods of reliance on the visual system as the primary mechanism of postural regulation. For example, Butterworth and Hicks [151] pointed out that visual feedback is again favored as the infant masters motor control of the trunk and starts to sit upright independently. Lee and Aronson [155] observed a similar pattern of visual predominance as the infant begins to stand.
From this material, it is logical to conclude that as a child is born, concerning progression from crawling, to sitting upright, to standing, reflexive head control seems to be the primary factor necessary for upright postural control. This sequence of postural development is predicated upon mastering reflex control of head position relative to gravity so that trunk and lower extremity control can be learned using a fixed reference point. This conclusion is further supported by evidence that after reflex control of the head is learned, standing requires a coordinated response of the lower extremity musculature to balance the position of the head over the base of support. As Woollacott et al observed [154], neuromuscular responses of the lower extremity are coordinated much earlier than trunk and upper extremity muscles. The neuromotor responses of the pelvic girdle and lower extremity are collectively termed the pelvo-ocular reflex [156], which serves to orient the body region in response to head position and anticipatory visual reference cues. The significance of this reflex may be attributed to the early development of hip and leg coordination. Neuromuscular coordination of the trunk may not be fully developed until the child reaches 7–10 years of age [154].
Although visual input for postural control seems to predominate in early life, they may be some explanation as to why this occurs. Because the visual system functions independent of gravity [157], this system is not affected by gravitoinertial changes. Therefore, it can provide the most consistent reference point from which to orient the head and neck. Additionally, previous studies have demonstrated that infants cannot process and integrate postural input from multiple sources, such as from joint mechanoreceptors and the vestibular system. In a study of 4–6 month old infants, Woollacott et al [154] found that infants using both visual and vestibular cues were able to correctly orient to a moving platform 60% of the time. However, when the infants were blindfolded using goggles, their postural responses were correctly oriented 100% of the time, suggesting an inability of the infant to process two different sources of postural stimulus simultaneously. By 8–14 months, however, infants appeared to consistently adapt to postural stimuli from both sources of sensory input.
Biomechanical Development of Postural Control
Aside from the neurological development of postural control, it is important to discuss the biomechanical development of postural control, especially as it relates to the spine. Since the spine is the literal backbone of upright postural support, structural and functional development of the spine also appear to be consequences of upright adaptation to a gravitational environment. The sagittal curves of the spine allow for a balance between strength and flexibility, while also resisting the axial compressive force of gravity [158]. These sagittal curves are not fully developed at birth. Rather, they are formed as a consequence of adaptation to the external environment (gravity). In utero, the fetal spine is shaped more as a C-shaped curve. This shape is more suited to adapt to a microgravitational environment. However, as the fetus grows and occupies more of the uterus, much of the watery environment is lost. Therefore, the fetal spine begins to adapt and take on a structure more suited for gravitational adaptation. Bagnall et al [159] suggested that the cervical curve is fully developed in utero. However, their study used postmortem fetuses artificially positioned and radiographed, although the authors note that much attention was given to replicating the fetal position in the uterus. Although they note no visual abnormalities, no information is given as to the cause of fetal demise or maternal history. Therefore, it is possible that these fetuses are not representative of the average healthy fetal population. Panattoni and Todros [160] demonstrated through ultrasonography that both the cervical and lumbar curves are visually developed by the 24th–26th week of gestation. This may be due to the morphological development of the cervical facet joints and discs.
The extrauterine environment changes the compressive force and force vectors upon the spine. As previously mentioned, the newborn muscle strength is not sufficient to maintain upright head posture. From a mechanical standpoint, creating more of a mechanical advantage can compensate for this lack of muscle strength. The C-shaped spine provides two intrinsic lever arms from which the paraspinal muscles attach and initiate movement. However, since the spinal muscles are too weak to maintain upright head position, shorter lever arms must be developed to overcome this muscular deficit. The forward cervical curve creates two more functional lever arms, giving the cervical spine muscles the mechanical advantage necessary for upright head stabilization and movement. Figure 1 depicts these developmental stages. As the child begins to crawl, the lumbar curve is developed as a result of the downward pull of gravity. Once the lumbar curve is developed, two more lever arms are created, providing the lumbopelvic musculature with the leverage necessary to allow an upright standing posture. From an engineering perspective, the resistance of a curved column is directly proportional to the square of the number of curves plus one [158]. Therefore, a C-shaped fetal spine contains two lever arms, for a resistance of 5 (2 × 2 + 1 = 5). When the child develops a cervical curve, two additional lever arms are created, thereby increasing the resistance to 17 (4 × 4 + 1 = 17). Finally, the lumbar curve development further increases the resistance of the spine to 37. An illustration of this is shown in Figure 2. The creation of these lever arms allows the spinal muscles to maintain upright posture more efficiently. Initially, upright postural control is a voluntary muscular task. However, as the spinal muscles are repeatedly required to perform these tasks, the tasks become automated. As a nerve impulse passes through a set of neurons at the exclusion of others, it will take the same course on future occasions and each time it traverses this path the resistance will be smaller [161,162]. As these postural neuromotor pathways are facilitated, they become the basis for the neurophysiologic reflexes governing involuntary postural regulation. Harbst [163] previously reported that repeated voluntary performance of a postural task becomes faster and easier to perform as neuromotor pathways are reinforced. As the infant begins to hold his/her head upright, the postural muscles required to perform that task are activated. As this task is repeated, holding the head upright becomes an involuntary, automated process under the direction and control of the postural reflexes via the facilitated neuromotor pathways. The same neurological learning processes are invoked as the child begins attempting to stand upright. At this point, postural muscles are required to perform many functions simultaneously. Joints of the thoracic and lumbopelvic spine, as well as the hips, knees and ankles, must all be actively stabilized by the surrounding regional musculature. Meanwhile, the global spine and posture must balance the position and weight of the head, neck and trunk above their base of support, the feet. This active muscular stabilization also increases the stress on musculotendinous junctions and osseous attachments, increasing the rate at which the skeletal frame ossifies. As the process of standing is repeated, neuromotor control of the lower extremity and spinal muscles is coordinated with the postural reflexes of the head and neck through cerebellar integration, thereby developing a cohesive network of involuntary postural control.
Figure 1 This figure illustrates the development of the sagittal spinal curves. In the womb, the fetal spine is more of a C-shape (left). As the child begins to hold his head up, the cervical curve is developed and reinforced (middle). Finally, as the child begins to crawl, gravity helps to develop the lumbar curve, a requisite for a bipedal upright stance (right).
Figure 2 The resistance (R) of any curved column to compression forces is directly proportional to the square of the number of curves (N) plus one (R = N2 + 1). Therefore, the fetal spine with its single curve has a resistance value of 2 (12 + 1 = 2). This is not enough to resist the forces of gravity against the head, neck, and upper trunk as we saw in Figure 1. The development of the cervical curve increases the resistance value of the spine by 2.5 times (22 + 1 = 5).
Reflex Hierarchy
Human upright posture is developed and maintained in response to earth's gravity. Perhaps the best way to study the effects of gravity upon the human spine and nervous system is to study humans as they actively adapt to environments where the gravitational field is altered or absent. Clues to reflex hierarchy and reference may be determined as a consequence of forced adaptation to a new external environment.
Since space travel has become a reality, several studies have demonstrated the effects of microgravity on human posture. Perhaps most importantly, it would appear that a postural reflex hierarchy may exist irrespective of the external environment. For example, a study by Baroni et al [164] evaluated two astronauts during space flight using kinematic analysis. The astronauts were instructed to perform specific axial movements from an erect, upright posture. Their postures and movements were recorded before, during, and after the movement performance. The authors found a pronounced forward trunk lean when the eyes were closed compared to eyes open. They suggest that visual input for postural control may be independent of gravity-based postural cues. This conclusion is also supported by research from other authors. Koga [157] studied the eye movements of humans during spaceflight. He found that purposeful eye movements showed similar accuracy of target fixation and saccade compared to pre-flight eye movements. Further, Koga [157] reported that neck muscle activity was not coordinated with ocular movement during spaceflight, although oculocervical coordination was observed under Earth's gravity. These findings demonstrate a visual preference for postural control in altered external environments, and that cervical spine afferents are gravity dependent. More specifically, extraocular muscular afferents are highly coordinated and function independently of gravitational changes. Regardless of the gravitational environment, visual afferents and cues provide an external reference for maintaining upright posture, even when somatosensory afferents and internal references are absent or conflicting [165-167].
In the presence of Earth's gravity, the vestibular system plays a major role in monitoring changes in head position, primarily through the utricle and saccule [168]. However, initial exposure to microgravitational environments reduces the effects of the vestibular organs on posture regulation. Clarke et al [168] showed that vestibular control of posture recovers only after prolonged exposure to microgravity. They suggest that cervical spine afferents may play a role in vestibular recovery.
Postural Body Scheme
Another method for observing and documenting the interactions of postural reflexes is to study the causes and factors associated with space motion sickness. This sickness is simply a result of conflicting postural input into the central nervous system. This sickness is common in the first days of spaceflight, and resolves as adaptation to microgravity occurs [169]. The occurrence of space motion sickness provides a framework from which postural control theories attempt to explain upright posture regulation in direct response to gravity.
A conceptual model called the postural body scheme [165,170-173] represents the internal reference point by which upright posture is regulated. Vertical body orientation, corrective postural reactions, and anticipatory postural adjustments are all organized based on this internal representation [170]. This postural body scheme remains stable during gravitational changes, even when mechanoreceptive and vestibular inputs are significantly decreased [174]. During periods of microgravity, space motion sickness is attributed to a conflict between visual postural inputs and afferents from the vestibular and somatosensory systems. The postural body scheme is centered around gravity acting as the vertical axis of space while in earth's gravity. However, this vertical axis is not present in microgravity, effectively eliminating the external reference point for many of the postural reflexes [175]. To observe these effects, Takahashi et al [176] performed Coriolis stimulation on five healthy subjects before and during space flight. The subjects were instructed to tilt their heads at varying speeds in both gravitational environments. Observations were recorded regarding eye movement, body sway, and motion sickness. They found that nystagmus was present under both conditions, although its duration was shorter in microgravity. However, body sway and sickness was not observed in microgravity, although they were apparent in normal gravity. Their findings provide evidence that visual control of posture is defined by an internal reference frame within the brain, not subject to changes in external environment. In an experiment conducted by Amblard et al [177], they recorded movements associated with head stabilization in two subjects during space flight. Their results also suggest a postural reflex hierarchy, with visual input, vestibular input, and postural body scheme among the most important. The main underlying commonality among all of these cited studies is the predominance of visual input and afferents in regulating upright postural control despite changes in the external gravitational environment. Finally, although the postural body scheme is an internal reference point for postural control, it may receive much of its information from visual input. Yakushin et al [178] performed vestibular stimulation on five subjects while their heads were immobilized. The subjects were then placed in a side lying position. Three-dimensional ocular movements were recorded during vestibular stimulation. The authors found that adaptations of the vestibulo-ocular reflex are gravity-dependent, and appear to be stored as a sort of short-term posture memory. They suggest that the vestibular nuclei may be responsible for the storage of this gravity-dependent posture information, since these nuclei form direct connections between vestibular and visual afferents.
Not only does the postural body scheme provide an internal framework for maintaining upright posture, but it also serves as a stable internal representation of biomechanical properties to guide and organize anticipatory postural adjustments and voluntary motor movements [179,180]. Understanding the conceptual model of the postural body scheme can be clinically beneficial to manipulative medical clinicians in that biomechanical functional improvement may ultimately rely on the patient's ability to learn novel neuromotor strategies for upright posture and gait.
Discussion
This review has focused mainly upon the postural reflexes associated with the cervical spine and its constituent parts, and the special senses, specifically the eyes and inner ear. Obviously, there are other postural reflexes we did not cover in this review, including skin and surface receptors in the extremities [181,182], somatic graviceptors [183-185] located within the viscera, and baroreceptors located within the circulatory system [186,187]. However, they were not covered in this review for specific reasons. Skin and surface receptors in the lower extremity have been well illustrated in the chiropractic literature, especially in regards to postural control. Various authors have already shown improvements in postural control, via balance testing, using molded foot orthotics [188,189], for example. Given the extensive information already published [181,182] on this aspect of postural control, we did not address it here. However, this omission does not diminish its importance for postural control. Additionally, while somatic graviceptors and vascular baroreceptors also maintain certain postural regulatory functions, these components are not readily modifiable by manual medical methods. Our review has focused upon those postural reflexes that can be predicted and recruited by clinicians in chiropractic, physical therapy, physical medicine, and osteopathic medicine to ultimately help execute specific postural rehabilitation programs.
We attempted to review and discuss the anatomy and interactions of the various postural reflexes. However, with the amount of overlap found in many neurological processes, it is likely impossible to identify and outline each and every postural reflexive behavior. Also, as biomedical technology and research uncovers new areas of neurobiology and neurophysiology, we will no doubt find our present review of postural reflexes to be inadequate. The complexity of the nervous system has probably not yet been fully appreciated by medical and allied health practitioners, nor are its connections to all other physiological processes, including posture, fully documented and understood.
This review is important in that much of this information has largely been ignored in chiropractic and manual medicine. Although posture correction has gained significant popularity more recently, the neurological control of posture has been largely omitted. While previous reviews have outlined potential pathogenetic biomechanical configurations of the spine and spinal cord [18-20], neurophysiologic adaptation to normal and abnormal posture has not been extensively detailed. Given both the historical and clinical importance of the nervous system to overall health and well being [2,10], its involvement in something as important as postural control should be emphasized in future chiropractic literature.
Traditional chiropractic principles maintain that the nervous system is responsible for coordinating all other body systems [190]. Typically speaking, this perspective is applied to the other specific body systems, such as the cardiovascular, digestive, endocrine, and respiratory systems. However, in this paper, we detail the mechanisms and adaptive processes by which the nervous system also controls and coordinates our upright posture. In addition to regulating our internal environment, our nervous system, through its various postural reflexes, observes, analyzes, predicts, and adapts to changes in our external environment. Primarily, this external environment is gravity.
Chiropractors aim to evaluate and treat articular dysfunction of the spine to restore function, reduce pain, and encourage normal nervous system "integrity." [191] This has typically been performed by trying to determine spinal segmental alterations in alignment in relation to the vertebral segments above and below. However, from the data presented here, we suggest that the spine, as a singularly functioning entity, is subservient to the reflex adaptations made by the nervous system in relation to gravity. So, it may be worth trying to identify potentially putative postural reflex function(s). A theoretical example may be a thoracolumbar pain caused by a swayback posture. It is postulated that the swayback posture may be the result of a forward head posture relative to the trunk, thus causing a forward shift of the pelvic complex. This forward pelvic shift is mediated by the pelvo-ocular reflex [156], as outlined in our previous review. However, the underlying cause of this theoretical thoracolumbar pain and dysfunction may be the forward head posture forcing a compensatory swayback posture.
There are a couple of specific points made in this paper that we wish to highlight and relate to clinical practice, since these concepts are not currently explored or discussed in the chiropractic literature. First, as discussed earlier in this paper, we reviewed previous work by Bove et al [100] testing the cervical spine contribution to locomotion and static posture through unilateral sternocleidomastoid vibration. Again, their findings suggest that the cervical spine plays a larger role in locomotion and a smaller role in static upright posture. This information has important clinical implications that manual practitioners should consider. Locating specific postural reflex deficits may be achieved by subjecting patients to various postural and locomotive balance tasks to identify which function is being compromised. Although unproven, isolation and treatment of the specific deficient postural reflex may well mean the difference between treatment success and failure.
Two separate studies by Karnath et al [103] and Vuillerme et al [112] compared the afferentation of a fully active muscle and a fatigued muscle. The collective results of these studies may have very important clinical relevance. If fatigued muscles are not able to transmit somatosensory information to the central nervous system, then upright postural control may be compromised if maintaining a given static posture requires a large amount of constant isometric muscular contraction. Certain subpopulations may therefore be advised to undergo posture correction, such as those elderly who are at risk for balance failures and/or hip fractures. These balance failures may be at least partially attributed to lack of somatosensory input from fatigued postural muscles. This hypothesis is certainly worthy of research.
While neurological disturbances have been well documented in other contexts, such as whiplash-associated disorders [76,89,90,92,192], neurological disturbances resulting from chronic abnormal posture have not been elucidated. This may be due to a more narrow focus upon only the mechanical components of the spine and posture [16,17,22-27]. We hope that this review will help to shed some light upon the postural adaptations and responses that may not only cause neurophysiological dysfunction, but also those that may occur because of it.
Conclusion
Upright human posture is maintained reflexively by a vast network of peripheral and central nervous pathways designed to provide instantaneous input regarding both internal and external environmental factors. In this review, we outlined those postural reflexes related to pathways and structures involving the cervical spine, the eyes, and the inner ear. How these structures and pathways obtain somatosensory input, interact with each other, and modulate postural changes and corrections has been described here. While there are many other postural control mechanisms we did not discuss in this review, we chose to outline those reflexes that may be of primary importance to practitioners within the manual healing arts. This review may shed some light upon the idea that vertebral misalignments or fixations are not random injury- or activity-induced events. Rather, they may be a consequence of an adaptive postural process to the external environment mediated by the nervous system through its extensive network of postural reflexes. Research into this concept is necessary before clinical utility is determined.
==== Refs
Gray H Gray's Anatomy: The Classic Collector's Edition 1977 Gramercy Books New York
Palmer DD The Chiropractor's Adjuster The Science, Art, and Philosophy of Chiropractic 1910 Portland, OR: Portland Printing House Company
Berthoud HR Multiple neural systems controlling food intake and body weight Neurosci Biobehav Rev 2002 26 393 428 12204189 10.1016/S0149-7634(02)00014-3
Bellinger LL Bernardis LL The dorsomedial hypothalamic nucleus and its role in ingestive behavior and body weight regulation: lessons learned from lesioning studies Physiol Behav 2002 76 431 42 12117580 10.1016/S0031-9384(02)00756-4
Downing JE Miyan JA Neural immunoregulation: emerging roles for nerves in immune homeostasis and disease Immunol Today 2000 21 281 289 10825740 10.1016/S0167-5699(00)01635-2
Kumar VM Neural regulation of glucose homeostasis Indian J Physiol Pharmacol 1999 43 415 24 10776456
Shamsuzzaman AS Somers VK Cardiorespiratory interactions in neural circulatory control in humans Ann NY Acad Sci 2001 940 488 499 11458704
Streilein JW Okamoto S Sano Y Taylor AW Neural control of ocular immune privilege Ann NY Acad Sci 2000 917 297 306 11268357
Tracey KJ Czura CJ Ivanova S Mind over immunity FASEB J 2001 15 1575 1576 11427490 10.1096/fj.01-0148hyp
Strang VV Essential Principles of Chiropractic 1984 Palmer College of Chiropractic Davenport IA
Keating JC Scientific basis for the chiropractic subluxation complex J Manipulative Physiol Ther 1986 9 52 56 3701227
Leboeuf-Yde C How real is the subluxation? A research perspective J Manipulative Physiol Ther 1998 21 492 494 9777550
Seaman DR Winterstein JF Dysafferentation: a novel term to describe the neuropathophysiological effects of joint complex dysfunction. A look at likely mechanisms of symptom generation J Manipulative Physiol Ther 1998 21 267 280 9608382
Silberstein SD Lipton RB Goadsby PJ Headache in clinical practice 2002 2 Martin Dunitz Ltd 1, 115 123
Jones GM Kandel ER, Schwartz JH, Jessell TM Posture Principles of Neural Science 2000 4 McGraw-Hill Companies
Dulhunty J A simplified conceptual model of the human cervical spine for evaluating force transmission in upright static posture J Manipulative Physiol Ther 2002 25 306 317 12072851 10.1067/mmt.2002.124421
Harrison DD Janik TJ Troyanovich SJ Harrison DE Colloca CJ Evaluation of the assumptions used to derive an ideal normal cervical spine model J Manipulative Physiol Ther 1997 20 246 254 9168409
Harrison DE Cailliet R Harrison DD Troyanovich SJ Harrison SO A review of biomechanics of the central nervous system-part I: spinal canal deformations resulting from changes in posture J Manipulative Physiol Ther 1999 22 227 234 10367759
Harrison DE Cailliet R Harrison DD Troyanovich SJ Harrison SO A review of biomechanics of the central nervous system-part II: spinal cord strains from postural loads J Manipulative Physiol Ther 1999 22 322 332 10395435
Harrison DE Cailliet R Harrison DD Troyanovich SJ Harrison SO A Review of biomechanics of the central nervous system-part III: Spinal cord stresses from postural loads and their neurologic effects J Manipulative Physiol Ther 1999 22 399 410 10478773
Harrison DE Harrison DD Colloca CJ Betz J Janik TJ Holland B Repeatability over time of posture, radiograph positioning, and radiograph line drawing: an analysis of six control groups J Manipulative Physiol Ther 2003 26 87 98 12584507 10.1067/mmt.2003.15
Harrison DE Harrison DD Janik TJ Jones EW Cailliet R Normand M Comparison of axial and flexural stresses in lordosis and three buckled configurations of the cervical spine Clin Biomech 2001 16 276 284 10.1016/S0268-0033(01)00006-7
Harrison DE Harrison DD Troyanovich SJ Harmon S A normal spinal position: it's time to accept the evidence J Manipulative Physiol Ther 2000 23 623 644 11145804 10.1067/mmt.2000.110941
Harrison DE Janik TJ Harrison DD Cailliet R Harmon SF Can the thoracic kyphosis be modeled with a simple geometric shape? The results of circular and elliptical modeling in 80 asymptomatic patients J Spinal Disord 2002 15 213 220
Harrison DE Jones EW Janik TJ Harrison DD Evaluation of the axial and flexural stresses in the vertebral body cortex and trabecular bone in lordosis and two sagittal cervical translation configurations with an elliptical shell model J Manipulative Physiol Ther 2002 25 391 401 12183697 10.1067/mmt.2002.126128
Janik TJ Harrison DD Cailliet R Troyanovich SJ Harrison DE Can the sagittal lumbar curvature be closely approximated by an ellipse? J Orthop Res 1998 16 766 770 9877403 10.1002/jor.1100160620
Weigand R Kettner NW Brahee D Marquina N Cervical spine geometry correlated to cervical degenerative disease in a symptomatic group J Manipulative Physiol Ther 2003 26 341 346 12902961 10.1016/S0161-4754(03)00068-X
Baloh RW Honrubia V Clinical Neurophysiology of the Vestibular System 1989 2 Philadelphia: FA Davis
Wahler-Luck M Schutz T Kretschmann HJ A new anatomical representation of the human visual pathways Graefe's Arch Clin Exp Ophthalmol 1991 229 201 205 1869052 10.1007/BF00167867
Buchanan JJ Horak FB Emergence of postural patterns as a function of vision and translation frequency J Neurophysiol 1999 81 2325 2339 10322069
Catanzariti JF Salomez E Bruandet JM Thevenon A Visual deficiency and scoliosis Spine 2001 26 48 52 11148645 10.1097/00007632-200101010-00010
Desmurget M Vindras P Grea H Viviani P Grafton ST Proprioception does not quickly drift during visual occlusion Exp Brain Res 2000 134 363 377 11045361 10.1007/s002210000473
Guerraz M Thilo KV Bronstein AM Gresty MA Influence of action and expectation on visual control of posture Cognitive Brain Res 2001 11 259 266 10.1016/S0926-6410(00)00080-X
Guerraz M Yardley L Bertholon P Pollak L Rudge P Gresty MA Bronstein AM Visual vertigo: symptom assessment, spatial orientation and postural control Brain 2001 124 1646 1656 11459755 10.1093/brain/124.8.1646
Havertape SA Cruz OA Abnormal head posture associated with high hyperopia J AAPOS 1998 2 12 16 10532361
Nucci P Rosenbaum A Acquired anomalous head posture following loss of vision in one eye Acta Ophthalmol Scand 2002 80 109 112 11906317 10.1034/j.1600-0420.2002.800123.x
Pyykko I Vesikivi M Ishizaki H Magnusson M Juhola M Postural control in blinds and in usher's syndrome Acta Otolaryngol 1991 603 606
Stevens DJ Hertle RW Relationship between visual acuity and anomalous head posture in patients with congenital nystagmus J Pediatr Ophthalmol Strabismus 2003 40 259 264 14560831
Hills-Willford C Kisner C Glenn TM Sachs L The interaction of wearing multifocal lenses with head posture and pain JOSPT 1996 23 194 199 8919398
Soechting JF Berthoz A Dynamic role of vision in the control of posture in man Exp Brain Res 1979 36 551 561 477782 10.1007/BF00238522
DiZio P Li W Lackner JR Matin L Combined influences of gravitoinertial force level and visual field pitch on visually perceived eye level J Vestib Res 1997 7 381 392 9376912 10.1016/S0957-4271(96)00163-2
Bertenthal BI Rose JL Bai DL Perception-action coupling in the development of visual control of posture J Exp Psychol: Hum Percept Perform 1997 23 1631 1643 9425672 10.1037//0096-1523.23.6.1631
Buttner-Ennever JA Horn AKE The neuroanatomical basis of oculomotor disorders: the dual motor control of extraocular muscles and its possible role in proprioception Curr Opin Neurol 2002 15 35 43 11796949 10.1097/00019052-200202000-00007
Buttner-Ennever JA Cohen B Horn AKE Efferent pathways of the nucleus of the optic tract in monkey and their role in eye movements J Comp Neurol 1996 373 90 107 8876465 10.1002/(SICI)1096-9861(19960909)373:1<90::AID-CNE8>3.0.CO;2-8
Buttner-Ennever JA Horn AKE Graf W Ugolinf AG Modern concepts of brainstem anatomy Ann NY Acad Sci 2002 956 75 84 11960795
Distler C Hoffman KP Cortical input to the nucleus of the optic tract and dorsal terminal nucleus (NOT-DTN) in macaques: a retrograde tracing study Cerebral Cortex 2001 11 572 589 11375918 10.1093/cercor/11.6.572
Klier EM Wang H Crawford JD Three-dimensional eye-head coordination is implemented downstream from the superior colliculus J Neurophysiol 2003 89 2839 2853 12740415
Stuphorn V Bauswein E Hoffman KP Neurons in the primate superior colliculus coding for arm movements in gaze-related coordinates J Neurophysiol 2000 83 1283 1299 10712456
Chen B May PJ The feedback circuit connecting the superior colliculus and central mesencephalic formation: a direct morphological demonstration Exp Brain Res 2000 131 10 21 10759167 10.1007/s002219900280
Goldberg ME Hudspeth AJ Kandel ER, Schwartz JH, Jessell TM The Vestibular System Principles of Neural Science 2000 4 McGraw-Hill Companies
Buttner-Ennever JA A review of otolith pathways to brainstem and cerebellum Ann NY Acad Sci 1999 871 51 64 10372062
Creath R Kiemel T Horak F Jeka JJ Limited control strategies with the loss of vestibular function Exp Brain Res 2002 145 323 333 12136382 10.1007/s00221-002-1110-0
Horak FB Buchanan J Creath R Jeka J Vestibulospinal control of posture Sensorimotor control of movement and posture 2002 Kluwer Academic/Plenum Publishers
Hunter MC Hoffman MA Postural control: visual and cognitive manipulations Gait Posture 2001 13 41 48 11166553 10.1016/S0966-6362(00)00089-8
Schweigart G Chien RD Mergner T Neck proprioception compensates for age-related deterioration of vestibular self-motion perception Exp Brain Res 2002 147 89 97 12373373 10.1007/s00221-002-1218-2
Angelaki DE Hess BJ Suzuki J Differential processing of semicircular canal signals in the vestibule-ocular reflex J Neurosci 1995 15 7201 7216 7472475
Brandt T Modelling brain function: the vestibulo-ocular reflex Curr Opin Neurol 2001 14 1 4 11176210 10.1097/00019052-200102000-00001
Highstein SM Role of the flocculus of the cerebellum in motor learning of the vestibule-ocular reflex Otolaryngol Head Neck Surg 1998 119 212 220 9743077
Koizuka I Adaptive plasticity in the otolith-ocular reflex Auris Nasus Larynx 2003 30 S3 6 12543152 10.1016/S0385-8146(02)00117-7
Pettorossi VE Errico P Ferraresi A Manni E Influence of the extraocular muscle proprioceptors on the orientation of the vestibule-ocular reflex Acta Otolaryngol 1996 116 198 200 8725513
Raphan T Cohen B The vestibulo-ocular reflex in three dimensions Exp Brain Res 2002 145 1 27 12070741 10.1007/s00221-002-1067-z
Ratham R Anastasio TJ Evidence for a cooperative learning mechanism in the vestibule-ocular reflex Neuroreport 1995 6 2129 2133 8595186
Rub U Brunt ER de Vos RA Del Turco D Del Tredici K Gierga K Schultz C Ghebremedhin E Burk K Auburger G Braak H Degeneration of the central vestibular system in spinocerebellar ataxia type 3 (SCA3) patients and its possible clinical significance Neuropathol Appl Neurobiol 2004 30 402 414 15305986 10.1111/j.1365-2990.2004.00554.x
Bosco G Poppele RE Modulation of dorsal spinocerebellar responses to limb movement: II. Effect of sensory input J Neurophysiol 2003 90 3372 3383 14615435
Bosco G Poppele RE Reference frames for spinal proprioception: kinematics based or kinetics based? J Neurophysiol 2000 83 2946 2955 10805690
Ghez C Thach WT Kandel ER, Schwartz JH, Jessell TM The Cerebellum Principles of Neural Science 2000 4 McGraw-Hill Companies
Poppele RE Bosco G Rankin AM Independent representations of limb axis length and orientation in spinocerebellar response components J Neurophysiol 2002 87 409 422 11784759
Poppele RE Rankin A Eian J Dorsal spinocerebellar tract neurons respond to contralateral limb stepping Exp Brain Res 2003 149 361 370 12632238
Schoenen J Grant G Spinal Cord: Connections The Human Nervous System 2004 2 San Diego, CA Elsevier Academic Press
Kulkarni V Chandy MJ Babu KS Quantitative study of muscle spindles in suboccipital muscles of human foetuses Neurol India 2001 49 355 259 11799407
Sjolander P Johansson H Djupsjobacka M Spinal and supraspinal effects of activity in ligament afferents J Electro Kinesio 2002 12 167 176 10.1016/S1050-6411(02)00017-2
Grod JP Diakow PR Effect of neck pain on verticality perception: A cohort study Arch Phys Med Rehabil 2002 83 412 415 11887124 10.1053/apmr.2002.29660
Ohtori S Takahashi K Chiba T Yamagata M Sameda H Moriya H Sensory innervation of the cervical facet joint in rats Spine 2001 26 147 150 11154533 10.1097/00007632-200101150-00007
Sterling M Jull G Vicenzino B Kenardy J Sensory hypersensitivity occurs soon after whiplash injury and its associated with poor recovery Pain 2003 104 509 517 12927623 10.1016/S0304-3959(03)00078-2
Strasmann TJ Feilscher TH Baumann KI Halata Z Distribution of sensory receptors in joints of the upper cervical column in the laboratory marsupial monodelphis domestica Anat Anz 1999 181 199 206 10.1016/S0940-9602(99)80008-9
Treleaven J Jull G Sterling M Dizziness and unsteadiness following whiplash injury: characteristic features and relationship with cervical joint position error J Rehabil Med 2003 35 36 43 12610847 10.1080/16501970306109
Winkelstein BA McLendon RE Barbir A Myers BS An anatomical investigation of the human cervical facet capsule, quantifying muscle insertion area J Anat 2001 198 455 461 11327207 10.1046/j.1469-7580.2001.19840455.x
Winkelstein BA Nightingale RW Richardson WJ Myers BS The cervical facet capsule and its role in whiplash injury: a biomechanical investigation Spine 2000 25 1238 1246 10806500 10.1097/00007632-200005150-00007
Lin CCK Crago PE Structural model of the muscle spindle Ann Biomed Eng 2002 30 68 83 11874143 10.1114/1.1433488
Schieppati M Nardone A Schmid M Neck muscle fatigue affects postural control in man Neuroscience 2003 121 277 285 14521987 10.1016/S0306-4522(03)00439-1
Bartsch T Goadsby PJ Increased responses in trigeminocervical nociceptive neurons to cervical input after stimulation of the dura mater Brain 2003 126 1801 1813 12821523 10.1093/brain/awg190
Metherate RS da Costa DC Herron P Dykes RW A thalamic terminus of the lateral cervical nucleus: the lateral division of the posterior nuclear group J Neurophysiol 1986 56 1498 1520 3027272
Pompeiano O Allum JHJ Vestibulospinal control of posture and locomotion; progress in brain research 1988 76 Elsevier Science Publishers
Schwabe A Drepper J Maschke M Diener H-C Timmann D The role of the human cerebellum in short- and long-term habituation of postural responses Gait Posture 2004 19 16 23 14741300 10.1016/S0966-6362(03)00006-7
McLain RF Mechanoreceptor endings in human cervical facet joints Spine 1994 19 495 501 8184340
Yoganandan N Knowles SA Maiman DJ Pinter FA Anatomic study of the morphology of human cervical facet joint Spine 2003 28 2317 2323 14560077 10.1097/01.BRS.0000085356.89103.A5
Wyke B Neurology of the cervical spinal joint Physiotherapy 1979 65 72 76 451044
Ouchi Y Okada H Yoshikawa E Nobezawa S Futatsubashi M Brain activation during maintenance of standing postures in humans Brain 1999 122 329 338 10071060 10.1093/brain/122.2.329
Igarashi M Alford BR Watanabe T Maxian CM Role of neck proprioceptors for the maintenance of dynamic bodily equilibrium in the squirrel monkey Laryngoscope 1969 69 1713 1727 4981231
Raymond G Disturbances of Nervous Function 1969 1 New York: John Willey and Son
Wyke BD Neuromuscular mechanisms influencing mandibular posture: a neurologist's review of current concepts J Dentistry 1973 2 11 20
Rubin AM Woolley SM Dailey VM Goebel JA Postural stability following mild head or whiplash injuries Am J Otol 1995 16 216 221 8572122 10.1016/0196-0709(95)90108-6
Brolin K Halldin P Development of a finite element model of the upper cervical spine and a parameter study of ligament characteristics Spine 2004 29 376 385 15094533 10.1097/01.BRS.0000090820.99182.2D
Jiang H Russell GG Raso VJ Moreau MJ Hill DL Bagnall KM The nature and distribution of the innervation of human supraspinal and interspinal ligaments Spine 1995 20 869 876 7644951
Johnson GM The sensory and sympathetic nerve supply within the cervical spine: review of recent observations Man Ther 2004 9 71 76 15040965 10.1016/S1356-689X(03)00093-6
Yamada H Honda T Yaginuma H Kikuchi S Sugiura Y Comparison of sensory and sympathetic innervation of the dura mater and posterior longitudinal ligament in the cervical spine after removal of the stellate ganglion J Comp Neurol 2001 434 86 100 11329131 10.1002/cne.1166
Jiang H Moreau MJ Raso J Russell G Bagnall K Identification of the location, extent, and pathway of sensory neurologic feedback after mechanical stimulation of a lateral spinal ligament in chickens Spine 1997 22 17 25 9122777 10.1097/00007632-199701010-00004
Jiang H Moreau MJ Raso VJ Russell GG Bagnall KM A comparison of spinal ligaments. Differences between bipeds and quadrupeds J Anat 1995 187 85 91 7591988
Betts GA Barone M Karlberg M MacDougall H Curthoys IS Neck muscle vibration alters visually-perceived roll after unilateral vestibular loss Neuroreport 2000 11 2659 2662 10976939
Bove M Courtine G Schieppati M Neck muscle vibration and spatial orientation during stepping in place in humans J Neurophysiol 2002 88 2232 2241 12424265
Han Y Lennerstrand G Changes of visual localization induced by eye and neck muscle vibration in normal and strabismic subjects Graefes Arch Exp Ophthalmol 1999 237 815 823 10.1007/s004170050318
Ivanenko YP Grasso R Lacquaniti F Neck muscle vibration makes walking humans accelerate in the direction of gaze J Physiol 2000 525 803 814 10856131 10.1111/j.1469-7793.2000.t01-1-00803.x
Karnath HO Konczak J Dichgans J Effect of prolonged neck muscle vibration on lateral head tilt in severe spasmodic torticollis J Neurol Neurosurg Psychiatry 2000 69 658 660 11032623 10.1136/jnnp.69.5.658
Karnath HO Reich E Rorden C Fetter M Driver J The perception of body orientation after neck-proprioception stimulation. Effects of time and of visual cueing Exp Brain Res 2002 143 350 358 11889513 10.1007/s00221-001-0996-2
Kavounoudias A Gilhodes JC Roll R Roll JP From balance regulation to body orientation: two goals for muscle proprioceptive information processing? Exp Brain Res 1999 124 80 88 9928792 10.1007/s002210050602
Ledin T Hafstrom A Fransson PA Magnusson M Influence of neck proprioception on vibration-induced postural sway Acta Otolaryngol 2003 123 594 599 12875581 10.1080/00016480310001835
Lewald J Karnath HO Ehrenstein WH Neck-proprioceptive influence on auditory lateralization Exp Brain Res 1999 125 389 396 10323284 10.1007/s002210050695
McKenna GJ Peng GC Zee DS Neck muscle vibration alters visually perceived roll in normals J Assoc Res Otolaryngol 2004 5 25 31 14569429 10.1007/s10162-003-4005-2
Popov KE Lekhel H Faldon M Bronstein AM Gresty MA Visual and oculomotor responses induced by neck vibration in normal subjects and labyrinthine-defective patients Exp Brain Res 1999 128 343 352 10501806 10.1007/s002210050854
Ribot-Ciscar E Trefouret S Aimonetti JM Attarian S Pouget J Roll JP Is muscle spindle proprioceptive function spared in muscular dystrophies? A muscle tendon vibration study Muscle Nerve 2004 29 861 866 15170619 10.1002/mus.20044
Strupp M Arbusow V Borges Pereira C Dieterich M Brandt T Subjective straight-ahead during neck muscle vibration: effects of ageing Neuroreport 1999 10 3191 3194 10574558
Vuillerme N Danion F Forestier N Nougier V Postural sway under muscle vibration and muscle fatigue in humans Neurosci Lett 2002 333 131 135 12419498 10.1016/S0304-3940(02)00999-0
Bogduk N Cervicogenic headache: anatomic basis and pathophysiologic mechanisms Curr Pain Headache Rep 2001 5 382 386 11403743
Bogduk N The anatomical basis for cervicogenic headache J Manipulative Physiol Ther 1992 15 67 70 1740655
Bogduk N The neck and headaches Neurol Clin 2004 22 151 171 15062532 10.1016/S0733-8619(03)00100-2
Bogduk N The anatomical basis for spinal pain syndromes J Manipulative Physiol Ther 1995 18 603 605 8775022
Bogduk N The anatomy and pathophysiology of neck pain Phys Med Rehabil Clin N Am 2003 14 455 472 12948338
McPartland JM Brodeur RR Hallgren RC Chronic neck pain, standing balance, and suboccipital muscle atrophy – a pilot study J Manipulative Physiol Ther 1997 20 24 29 9004119
Michaelson P Michaelson M Jaric S Latash ML Sjolander P Djupsjobacka M Vertical posture and head stability in patients with chronic neck pain J Rehabil Med 2003 35 229 235 14582555 10.1080/16501970306093
Sjostrom H Allum JH Carpenter MG Adkin AL Honegger F Ettlin T Trunk sway measures of postural stability during clinical balance tests in patients with chronic whiplash injury symptoms Spine 2003 28 1725 1734 12897500 10.1097/00007632-200308010-00018
Jurgens R Mergner T Interaction between cervico-ocular and vestibulo-ocular reflexes in normal adults Exp Brain Res 1989 77 381 390 2792284 10.1007/BF00274995
Skliut IA Likhachev SA Skliut MI Cervico-ocular reflex in healthy humans Aviakosm Ekolog Med 1999 33 46 48 10590809
Thoden U Doerr M Leopold HC Motion perception of head or trunk modulates cervico-ocular reflex (COR) Acta Otolaryngol 1983 96 9 14 6613555
Dutia MB Interaction between vestibulocollic and cervicocollic reflexes: automatic compensation of reflex gain by muscle afferents Prog Brain Res 1988 76 173 180 3064144
Keshner EA Hain TC Chen KJ Predicting control mechanisms for human head stabilization by altering the passive mechanics J Vestib Res 1999 9 423 434 10639027
Peterson BW Choi H Hain T Keshner E Peng GC Dynamic and kinematic strategies for head movement control Ann NY Acad Sci 2001 942 381 393 11710479
Peterson BW Goldberg J Bilotto G Fuller JH Cervicocollic reflex: its dynamic properties and interaction with vestibular reflexes J Neurophysiol 1985 54 90 109 3162006
Allum JH Gresty M Keshner E Shupert C The control of head movements during human balance corrections J Vestib Res 1997 7 189 218 9178224 10.1016/S0957-4271(97)00029-3
Boyle R Vestibulospinal control of reflex and voluntary head movement Ann NY Acad Sci 2001 942 364 380 11710478
Colebatch JG Di Lazzaro V Quartarone A Rothwell JC Gresty M Click-evoked vestibulocollic reflexes in torticollis Mov Disord 1995 10 455 459 7565826 10.1002/mds.870100409
Colebatch JG Rothwell JC Motor unit excitability changes mediating vestibulocollic reflexes in the sternocleidomastoid muscle Clin Neurophysiol 2004 115 2567 2573 15465446 10.1016/j.clinph.2004.06.012
Gdowski GT McCrea RA Integration of vestibular and head movement signals in the vestibular nuclei during whole-body rotation J Neurophysiol 1999 82 436 449 10400970
Ito Y Corna S von Brevern M Bronstein A Rothwell J Gresty M Neck muscle responses to abrupt free fall of the head: comparison of normal with labyrinthine-defective human subjects J Physiol 1995 489 911 916 8788954
Pettorossi VE Santarelli R Draicchio F Errico P Ferreresi A Gravity load related asymmetries in the sagittal vestibulo-collic reflex Acta Otolaryngol 1993 113 239 243 8517120
Uchino Y Role of cross-striolar and commissural inhibition in the vestibulocollic reflex Prog Brain Res 2004 143 403 409 14653183
Watson SR Colebatch JG Vestibulocollic reflexes evoked by short-duration galvanic stimulation in man J Physiol 1998 513 587 597 9807006 10.1111/j.1469-7793.1998.587bb.x
Welgampola MS Colebatch JG Vestibulocollic reflex: normal values and the effect of age Clin Neurophysiol 2001 112 1971 1979 11682335 10.1016/S1388-2457(01)00645-9
Welgampola MS Colebatch JG Vestibulospinal reflexes: quantitative effects of sensory feedback and postural task Exp Brain Res 2001 139 345 353 11545473 10.1007/s002210100754
Wilson VJ Boyle R Fukushima K Rose PK Shinoda Y Sugiuchi Y Uchino Y The vestibulocollic reflex J Vestib Res 1995 5 147 170 7627376 10.1016/0957-4271(94)00035-Z
Wilson VJ Schor RH The neural substrate of the vestibulocollic reflex. What needs to be learned Exp Brain Res 1999 129 483 493 10638422 10.1007/s002210050918
Wilson VJ Vestibulospinal reflexes and the reticular formation Prog Brain Res 1993 97 211 217 8234748
Wilson VJ Yamagata Y Yates BJ Schor RH Nonaka S Response of vestibular neurons to head rotations in vertical planes. III. Response of vestibulocollic neurons to vestibular and neck stimulation J Neurophysiol 1990 64 1695 1703 2074457
Chambers BR Mai M Barber HO Bilateral vestibular loss, oscillopsia, and the cervico-ocular reflex Otolaryngol Head Neck Surg 1985 93 403 407 3927239
Bronstein AM Hood JD The cervico-ocular reflex in normal subjects and patients with absent vestibular function Brain Res 1986 373 399 408 3487371 10.1016/0006-8993(86)90355-0
Heimbrand S Bronstein AM Gresty MA Faldon ME Optically induced plasticity of the cervico-ocular reflex in patients with bilateral absence of vestibular function Exp Brain Res 1996 112 372 380 9007539 10.1007/BF00227943
Bronstein AM Mossman S Luxon LM The neck-eye reflex in patients with reduced vestibular and optokinetic function Brain 1991 114 1 11 1998877
Schubert MC Das V Tusa RJ Herdman SJ Cervico-ocular reflex in normal subjects and patients with unilateral vestibular hypofunction Otol Neurotol 2004 25 65 71 14724495 10.1097/00129492-200401000-00013
Rijkaart DC van der Geest JN Kelders WP de Zeeuw CI Frens MA Short-term adaptation of the cervico-ocular reflex Exp Brain Res 2004 156 124 128 15014925 10.1007/s00221-004-1878-1
Gresty MA Bronstein AM Mergner T, Hlavacka F Control of human head posture Multisensory Control of Posture 1995 New York: Plenum Press
Bullinger A Cognitive elaboration of sensorimotor behavior Infancy and epistemology: An evaluation of Piaget's theory 1981 London: Harvester Press
Butterworth G Hicks L Visual proprioception and postural stability in infancy: A developmental study Perception 1977 6 255 262 866081
Prechtl HFR Prenatal motor development Motor development in children: Aspects of coordination and control 1986 Boston:Martinus Nijhoff
Pope MJ Visual proprioception in infant postural development PhD dissertation 1986 University of Southampton
Woollacott M Debu B Mowatt M Neuromuscular control of posture in the infant and child J Motor Behav 1987 19 167 186
Lee DN Aronson E Visual proprioceptive control of standing in human infants Percept Psychophys 1974 15 529 532
Lewit K Muscular and articular factors in movement restriction Manual Medicine 1985 1 83 85
Koga K Gravity cue has implicit effects on human behavior Avait Space Environ Med 2000 71 A78 A86
Kapandji IA The Physiology of the Joints: The Trunk and the Vertebral Column 1974 3 New York: Churchill Livingstone
Bagnall KM Harris PF Jones P A radiographic study of the human fetal spine. I: The development of the secondary cervical curvature J Anat 1977 123 777 782 885790
Panattoni GL Todros T Fetal motor activity and spine development Panminerva Med 1989 31 183 186 2699021
Vannerson JF Specificity and the law of facilitation in the nervous system Receptor 1971 2 1
Nimo RL Technique for immediate relief of headache and neck pain, including whiplash Receptor 1986 2 1 15
Harbst KB Wilder PA Neurophysiologic, motor control, and motor learning basis of closed kinetic chain exercise Orthop Phys Ther Clin N Am 2000 9 137 149
Baroni G Ferrigno G Rabuffetti M Pedotti A Massion J Long-term adaptation of postural control in microgravity Exp Brain Res 1999 128 410 416 10501814 10.1007/s002210050862
Clement G Gurfinkel VS Lestienne F Lipshits MI Popov KE Adaptation of postural control to weightlessness Exp Brain Res 1984 57 61 72 6519230
Young LR Jackson DK Groleau N Modestino S Multisensory integration in microgravity Ann N Y Acad Sci 1992 656 340 353 1599154
Clement G Lestienne F Adaptive modifications of postural attitude in conditions of weightlessness Exp Brain Res 1988 72 381 389 3224649
Clarke AH Grigull J Scherer H The three-dimensional vestibulo-ocular reflex during prolonged microgravity Exp Brain Res 2000 134 322 334 11045357 10.1007/s002210000476
Jennings RT Managing space motion sickness J Vestib Res 1998 8 67 70 9416592 10.1016/S0957-4271(97)00042-6
Massion J Postural control system Curr Opin Neurobiol 1994 4 877 887 7888772 10.1016/0959-4388(94)90137-6
Gurfinkel VS The mechanisms of postural regulation in man Sov Sci Rev F Phys Gen Biol 1994 7 59 89
Gurfinkel VS Levik YuS Popov KE Smetanin BN Gurfinkel VS, Ioffe ME, Massion J, Roll JP Body scheme in the control of postural activity Stance and motion: facts and concepts 1988 Plenum Press, New York
Amblard B Assaiante C Fabre JC Mouchino L Massion J Voluntary head stabilization in space during oscillatory trunk movements in the frontal plane performed in weightlessness Exp Brain Res 1997 114 214 225 9166911
Fisk J Lackner JR DiZio P Gravitoinertial force level influences arm movement control J Neurophysiol 1993 69 504 511 8459281
Glasauer S Mittelstaedt H Perception of spatial orientation in microgravity Brain Res Brain Res Rev 1998 28 185 193 9795210 10.1016/S0165-0173(98)00038-1
Takahashi M Sekine M Ikeda T Watanuki K Hakuta S Takeoka H Effect of microgravity on spatial orientation and posture regulation during coriolis stimulation Acta Otolaryngol 2004 124 495 501 15224882 10.1080/00016480410016315
Amblard B Assaiante C Vaugoyeau M Baroni G Ferrigno G Pedotti A Voluntary head stabilization in space during oscillatory trunk movements in the frontal plane performed before, during, and after a prolonged period of weightlessness Exp Brain Res 2001 137 170 179 11315545 10.1007/s002210000621
Yakushin SB Palla A Haslwanter T Bockisch CJ Straumann D Dependence of adaptation of the human vertical angular vestibulo-ocular reflex on gravity Exp Brain Res 2003 152 137 142 12879171 10.1007/s00221-003-1543-0
Massion J Movement, posture, and equilibrium: interaction and coordination Progr Neurobiol 1992 38 35 56 1736324 10.1016/0301-0082(92)90034-C
Lestienne F Gurfinkel VS Postural control in weightlessness: a dual process underlying adaptation to an unusual environment Trends Neurosci 1988 11 359 363 2469196 10.1016/0166-2236(88)90058-6
Inglis JT Kennedy PM Wells C Chua R Gandevia SC, Proske U, Stuart DG The role of cutaneous receptors in the foot Sensorimotor Control of Movement and Posture 2002 Kluwer Academic/Plenum Publishers
Jacobs R Burleigh-Jacobs A Winters JM, Crago PE Neuromuscular control strategies in postural coordination Biomechanics and Neural Control of Posture and Movement 2000 New York: Springer-Verlag
Clement G Moore ST Raphan T Cohen B Perception of tilt (somatogravic illusion) in response to sustained linear acceleration during space flight Exp Brain Res 2001 138 410 8 11465738 10.1007/s002210100706
Jarchow T Wirz M Haslwanter T Dietz V Straumann D Perceived horizontal body position in healthy and paraplegic subjects: effect of centrifugation J Neurophysiol 2003 90 2973 7 12878718
Vaitl D Mittelstaedt H Saborowski R Stark R Baisch F Shifts in blood volume alter the perception of posture: further evidence for somatic graviception Int J Psychophysiol 2002 44 1 11 11852154 10.1016/S0167-8760(01)00184-2
McIntyre D Ring C Carroll D Effects of arousal and natural baroreceptor activation on the human muscle stretch reflex Psychophysiology 2004 41 954 60 15563348 10.1111/j.1469-8986.2004.00235.x
Moffitt JA Schadt JC Hasser EM Altered central nervous system processing of baroreceptor input following hindlimb unloading in rats Am J Physiol 1999 277 H2272 9 10600846
Ball KA Afheldt MJ Evolution of foot orthotics – part 1: coherent theory or coherent practice? J Manipulative Physiol Ther 2002 25 116 24 11896381 10.1067/mmt.2002.121415
Ball KA Afheldt MJ Evolution of foot orthotics – part 2: research reshapes long-standing theory J Manipulative Physiol Ther 2002 25 125 34 11896382 10.1067/mmt.2002.121416
Moore J Chiropractic in America The History of a Medical Alternative 1993 Baltimore: Johns Hopkins University Press
Meeker WC Haldeman S Chiropractic: a profession at the crossroads of mainstream and alternative medicine Ann Intern Med 2002 136 216 227 11827498
Foreman SM Croft AC Whiplash Injuries: The Cervical Acceleration-Deceleration Syndrome 1995 2 Williams & Wilkins, Baltimore, MD
|
16091134
|
PMC1198239
|
CC BY
|
2021-01-04 16:38:23
|
no
|
Chiropr Osteopat. 2005 Aug 9; 13:16
|
utf-8
|
Chiropr Osteopat
| 2,005 |
10.1186/1746-1340-13-16
|
oa_comm
|
==== Front
Clin Pract Epidemiol Ment HealthClinical Practice and Epidemiology in Mental Health : CP & EMH1745-0179BioMed Central 1745-0179-1-101605352410.1186/1745-0179-1-10ResearchHepatitis B virus infection among inpatients of a psychiatric hospital of Mexico Alvarado Esquivel Cosme [email protected] Valenzuela Miguel Ángel [email protected] Suárez Miguel Francisco [email protected] Andrade Francisco [email protected] Facultad de Medicina, Universidad Juárez del Estado de Durango. Durango, Dgo. México2 Hospital Psiquiátrico Dr. Miguel Vallebueno. Secretaria de Salud. Durango, Dgo. México3 Hospital Regional Dr. Ignacio Téllez. IMSS. Durango, Dgo. México4 Hospital de Nuevo Ideal. Secretaria de Salud. Nuevo Ideal, Dgo. México2005 29 7 2005 1 10 10 12 4 2005 29 7 2005 Copyright ©2005 Esquivel et al; licensee BioMed Central Ltd.2005Esquivel et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Background
The epidemiology of the hepatitis B virus (HBV) infection in psychiatric patients from developing countries is poorly studied. Therefore, we sought to determine the frequency of HBV surface antigen (HBsAg) and HBV surface antibody (HBsAb) serological markers of HBV infection in a population of patients of a psychiatric hospital in Durango City, Durango, Mexico, and to determine whether there are any epidemiological characteristics of the subjects associated with the infection.
Methods
Out of 150 patients of the psychiatric hospital of Durango City, 99 were examined for HBsAg and HBsAb by AUSZYME MONOCLONAL (Abbott Laboratories, Abbott Park, IL, USA) assay and AUSAB (Abbott Laboratories, Abbott Park, IL, USA) assay, respectively. Epidemiological data from each participant was also obtained. For comparison purposes, 2505 blood donors were examined for HBsAg seropositivity.
Results
Out of the 99 patients studied, twelve showed serological evidence of HBV infection (12.1%); 7 of them (7.1%) were positive for HBsAg, and 5 (5.1%) were positive for HBsAb. Out of the 2505 blood donors, 2 (0.0008%) were HBsAg positive. Seropositivity to HBV markers was associated with an age of 45 years and older (OR = 4.27; 95%CI = 1.02–18.78). Other characteristics as gender, number of hospitalizations, duration of the last hospitalization, and clinical diagnosis were not associated with seropositivity to HBV infection markers. Patients showed a significantly higher HBsAg seropositivity than blood donors (p < 0.0000001)
Conclusion
HBV was found to be an important infectious agent in the Mexican psychiatric inpatient population studied. Health care strategies for prevention and control of HBV infection in psychiatric hospitals should pay special attention to patients aged forty-five years and older.
==== Body
Background
Hepatitis B virus (HBV) infections are an important cause of morbidity and mortality worldwide. More than 400,000,000 persons are chronically infected by HBV [1,2]. Moreover, the virus is responsible for more than 300,000 cases of liver cancer every year and for similar numbers of gastrointestinal haemorrhage and ascites [2]. In the United States, chronic HBV infection is responsible for about 5,000 annual deaths from cirrhosis and hepatocellular carcinoma [1]. In Latin America, statistical estimates of HBV-related morbidity showed that greater than 150,000 acute HBV cases occur per year [3].
Reports indicate that patients with mental illnesses have a high prevalence of HBV infection [4-7], and the prevalence of this infection in psychiatric patients varies substantially in different countries. For instance, in a Brazilian study, 22.4% of the psychiatric patients studied were positive for HBV infection [8]. In European countries, a study carried out in Spain showed that HBV seropositivity in mentally handicapped adults was 81.3% [9]. In contrast, in Northern Ireland, only 4.5% of the residents of an institution for the mentally handicapped carried any HBV marker [10]. In the U.S.A., a study performed on people with severe mental illness showed a 23.4% prevalence rate of HBV infection [11]. In Asian countries, a Taiwanese study reported that 18.1% of institutionalized psychiatric patients were positive for HBV surface antigen (HBsAg) [12]. Furthermore, a study performed on psychiatric inpatients in Singapore showed that 12.7% were positive for HBsAg, 63.4% were positive for HBV surface antibody (HBsAb), and 69% were positive for HBV core antibody [13].
There is a lack of information concerning the epidemiology of HBV infection in psychiatric patients in Mexico. Therefore, we conducted a cross-sectional, prospective, descriptive survey in order to determine the frequency of HBsAg and HBsAb in a population of inpatients of a psychiatric hospital of Durango, a northern Mexican city. In addition, we sought to determine whether any epidemiological data of the patients correlated with HBV infection.
Methods
Study population
This study was performed in a psychiatric hospital of Durango City, Durango, Mexico during the years 1995 and 1996. A sample size calculation was performed based on the reported prevalence of HBV infection in a comparable Latin-American population [8]. A 99% confidence level sample size of 74 patients was obtained. From a total of 150 inpatients, 105 were randomly chosen and invited to participate in the study. Out of the 105 chosen inpatients, 3 were excluded since they were severely ill and unable to decide to participate in the study. Of the remaining 103 chosen inpatients, 99 agreed to participate. This study was evaluated and accepted by the Institutional Ethical Committee. A written informed consent was obtained from all the individuals participating in the study. For comparison purposes of HBsAg seropositivity, we studied 2505 blood donors that attended a local blood bank during the same study period.
Serology for HBsAg and HBsAb
Sera, from the 99 patients were analyzed for HBsAg and HBsAb by AUSZYME MONOCLONAL (Abbott Laboratories, Abbott Park, IL, USA) assay and AUSAB (Abbott Laboratories, Abbott Park, IL, USA) assay, respectively. Sera, from blood donors, were analyzed for HBsAg only by using AUSZYME MONOCLONAL (Abbott Laboratories, Abbott Park, IL, USA) assay.
Epidemiological data
Epidemiological data including age, gender, diagnosis, number of hospitalizations, duration of last hospitalization, and history of drug use, blood transfusion, sexual promiscuity, and surgery from all ninety-nine patients studied were obtained. Data was obtained from the patients, medical examination records, and informants. Classification of mental illnesses was performed according to the ICD-10 criteria [14].
Statistical analysis
Sample size calculation as well as analysis of results were performed with the aid of the software Epi Info 6. To assess the association between the characteristics of the subjects and the disease, the crude odds ratio with a 95% exact confidence interval was used. We calculated the exact confidence interval because the cell value (number of cases) was less than 5 in some comparisons. For comparison of the frequencies among the groups, the Fisher exact test was used. A p value of less than 0.05 was considered significant.
Results
Serology for HBsAg and HBsAb
Out of the ninety-nine patients studied, twelve (12.1%) showed serological evidence of HBV infection. HBsAg was detected in seven patients (7.1%) while HBsAb was detected in five patients (5.1%).
Out of the 2505 blood donors studied, 2 (0.0008%) were positive for HBsAg. This frequency of HBsAg seropositivity in patients was significantly higher than that found in blood donors (p < 0.0000001).
Epidemiological data
Of the ninety-nine psychiatric patients studied, thirty were women and sixty-nine were men. The mean age was 39.4 years (range: 18 to 80 years). We were able to trace the number of hospitalizations and duration of the last hospitalization in ninety-three out of the ninety-nine patients. Of these patients, eighty two had been hospitalized once while eleven patients had been hospitalized from two to five times. The mean duration of the last hospitalization was 7.5 years (range: 0.1 to 36 years). Patients suffered from a number of mental illnesses including schizophrenia (F20–29, n = 35), organic delusional [schizophrenia-like] disorder (F06.2, n = 23), mental retardation (F70–79, n = 18), alcohol abuse (F10, n = 5), drug dependence (F19, n = 4), dementia of the Alzheimer type (F00, n = 2), vascular dementia (F01.0-1, n = 2), depression (F32, n = 2), personality disorder (F60.9, n = 1), and hyperkinetic disorders (F90, n = 1). In six patients diagnosis was not available. Of the participants, four (4.0%) had risk factors for hepatitis B: three had received blood transfusions, and one had sexual promiscuity; eleven more patients were drug users, but they denied the use of intravenous drugs. Similarly, patients denied any history of surgery. The twelve patients with HBV positive markers had a mean age of 42.7 years (range: 18 to 65 years), nine were male and three female. Separately, HBsAg positive patients had a mean age of 39.6 years (range: 18 to 65 years), and HBsAb positive patients had a mean age of 47.0 years (range: 25 to 56 years). Eleven patients had been hospitalized once, and one patient three times. The mean duration of their hospitalizations was 9.7 years (range: 0.1 to 35 years). Out of the twelve patients, two were drug users. Furthermore, of the 12 patients with HBV positive markers, three suffered from organic delusional [schizophrenia-like] disorder, four schizophrenia, three mental retardation, one vascular dementia, and in one patient the diagnosis was not available. Table 1 shows the main epidemiological data of the patients and its relation to HBV seropositivity. The patients with negative results for HBV markers had a mean age of 38.9 years (range: 18 to 80 years), of whom, sixty were male and twenty-seven were female. We were able to obtain the number of hospitalizations in eighty-one out of the eighty seven HBV negative patients. Of these, seventy-one of them had been hospitalized once, while ten had been hospitalized from two to five times. The mean duration of their hospitalization was 7.1 years (range: 0.1 to 36 years). Out of the eighty-seven HBV negative subjects, nine were drug users, three had received blood transfusions and one had sexual promiscuity. Of these patients, thirty-one suffered from schizophrenia, twenty organic delusional [schizophrenia-like] disorder, fifteen mental retardation, five alcohol abuse, four drug dependence, two dementia of the Alzheimer type, one vascular dementia, two depression, one personality disorder, one hyperkinetic disorder, and in five patients, the diagnosis was not available. Table 2 shows a comparison between the clinical diagnosis in patients with positive HBV markers and patients with negative HBV markers.
Table 1 Comparison of epidemiological characteristics of the psychiatric patients with positive HBV markers and the psychiatric patients with negative HBV markers.
Epidemiological Characteristics Patients with positive markers N = 12 (%) Patients with negative markers n = 87 (%)c ORa 95% ECIb
Age
less than 45 years 5 (41.7%) 58 (71.6%)
45 years and older 7 (58.3%) 23 (28.4%) 4.27 1.02–18.78
Gender
Female 3 (25.0%) 27 (31.0%)
Male 9 (75.0%) 60 (69.0%) 1.35 0.30–8.33
Number of hospitalizations
One 11 (91.7%) 71 (87.7%)
More than one 1 (8.3%) 10 (12.3%) 0.65 0.01–5.45
Duration of last hospitalization
Equal or less than 5 years 5 (41.7%) 46 (56.8%)
More than 5 years 7 (58.3%) 35 (43.2%) 1.84 0.46–7.96
History of blood transfusion 0 (0.0%) 3 (3.4%) 0 0.00–18.39
History of drug use 2 (16.7%) 9 (10.3%) 1.73 0.16–10.25
aOR: Odds ratio.
bECI: Exact confidence interval.
cSome data were available in only 81 subjects.
Table 2 Comparison of predominant clinical diagnosis of the psychiatric patients with positive HBV serological markers and the psychiatric patients with negative HBV serological markers.
Clinical Diagnosis Patients with positive markers Patients with negative markers ORa 95% ECIb
Organic delusional disorder 3 (27.3%) 20 (24.4%) 1.16 0.18–5.46
Schizophrenia 4 (36.4%) 29 (35.4%) 1.04 0.21–4.52
Mental retardation 2 (18.2%) 13 (15.9%) 1.18 0.11–6.71
aOR: Odds ratio.
bECI: Exact confidence interval.
Discussion
In this study, 12.1% of the psychiatric inpatients showed seropositivity to either HBsAg or HBsAb. In addition, we found a remarkable and significantly higher HBsAg seropositivity in patients than in blood donors. The frequency of HBsAg seropositivity found in our psychiatric population was also much higher than those reported in Mexican blood donors [15-17]. When comparing our results with those reported in similar psychiatric populations of other countries, our frequencies were only comparable with that found in northern Ireland [10]. However, our figure was much lower than those reported in Brazil [8], the U.S.A. [11], Spain [9], Singapore [13], and Taiwan [12]. A likely explanation for this finding could be differences in the characteristics of the patients studied. For instance, intravenous drug use, an important risk factor, which was not present in our patients studied. Similarly, other risk factors such as blood transfusion and sexual promiscuity were only present in three and one of our patients, respectively. When the characteristics of our patients were analyzed for any association with HBV infection, we observed that an age forty-five years and older were associated with HBV infection (OR = 4.27; 95% CI = 1.02–18.78). A possible explanation of this finding is that the longer the life of a subject, the higher the possibility to contract the virus. Age of psychiatric patients has been correlated with HBV infection [6], but our result conflicts with the lower HBsAg carrier rate in the aged found by Chang et al [12]. Other characteristics of the patients such as gender, clinical diagnosis, number of hospitalizations, duration of last hospitalization, history of drug use, blood transfusion, sexual promiscuity, and surgery were not associated with HBV infection. HBV infection was slightly lower in females than males in our study, and this result agrees with that reported by Chang et al [12]. With respect to clinical diagnosis, risk groups for HBV infection include chronic schizophrenia patients [18], and institutionalized psychotic patients [19]. Nevertheless, frequencies of schizophrenia patients and psychotic patients were not statistically different in HBV positive and HBV negative patients of our study. Likewise, although a long duration of hospitalization has been correlated with HBV infection [9], our HBV positive patients had a slightly longer but not statistically significant mean duration of hospitalization than the HBV negative patients (9.7 years vs. 7.2 years). This result agrees with that reported by Gmelin et al [12].
A high prevalence of HBV infection does not only occur in patients but also in personnel of psychiatric hospitals [20-22]. Therefore, our results highlight the need to take action to prevent and control HBV infection in psychiatric hospitals in Mexico. Immunization is by far the single most effective prevention measure for HBV infection [23]. Hepatitis B vaccine led to a decline in the incidence of hepatitis D virus [24], and reduces incidence of liver cancer [2]. Our results may be used as a guide when the hepatitis B vaccination is being considered for use in a psychiatric hospital. Besides vaccination programs, further recommendations include health education, HBV serologic screening of patients and personnel, and the elimination of overcrowding in psychiatric hospitals.
Conclusion
HBV is an important infectious agent in psychiatric inpatients in Durango, Mexico. Nevertheless, the incidence of HBV infection found is lower than the majority of those reported in other countries. HBV infection was associated with an age of forty-five years and older.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
Author 1 (CAE) conceived and designed the study protocol, participated in the coordination and management of the study, performed the data analysis and wrote the manuscript. Author 2 (MAAV) obtained the general data of the patients, and performed the clinical evaluation of the patients.
Author 3 (MFMS) analyzed blood donors, and performed the data analysis.
Author 4 (FEA) obtained the general data of the patients, and performed the data analysis.
Acknowledgements
The excellent assistance of Miss Maria del Rosario Alonso Almeida is greatly appreciated.
==== Refs
Lin KW Kirchner JT Hepatitis B Am Fam Physician 2004 69 75 82 14727820
Lai CL Ratziu V Yuen MF Poynard T Viral hepatitis B Lancet 2003 362 2089 94 10.1016/S0140-6736(03)15108-2 14697813
Fay OH Hepatitis B in Latin America: epidemiological patterns and eradication strategy. The Latin American Regional Study Group Vaccine 1990 8 Suppl S100 6 2139277
Feng CS Prevalence of hepatitis B in an adult psychiatric hospital J Am Geriatr Soc 1982 30 326 8 7077009
Franson TR Ksobiech LJ Simonsen HW Prevalence of hepatitis B carriers in a mental health in-patient facility: implications for employee screening and vaccination Psychiatr Hosp 1986 17 81 3 10278763
Gmelin K Doerr HW Middelhoff H von Ehrlich B Sann G Theilmann G Kommerell B Hepatitis markers in a psychiatric institution Dev Biol Stand 1983 54 545 8 6653899
Prats F Porta Serra M Yazbeck H Herrera R Gasso JM The prevalence of serological markers for the human immunodeficiency virus and the hepatitis B virus in a psychiatric hospital Gac Sanit 1990 4 179 83 2286497
de Souza MM Barbosa MA Borges AM Daher RR Martins RM Cardoso DD Seroprevalence of hepatitis B virus infection in patients with mental problems Rev Bras Psiquiatr 2004 26 35 8 15057838
Asensio F Bayas JM Bertran MJ Asenjo MA Prevalence of hepatitis B infection in long-stay mentally handicapped adults Eur J Epidemiol 2000 16 725 9 10.1023/A:1026793900057 11142500
Kee F McGinnity M Marriott C Calvert GJ Shanks OE O'Neill H Connolly J Hepatitis B screening in a northern Irish mental handicap institution: relevance to hepatitis B vaccination J Hosp Infect 1989 14 227 32 10.1016/0195-6701(89)90039-X 2575104
Rosenberg SD Goodman LA Osher FC Swartz MS Essock SM Butterfield MI Constantine NT Wolford GL Salyers MP Prevalence of HIV, hepatitis B, and hepatitis C in people with severe mental illness Am J Public Health 2001 91 31 7 11189820
Chang TT Lin H Yen YS Wu HL Hepatitis B and hepatitis C among institutionalized psychiatric patients in Taiwan J Med Virol 1993 40 2 170 3 8360635
Tey BH Oon CJ Kua EH Kueh YK Wong YW Chin JH Prevalence of hepatitis B markers in psychiatric in-patients in Singapore: a pilot study Ann Acad Med Singapore 1987 16 608 11 3446002
Clasificación Estadística Internacional de Enfermedades y Problemas relacionados con la Salud 1995 1–3 10 Organización Panamericana de la Salud. Washington, USA.
Carreto-Velez MA Carrada-Bravo T Martinez-Magdaleno A Seroprevalence of HBV, HCV, and HIV among blood donors in Irapuato, México Salud Publica Mex 2003 45 Supp 5 S690 3 14974281
Mendez-Sanchez N Baptista-Gonzalez H Sanchez-Gomez RH Bordes-Aznar J Uribe-Esquivel M The prevalence of hepatitis B and C in blood donors in a 3rd-level hospital of Mexico City Salud Publica Mex 1999 41 475 8 10634078
Alvarez-Munoz MT Bustamante-Calvillo ME Guiscafre-Gallardo JP Munoz O Hepatitis B and delta: the prevalence of seroepidemiological markers in volunteer blood donors and their families Gac Med Mex 1991 127 399 404 1790848
Said WM Saleh R Jumaian N Prevalence of hepatitis B virus among chronic schizophrenia patients East Mediterr Health J 2001 7 526 30 12690775
Chaudhury S Chandra S Augustine M Prevalence of Australia antigen (HBsAg) in institutionalised patients with psychosis Br J Psychiatry 1994 164 542 3 8038945
Di Nardo V Petrosillo N Ippolito G Bonaventura ME Puro V Chiaretti B Tosoni M Prevalence and incidence of hepatitis B virus, hepatitis C virus and human immunodeficiency virus among personnel and patients of a psychiatric hospital Eur J Epidemiol 1995 11 239 42 10.1007/BF01719496 7672084
Almi P Toscano L Rubino M Toti M Galluzzi P Epidemiology of hepatitis B virus infection in the personnel of a psychiatric hospital Minerva Med 1989 80 1011 4 2530470
Ares Camerino A Terron Pernia A Sainz Vera B Mira Gutierrez J Rodríguez Iglesias M Zafra Mezcua J Jesus de la Calle I Prevalence of hepatitis B markers in the personnel of psychiatric hospitals Rev Clin Esp 1989 184 16 9 2704866
Alter MJ Epidemiology and prevention of hepatitis B Semin Liver Dis 2003 23 39 46 10.1055/s-2003-37583 12616449
Geller SA Hepatitis B and hepatitis C Clin Liver Dis 2002 6 317 34 10.1016/S1089-3261(02)00008-9 12122858
|
16053524
|
PMC1198240
|
CC BY
|
2021-01-04 18:01:01
|
no
|
Clin Pract Epidemiol Ment Health. 2005 Jul 29; 1:10
|
utf-8
|
Clin Pract Epidemiol Ment Health
| 2,005 |
10.1186/1745-0179-1-10
|
oa_comm
|
==== Front
Epidemiol Perspect InnovEpidemiologic perspectives & innovations : EP+I1742-5573BioMed Central London 1742-5573-2-81605352210.1186/1742-5573-2-8Analytic PerspectiveCausal thinking and causal language in epidemiology: it's in the details Lipton Robert [email protected]Ødegaard Terje [email protected] Research Scientist, Prevention Research Center, 1995 University Ave. Suite 450, Berkeley, CA 94704, USA2 Faculty of Health and Social Work, Lillehammer University College, N-2626, Lillehammer, Norway2005 29 7 2005 2 8 8 24 9 2004 29 7 2005 Copyright © 2005 Lipton and Ødegaard; licensee BioMed Central Ltd.2005Lipton and Ødegaard; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Although epidemiology is necessarily involved with elucidating causal processes, we argue that there is little practical need, having described an epidemiological result, to then explicitly label it as causal (or not). Doing so is a convention which obscures the valuable core work of epidemiology as an important constituent of public health practice. We discuss another approach which emphasizes the public health "use value" of research findings in regard to prediction and intervention independent from explicit metaphysical causal claims. Examples are drawn from smoking and lung cancer, with particular focus on the original 1964 Surgeon General's report on smoking and the new version released in 2004. The intent is to help the epidemiologist focus on the pertinent implications of research, which, from a public health point of view, in large part entails the ability to predict and to intervene. Further discussion will center on the importance of differentiating between technical/practical uses of causal language, as might be used in structural equations or marginal structural modeling, and more foundational notions of cause. We show that statistical/epidemiological results, such as "smoking two packs a day increases risk of lung cancer by 10 times" are in themselves a kind of causal argument that are not in need of additional support from relatively ambiguous language such as "smoking causes lung cancer." We will show that the confusion stemming from the use of this latter statement is more than mere semantics. Our goal is to allow researchers to feel more confident in the power of their research to tell a convincing story without resorting to metaphysical/unsupportable notions of cause.
==== Body
Introduction
Causal thinking and causal language in epidemiology
A primary goal of epidemiological research is the ability to determine how exposures are related to outcomes. We are interested, at the population level, in what caused the cancer, the heart attack, the cholera epidemic or the food poisoning. Our methods have developed rapidly over the last four decades to account for, among other things, confounders, retrospective and longitudinal data, and bias. In an effort to systematize the causal enterprise, similar to efforts in other relatively young fields of scientific inquiry, epidemiologists have sought to tie such methods to an overarching causal rubric such as Popperian falsification, Mill's analysis of causation in terms of necessity and sufficiency, ceteris paribus conditions/control of confounding [1] and/or counterfactuals. Such efforts, while being very useful in advancing the field and providing guidance for understanding exposure and outcome relationships, have tended to ignore the claim of Hume, among other writers, that causal connections cannot be observed or objectively proven. Thus, on the one hand, a great deal of effort is spent to carefully develop methods aimed at revealing causal relationships, while on the other we are being told – rather persuasively – that we cannot ultimately determine causal relationships, or that we should refrain from attempts to establish causal relationships because these should be understood as different from nomological or probabilistic relations. Obviously, these tensions have not stopped scientific, let alone epidemiological, efforts from proceeding apace (nor should they).
Nevertheless, fundamental issues bearing on how the relationship between exposures and outcomes are assessed, interpreted and discussed, are left more ambiguous than necessary. And to be clear, this is not just a theoretical issue, since such ambiguity allows for real world problems to arise that, with a small amount of care, are easily avoided. In this essay, we will explore some limitations on obtaining causal information, and on how such epidemiological information should be disseminated, both to lay and professional audiences, in a more useful and less confusing manner than is often presented. The goal of our argument is to invite a less anxious and more humble, yet forceful, approach toward assessing epidemiological research. This approach will show that the process of examining exposures and outcomes is the important factor, in service to prediction and intervention, not an illusory ability to identify and articulate apparently more fundamental causal connections.
Analysis
Public health issues and causation-speak in the 2004 Surgeon General's report
Among the possible reasons so much has been written about causation and epidemiology is that in significant ways epidemiology is a science, and as such is definitionally interested in causation. If there is a shared discourse in epidemiology as a field it revolves around the manner in which exposures are related to outcomes in service to analyzing truly pressing public health issues.
Recently, a new Surgeon General's report on smoking, responding to and expanding on the original 1964 report, included a section explicitly discussing issues of causal claims and providing guidelines for determining the strength of causal relations [2]. The new report quotes the original 1964 report: "after vigorous discussions they could neither precisely define nor replace the word 'cause,' a reflection of the same problem that philosophers have confronted over the centuries." Further, the 1964 report noted that
when a relationship or an association between smoking ... and some condition in the host was noted, the significance of the association was assessed. The characterization of the assessment called for a specific term .... The word cause is the one in general usage in connection with matters considered in this study, and it is capable of conveying the notion of a significant, effectual relationship between an agent and an associated disorder or disease in the host. Granted that these complexities were recognized it is to be noted clearly that the Committee's considered decision to use the words ' "a cause" or "a major cause" or "a significant cause" or "a causal association" ' in certain conclusions about smoking and health affirms their conviction. [3] (p. 21)
The authors of the new report correctly point out that while the original report is quite useful and serves as one of the most important examples of comprehensive assessment of exposures and outcomes in public health history, there is some level of confusion associated with the language of causation. Indeed, the 1964 report is clearly struggling to articulate and justify its use of causal terms; e.g., in the passage quoted, the strained language of the last sentence is revealing. Circularity threatens when a choice of terminology for the purposes of describing one's findings is justified in part by a wish to "affirm convictions" in regard to the findings. Are we to suppose that the findings do not speak sufficiently eloquently for themselves?
In an attempt to address and even regiment the use of causation-speak, the authors of the new report, in addition to providing a very comprehensive list of causal statements related to smoking and health outcomes, discuss what they consider to be a less confusing approach towards using causal language and ascribing cause in epidemiology. Unfortunately, their efforts, while clearly useful as a guide to assessing possible implications of research, beg the question of whether explicit causal language is really needed in presenting and discussing research in the first place. Other begged questions concern how such causal language is necessarily linked to the substantive research and how hierarchies of causal strength are to be determined.
Starting on page 11 of the new report [2], their careful listing of causal statements from previous reports is strikingly idle in view of the fact that in many of the statements in the new report there is no explicit use of the word "cause" when these succinctly describe the current state of research. Examples are: "Autopsy studies suggest that cigarette smoking is associated with a significant increase in arteriosclerosis of the aorta and coronary arteries," "Recent autopsy studies confirm that pulmonary emphysema is much more frequent and severe in cigarette smokers than nonsmokers," or "Women cigarette smokers experience an increased risk for subarachnoid hemorrhage."
These statements can be contrasted with others such as "It is also more prudent to assume that the established association between cigarette smoking and coronary disease has causative meaning than to suspend judgment until no uncertainty remains" (p. 13, our emphasis). The discomfort on display in the last sentence is clear: responsible prudence apparently dictates the use of explicit causal language even though the findings, on their own, i.e. "established association" can be taken to be a meaningful statement of cause (more on this below). There is, further, an implied and somewhat ambiguous assumption that complete certainty, although not available here, is theoretically achievable, yet not needed, for causal information to be conveyed. Although this latter point is one with which we can strongly agree, we nevertheless argue that worrying about complete certainty is not useful for the simple reason that this level of certainty is not available. Indeed, if the information obtained from the "established associations" allows for effective prediction and/or intervention, then it is not clear what other information or language would be needed in terms of causal argumentation. Thus, the above remark seems to be intended to justify and/or motivate health policy (not necessarily a bad thing but off-point for the purposes of this discussion) rather than to improve our causal understanding of the relationship between smoking and coronary disease.
One problem is that, as Hume described and the authors confirm [2,4], while the use of causal language can be psychologically compelling, the causal nexus will never lend itself to be empirically detected or generally proven. Particularly in regard to the last statement from the Surgeon General's report, uncertainty will always remain. This uncertainty can be thought of as supporting a more probabilistic approach towards causation [5]. Parascandola and Weed point out that probabilistic models of causation are essentially more flexible than deterministic approaches. Their argument centers on the idea that since it is impossible to ever know all the constituent elements in a deterministic causal model, why not allow for some level of probabilistic ambiguity? The need to say anything definitive about this dichotomy, however, is not in the realm of the strictly scientific, nor is their discussion of what constitutes science and what constitutes public health policy, and why different notions of science might apply to the two. Although these are different contexts, the inability to "prove" or objectively "see" causation, however, still applies to both.
More importantly, a fundamental inability to determine cause is not necessarily a serious concern for epidemiologists because causal information can still be conveyed without getting bogged down in such epistemological and metaphysical issues. Thus, the struggle to develop a causal context relating tobacco to illness in the new report displays a level of anxiety that would be unnecessary if a more pragmatic approach toward causal information were used.
A short caveat on realism in science
We are not in this essay attempting to revisit the long-standing debate between realism and pragmatism in science. (A relatively current, although ultimately unconvincing, exploration of realism in epidemiology was discussed by Renton [6].) For the sake of making our argument, we accept the natural ontological attitude (NOA) developed by Arthur Fine [7] as being closely in line with our approach toward thinking about causation. His argument is, in fact, a generalization of what has been discussed here. That is, he asks what is the efficacy of having something be considered "real" in the same manner as something being determined as "causal." He is interested in the ability to manipulate the world, to predict and intervene. Being able to determine something as real, in a metaphysically emphatic sense, something he and we doubt can ever be accomplished, is beside the point when dealing with the actual process of doing science.
An alternative approach toward causal thinking
Once a famous epidemiologist, K, stated that causation is easy, "smoking causes lung cancer," adding a sarcastic "it's obvious" shrug of his shoulders to emphasize his point. This was in answer to a naïve query regarding how certain we could be about ever saying that X causes Y. K's response was a catalyst to our interest in epidemiologists' use of causal language, both in the day-to-day workings of any particular epidemiological project and in the more extended long-range meta-discussions bearing on causal thinking in epidemiology. It is our suggestion that K's remark, while presumably intended to lend scientific weight to the findings that he had in mind, might rather have done them a disservice.
What, if anything, would underwrite an explicit causal claim, in this kind of context? We shall consider an admittedly not uncontroversial discussion of causality by G.E.M. Anscombe [8]. She identifies a claim shared by received philosophical views about causal connections as being either a kind of necessary connection between events, or as instancing an exceptionless generalization – a universal claim – saying that a certain kind of event will always be preceded by certain others: "If an effect occurs in one case and a similar effect does not occur in an apparently similar case, there must be a relevant further difference." [8] (p. 88)
Versions of the associations of causality with either necessity or universality are found throughout the history of philosophical thinking about the subject – Anscombe mentions Aristotle, Spinoza, Hobbes, Hume, Kant and Russell – although the accounts vary greatly with respect to whether the focus is on necessity or universality. Aristotle, Spinoza and Hobbes go for necessity, with the latter expressing the connection as being logically rather than naturally compelling, and so does Kant, who secured necessity (and also invariability) by introducing the idea of causality as a rule governing our very ability to understand sequences of events. Hume famously saw no necessity at work in the collision between two billiard balls. He described how the potential for experiencing the same thing repeatedly – an experience of "constant conjunction" – provided the basis for an irresistible, species-wide, but ultimately psychological idea of necessity. This was considered an essential but not empirically justifiable part of the complex idea of causality. Hume explained the latter as an upshot of experienced exceptionless generality. (Kant's view was of course motivated by the "scandalousness" of this "too disastrous to be true" position.) Russell, at one stage, argues that it is for universality to explain the notion of causal connections as being necessary. What they all share, however, is that causation is about necessity or universality or both.
Anscombe challenges this fundamental view; i.e., she challenges the shared claim quoted above:
... it's not difficult to show it prima facie wrong to associate the notion of cause with necessity or universality in this way. For, it being much easier to trace effects back to causes with certainty than to predict effects from causes, we often know a cause without knowing whether there is an exceptionless generalization of the kind envisaged, or whether there is a necessity. [8] (pp. 136–137)
Thus if one, for example, has been intimate with someone developing mononucleosis, one might expect to contract it, and if one does one would assume that one knows the cause, but no doctor would venture to bet on one's coming down with the disease if invited to do so before a diagnosis.
What is Anscombe's point? She proposes that we may have causal knowledge without having clarified what is involved in causation, in any of the heavy-duty philosophical senses discussed above:
Compare the possibility of wanting clarification of 'valency' or 'long-run frequency,' which yet have been handled by chemists and statisticians without such clarification; and valencies and long-run frequencies, whatever the right way of explaining them, have been known. Thus one of the familiar philosophic analyses of causality, or a new one in the same line, may be correct, though knowledge of it is not necessary for knowledge of causes. [8] (p. 136, our emphasis)
Moreover, both necessity and universality fail to focus, she argues, on something "so obvious as to seem trite," and proposes to replace the shared feature of accounts of causality given above with the following:
... causality consists in the derivativeness of an effect from its causes. This is...the common feature of causality in its various kinds. Effects derive from, arise out of, come of, their causes. For example, everyone will grant that physical parenthood is a causal relation. Here the derivation is material, by fission. Now analysis in terms of necessity or universality does not tell us of this derivedness of the effect; rather, it forgets about that. For the necessity will be that of the laws of nature; through it we shall be able to derive knowledge of the effect from knowledge of the cause, or vice versa, but that does not show us the cause as source of the effect. Causation, then, is not to be identified with necessitation. [8]
Causal claims thus assert something other than the claim that the effect would not have occurred if the cause had not occurred; rather, they say something about how the effect was brought about by the cause. It is her claim that the philosophical tradition, by not attending to this, misses out on something fundamental to causality.
But doesn't this also accurately characterize a kind of question that epidemiology is normally not in a position to answer, and that it is also not part of its typical purview? The epidemiologist might find himself at home in the philosophical tradition that Anscombe is concerned to put to one side, or that part of it which looks to strict or statistical laws when attempting to articulate the essence of causal relations, and choose to dismiss her attempt to refocus philosophical awareness of causation as irrelevant. But to the extent that he finds Anscombe's argument intriguing, even if not compelling, to that same extent he is faced with reasons to refrain from couching his findings in causal language. This is of course not to say that there couldn't be causal statements about the relationship between, say, smoking and cancer, presumably uncovered by scientists in the areas of physiology and medicine, but they would presumably venture to articulate precisely how the effect is brought about by the cause, in given cases.
Anscombe remarks that our knowledge of causality is acquired through the learning of diverse causal concepts associated with actions and events. If talk about causes shows the possession of a concept "cause," this is a sophisticated addition to the language of someone mastering other causal concepts:
... the word 'cause' can be added to language in which are already represented many causal concepts. A small selection: scrape, push, wet, carry, eat, burn, knock over, keep off, squash, make (e.g. noises, paper boats), hurt. But if we care to imagine languages in which no causal concepts are represented, then no description of the use of a word in such languages will be able to represent it as meaning cause. [8] (p. 137)
In an epidemiological vein, when we say for lung cancer that smoking increases risk by ten times compared to those who do not smoke, the causal language is "increases risk by ...." What is the purpose then of adding "and this is likely to be a causal relationship"? Some may point out that that one statement can stand in for the other, and we would agree (under very specific circumstances), but this begs the question as to why epidemiologists seem to need to privilege one type of causal utterance over another, or to redundantly use language that explicitly uses the word "cause." Further, there is actually an asymmetry here between types of causal utterances. Claiming that something "increases risk" is, for the most part, less ambiguous than saying something causes another. Thus, in practice, different kinds of causal statements are not necessarily substitutable.
The knowledge of having hands
Readers with a surplus of philosophical patience might perhaps join us in also considering K's response against the background of G.E. Moore's famous attempt to argue that he had at least some bits of knowledge that were certain, and Wittgenstein's comments. Moore argued for the claim "I know I have two hands!" by first holding up one, while remarking, "Here is one hand," then the other [9,10]. This approach to knowledge, while intending to appear naïve, is actually under-girded by a sophisticated epistemological superstructure [11]. The thought was, roughly, that knowledge claims might be supported and skepticism about empirical knowledge refuted by providing examples of bits of such knowledge unquestionably available to a subject's mind in its perceptual encounter with his environment.
Wittgenstein was not convinced. His argument, much watered-down, went something like the following: let's say someone is playing a piano sonata, it is clearly a piece that is demanding two hands, but while he is playing he suddenly yells to his audience "I know I have two hands!" This phrase, apropos of nothing, is essentially meaningless; the knowledge of the two hands is, as it were, implicit in the playing of the sonata, the utterance of his claim to "know" this is idle. Indeed, if someone came up to you and said, completely out of the air, "I know I have two hands," there would be no context and no real information conveyed. He claimed in On Certainty [12] that Moore's attempt to justify his knowledge claim was misguided, and is interested in reinterpreting what is going on in such an example. He argues that although Moore's knowledge claims are indeed of a kind that it does not make sense to doubt, this is not because they are supported by irrefutable evidence:
The propositions, however, which Moore retails as examples of such known truths are indeed interesting. Not because anyone knows their truth, or believes he knows them, but because they all have a similar role in the system of empirical judgments. [12] (remark 137)
What is this "role"? Wittgenstein characterizes these claims metaphorically as, for example, what one can "discover ... like the axis around which a body rotates. ... [where] the movement around [the axis] determines its immobility" (remark 152), as the "rock bottom of one's convictions, ...one might almost say that these foundation walls are carried by the whole house" (remark 248), or "... the questions that we raise and our doubts depend on the fact that some propositions are exempt from doubt, are as it were like hinges on which those turn" (remark 341). These claims are then a kind of spin-off from the use of language, mistaken for empirical propositions, and treated by Moore as exempt from doubt. But,
I should like to say: Moore does not know what he asserts he knows, but it stands fast for him, as also for me; regarding it as absolutely solid is part of our method of doubt and enquiry. [12] (remark 151)
Wittgenstein was making the argument that knowledge is really not explicable unless tied to a process of doing things in the world. We know the pianist has two hands by virtue of his playing the sonata; the epidemiologist's research on exposures and effects, and his findings, make it impossible not to think of smoking as a cause of cancer, although the claim transpires rather than follows from the material.
The redundancy of emphatic causation claims
Let's look at an example of the effect of introducing an emphatic causality-claim into an epidemiological context. Consider the differences between saying
(1) Smoking causes lung cancer,
(2) If you smoked 2 packs a day for X amount of years, your chance of getting lung cancer would be 10 times greater than a non-smoker,
and finally,
(3) If you smoked 2 packs a day for X amount of years, your chance of getting lung cancer was 10 times greater than a non-smoker and it's causal! (Perhaps stamping one's feet for emphasis.)
Statement (1) needs the information in statement (2) to be useful from an epidemiological or public health standpoint. Statement (2) describes (of course in a somewhat sketchy way) an increased risk (in itself a causal statement) associated with exposure. (Information about attributable risk could also be included.) Does any other information need to be conveyed, beyond such description, such as in statement (3)? What is the nature of this last statement? Is the addition of "and it's causal" to statement (3) based on the content of statement (2) in a manner that makes (3) into an articulation of something that necessarily follows from (2), although it is not articulated there? Or does "and it's causal" convey some additional information that is useful/necessary, and if so, on which grounds?
Although epidemiologists may think that being able to say very specifically that smoking causes lung cancer is an important part of the research process, this kind of claim would not be underwritten by research findings. One might speculate that being able to make an explicitly causal claim is a desideratum for the professional culture, a desire or inculcated need to be able to make use of explicit causal language when stating conclusions or findings. After all, the use of causal language for purposes of summarizing or concluding may allow others to quickly ascertain whether this research is worth paying attention to or not, depending on whether causal claims are being made. That such claims might also be policy-driven, rather than demanded by the research effort itself, need not be germane to such a cultural trait.
The causal work actually done (i.e., the useful scientific information) is rather embedded in the longer detailed description or story, and this does not necessarily have to include any explicit language involving claims about causation. What we can say, with absolute certainty (taking into account different variables), is that a specific association was found. We cannot, with similar certainty, say that a causal relationship was found, nor do we need to do so. The former claim is accurate to the extent that the research methods were good and repeatable, the latter, explicitly causal claim, may not even be capable of being assessed.
More than semantics
This may seem, as mentioned before, to be just a semantic quibble, but it is not just that. Whether – as Hume argued – causation can neither be proven nor fully experientially justified, or the findings provide inadequate clues as to how the assumed effect was brought about by the assumed cause, we are in effect left with provisional approaches toward treating such relationships between exposures and outcomes [4] (pp. 4–10). To be clear, we are not arguing that people do not think in causal ways, but that when we try as epidemiologists very explicitly to say "X causes Y," we put ourselves in a position similar to that of G.E. Moore saying "I know I have two hands." The knowledge is in the doing; the causal information is in the explicit explanation of how smoking is related to lung cancer.
If we allow the extra language of causation, we need to ask how such an utterance is related to the research at hand. Is there something about research per se that demands causal language, along the lines of "X causes Y," be used when describing the results of the research? Or perhaps there is something that is demanded by the need to do the research in the first place, for example determining what caused the disease outbreak at the picnic. Or perhaps there even is something about the culture of epidemiological research (or all scientific research for that matter) that necessitates being able to say X causes Y.
It may appear that a researcher who has a great deal of expertise in an area of research, such as lung cancer and smoking, should be able to say simply and plainly that "smoking causes lung cancer," based on specific research findings. The problem is that the reasons for this "should be able to say" will not be directly supported by the research itself; it is not what makes this choice of words compelling. There is no more direct justification for this than for needing to say "I know I have two hands." The person uttering this may also think that they "should be able to say" that this is the case and provide a series of reasons, such as professional expertise, the demands of the piano culture, his comedy routine, etc.
The researcher's need to say "X causes Y" reveals something about his state of mind and his beliefs about what the research shows, but this information is not pertinent to the presentation of the research itself. Thus, we would argue that there are only two states the researcher is left with. If a knowledge or causal statement is uttered without context, whether the playing of the piano, or the setting out of a research process, such a claim is unsupportable. Within the particular context of epidemiological research, the claim is redundant and misleading.
The reader, at this point, may think we are being too harsh on the researcher who has many years of research experience and understands the literature, important alternative hypotheses, etc. All this may be true and the researcher may be contributing important information. The problem comes when the researcher is appealing to only the research itself to support the claim of causation, and the research itself is mute about causation per se, although eloquent, hopefully, about how, let's say, smoking increases the risk for lung cancer under specific conditions, controlling for confounders and avoiding biases. There is nothing separate from the results of the research that announces itself as causal. The researcher calling the relationship between smoking and lung cancer a "causal connection" will not be able to point to any element or grouping of elements in the research that unambiguously shows a causal relationship. The researcher can, and even hopefully will, use tools such as analyses of necessary and sufficient conditions, or counterfactual formulations in his research. Nevertheless, even competent handling of tools associated with the search for causal relations does not magically bestow a right to base causal claims on the findings.
Causal anxiety in the 2004 Surgeon General's report
In the new Surgeon General's report [2], the problem is most easily observed in the scale developed to gauge strength of language used in making causal statements about research. The four-level hierarchy for classifying the strength of causal inferences based on available evidence is as follows (page 18):
A. Evidence is sufficient to infer a causal relationship.
B. Evidence is suggestive but not sufficient to infer a causal relationship.
C. Evidence is inadequate to infer the presence or absence of a causal relationship (which encompasses evidence that is sparse, of poor quality, or conflicting).
D. Evidence is suggestive of no causal relationship.
Different causal methods may be used to choose a particular category among A-D, but no operational criteria for choosing among them is in fact being proposed; the authors instead appeal to a shared notion of what is appropriate in the field. Such an approach is of course fine as long as it is seen for what it is, and is not.
Further, the authors explicitly state that counterfactual claims provide the preferred basis for causal claims: "In this report, the definition of cause is based on the notions of a 'counterfactual' state." [2] (p. 19) Although counterfactuals do provide a powerful approach toward understanding how exposures and outcomes may be related, the specific claim that somehow this particular manual for the proper usage of causal terms might serve as the final arbiter of causal claims is troubling for a number of reasons.
First, there are many causal tools that are important in helping to determine relationships between exposures and outcomes, but what would justify the promotion of this particular approach compared to others? This is a highly provisional claim that fails to account for the fact that counterfactual approaches, like all causal approaches, cannot provide a generally applicable definition of cause (per se) [13]. There are always exceptions and ambiguities that finally point to the undeniable conclusion that methods to determine causal relations are not synonymous with observing objective cause.
Secondly, why does a causal claim need to be made explicit? The authors of the report are quite clear on this when they say, "Without the mantle of 'causal,' the identification of a 'risk factor' does not necessarily carry with it the certainty of disease prevention or delayed onset following exposure reduction or removal." They go on to say, somewhat confusedly, that "the characteristics of evidence that merit calling an association causal involve extra-statistical judgments. Because the claim is so central to disease prevention ...." (p. 19). There is a hiatus here between the information in the research findings and the need to come to causal conclusions for the purposes of using these research findings for public health intervention. Although we agree that intervention is an essential part of epidemiological and public health research, it is not clear that being able to say that X causes Y makes any sense, in this regard, except in a very highly contextualized/technical setting. The authors use an ends-justifying-the-means-argument that assumes that explicitly being able to say X causes Y is a necessary element of public health research.
Telling a good story
Our point is that although it is important to be able to use epidemiological research to predict and intervene at the public health level, to tell the best story possible about the research findings at hand, one doesn't have to say that X causes Y to achieve such an outcome. In fact, one cannot definitively claim such a relationship. We think the approach of the Surgeon General's report is commendable in detailing how one can obtain useful information from epidemiological data. Indeed, showing that smoking, controlling for a host of possible confounders in a cohort setting, increases risk for lung cancer is an adequate causal statement. There is nothing speculative in such a claim; we may accurately describe results in terms of estimates of effects, measures of statistical variance and control of confounders, hopefully replicable, in the best tradition of scientific research. All of this is non-controversial with regard to the practice of epidemiology. There is little room for ambiguity, although one may interpret data in many different ways. But neither of this requires, nor does it support the shift to causal language on the part of the authors; the conclusions and decisions that depend on beliefs about causation can be left to the readers.
When considering how we need to think causally in a public health setting, the salient points involve the usefulness of the information for prediction and possibly intervention. Thus, the usefulness and value of the long, patient description ultimately derives from how well people are convinced that this information provides a basis for some kind of intervention or prediction. Although much has been written about causation, we may, as Sosa and Tooley [12] and Cartwright [1] argue, never be able to have anything but a very specific "singularist" sense of causation; i.e., a sense of causal thinking that is not capable of being generalized with rules or methods, but is insurmountably contextual [8]. Far from being an obstacle, such an approach allows for a great deal more clarity regarding the interpretation of epidemiological research.
As mentioned above, this approach appears to be in line with the work of Parascandola and Weed [5] when they point out that probabilistic models of causation are essentially more flexible than deterministic approaches. Whether discussing determinism in causation or more humble, but no less important, issues about causal tools, there is no need to worry about generalizing the discussion. At best, these tools may act as guides that may make specific research more useful for the purposes of intervention or prediction, without providing access to posited objective causal relationships. That certain contexts, such as legal definitions of what constitutes cause, may force a specific notion of cause to come into play; e.g., as demanded by a rule making body like the office of the Surgeon General, or a Judge, provides no added significance to saying that X causes Y. Clearly, certain contexts may demand a very specific use of causal language. Such technical usage of "cause" etc., perhaps in a deterministic way, as might be demanded by a legal process, will occur in a specific setting. For example, a question such as "how much of the paralysis was caused by the faulty tires?" may be unambiguously germane for the purposes of adjudicating a tort case in which a specific notion of cause is introduced and accepted by all parties. Perhaps such uses of "cause" etc. need this level of description.
In another related example, one might ask, what of the situation when undertaking a marginal structural model (MSM) analysis in which the research differentiates between casual effects and mere effects? Is this not a justified use of cause? The answer is a qualified yes, because such a use is highly defined and limited in its meaning. For the sake of MSM analysis, a causal effect is differentiated from a non-causal effect as a function of how well relevant (a judgment call as to what is relevant) confounders and indirect effects are included in the model. The more complete the more "causal" argument is in regard to alternative hypotheses, the better – i.e., more causal – the model. Crude effects, on the other hand, are those that have included minimal, if any, control of relevant confounding and inclusion of indirect effects. There is no hard-and-fast test of when a mere effect becomes a causal effect. This assessment is up to those doing the research and those who assess it. Thus, what a causal effect seems to actually stand for is a more rigorous analysis. This rigorous analysis will hopefully yield more useful information than a less rigorous analysis in regard to intervention and/or prediction. The work here is not in the naming of something as causal, but in the actual rigor of the analysis. The causal language is thus a shortcut that denotes such rigor. Any foundational causal claims are, in fact, the result of circular reasoning. The main point here is that highly contextualized technical/statistical uses of causal language are not the same as making general causal claims about, for example, smoking causing lung cancer. We are always forced back into asking "under what conditions?"
Conclusion
We have argued that saying smoking causes lung cancer is either an empty or a redundant statement from a scientific perspective; implicitly or explicitly it belongs in the realm of health policy. Epidemiologists need to be constantly aware of the limits of causal language and also of the demands of making explicit causation claims.
When attention is not sufficiently paid to properly contextualizing causal claims and loosely using causal language, there are potentially real world consequences. For example, cigarette company lawyers were often heard to say that the case has not been definitively made that smoking caused lung cancer. They said this knowing full well that in the real world, there is nothing that can be definitively claimed. Nevertheless, this should not, in any way, be an obstacle for epidemiologists in the role, for example, of expert witnesses, who put forward the strongest possible account of a given research program, such as one that links smoking to lung cancer. Indeed, the best we can hope for here is to make the most compelling case, the most persuasive account, and hope that it will be more, rather than less, convincing. This is not a nihilistic throwing of the baby out with the bathwater. Not being able to say something is definitively causal does not mean that extremely useful information is not available; it is simply not available in the way that is traditionally demanded by this specific research community.
And here we must emphasize that there really is something different about implicit and explicit causal arguments. We can easily defend the claim that a ten-fold risk was found for two pack-a-day smokers compared to those who did not smoke. We simply cite the methods and research findings. We cannot defend the additional explicit claim that this is a causal relationship in the same manner. In fact, trying to justify such a claim results in circular explanations, as seen in the 2004 Surgeon General's report.
Unfortunately, even under the best conditions, we have no control over what rational or irrational processes a person will employ to assess usefulness or causation. For the cigarette executives, at one point in time, almost nothing could have been thought convincing. We could, however, imagine that as the power of the research findings mounted, there would be a decrease in the number of executives actually smoking. This, for us, is a promising kind of convincing causal argument, one that is based in actual changes made in the world. Thus, one could describe a narrative where risk is found to increase 10 times for smokers versus non-smokers and 20 years later tobacco executives were smoking in far reduced numbers. Obviously, similar changes did occur in the general population after the first Surgeon General's report appeared in 1964. There is, however, no final arbiter in this regard. Thus, we cannot create a fail-safe scale, or "causal" regime that will, simply by reaching a certain threshold, result in an uncontroversial notion of "causal" or causation, per se.
Efforts to establish standards for making causal claims, as in the new Surgeon General's report, should be encouraged as long as the focus is on developing a more coherent and shared sense of what makes specific research efforts, such as examining lung cancer and smoking, more useful for public health and medical purposes. Explicit causal language, if used in a very technical, agreed-upon sense, such as in the MSM modeling example above, could be similarly useful. But this use of technical causal language, a good use, in our estimation, must be recognized as simply a shorthand for better versus worse analyses, as judged by the author, and not a metaphysical statement about causation per se (which is beyond what we can learn from epidemiological findings). Given the difficulties described above, the establishment of an unambiguous meaningful and general notion of causal claims, besides being, for all practical purposes, unavailable, is unnecessary for the real world task of prediction and intervention at the public health level. Thus, claims such as "X is likely to be a causal factor for Y" should only be made if sufficient context and definition is provided, and omitted otherwise. For epidemiology, in particular, and science generally, the devil is in the details.
Acknowledgements
We would like to thank the editor-in-chief Carl V. Phillips and the anonymous reviewers for their comments and suggestions. They have made this a much stronger effort than it would have been otherwise. We would also like to thank Kathryn C. Dowling for her exceptional editorial assistance.
==== Refs
Cartwright N Teichman R An empiricist defence of singular causes Logic, Cause and Action: Essays in Honour of Elizabeth Anscombe, Royal Institute of Philosophy Supplement #46 2000 Cambridge , Cambridge University Press 47 58
US Department of Health and Human Services. Office of the Surgeon General The Health Consequences of Smoking: A Report of the Surgeon General 2004 Washington, D.C. , U.S. Department of Health, Education and Welfare, Public Health Service, CDC
U.S. Department of Health Education and Welfare Smoking and Health. Report of the Advisory Committee to the Surgeon General of the Public Health Service 1964 Washington D.C. , U.S. Department of Health, Education and Welfare, Public Health Service, CDC
Wilson F Hume's Defense of Causal Inference 1997 Toronto , University of Toronto Press 439
Parascandola M Weed DL Causation in epidemiology Journal of Epidemiology and Community Health 2001 55 905 912 11707485 10.1136/jech.55.12.905
Renton A Epidemiology and causation: a realist view Journal of Epidemiology and Community Health 1994 48 79 85 8138775
Fine A The Shaky Game: Einstein Realism and the Quantum Theory 1996 2nd Chicago , The University of Chicago Press 224
Anscombe GEM Causality and determination The Collected Philosophical Papers of GEM Anscombe, Volume Two, Metaphysics and the Philosophy of Mind 1981 Minneapolis , University of Minnesota Press 133 147. Reprinted in Causation. Edited by Sosa E, Tooley M. New York: Oxford University Press; 1993.
Moore GE Proof of an external world Proceedings of the British Academy 1939 25 273 300
Stroll A Moore and Wittgenstein on Certainty 1994 New York , Oxford University Press 206
Wittgenstein L Anscombe GEM, von Wright GH On Certainty 1969 New York , Harper & Row Publishers, Inc.
Sosa E Tooley M Sosa E, Tooley M Introduction Causation 1993 New York , Oxford University Press 1 35
|
16053522
|
PMC1198241
|
CC BY
|
2021-01-04 16:36:38
|
no
|
Epidemiol Perspect Innov. 2005 Jul 29; 2:8
|
utf-8
|
Epidemiol Perspect Innov
| 2,005 |
10.1186/1742-5573-2-8
|
oa_comm
|
==== Front
Front ZoolFrontiers in Zoology1742-9994BioMed Central London 1742-9994-2-111598753110.1186/1742-9994-2-11ResearchShape based assignment tests suggest transgressive phenotypes in natural sculpin hybrids (Teleostei, Scorpaeniformes, Cottidae) Nolte Arne W [email protected] H David [email protected] Institute for Genetics, Evolutionary Genetics, Weyertal 121, 50931 Cologne, Germany2 Dept. of Physics, Canisius College, 2001 Main St., Buffalo, NY 14208, USA2005 29 6 2005 2 11 11 20 4 2005 29 6 2005 Copyright © 2005 Nolte and Sheets; licensee BioMed Central Ltd.2005Nolte and Sheets; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Hybridization receives attention because of the potential role that it may play in generating evolutionary novelty. An explanation for the emergence of novel phenotypes is given by transgressive segregation, which, if frequent, would imply an important evolutionary role for hybridization. This process is still rarely studied in natural populations as samples of recent hybrids and their parental populations are needed. Further, the detection of transgressive segregation requires phenotypes that can be easily quantified and analysed. We analyse variability in body shape of divergent populations of European sculpins (Cottus gobio complex) as well as natural hybrids among them.
Results
A distance-based method is developed to assign unknown specimens to known groups based on morphometric data. Apparently, body shape represents a highly informative set of characters that parallels the discriminatory power of microsatellite markers in our study system. Populations of sculpins are distinct and "unknown" specimens can be correctly assigned to their source population based on body shape. Recent hybrids are intermediate along the axes separating their parental groups but display additional differentiation that is unique and coupled with the hybrid genetic background.
Conclusion
There is a specific hybrid shape component in natural sculpin hybrids that can be best explained by transgressive segregation. This inference of how hybrids differ from their ancestors provides basic information for future evolutionary studies. Furthermore, our approach may serve to assign candidate specimens to their source populations based on morphometric data and help in the interpretation of population differentiation.
==== Body
Background
Although hybridization has long been considered important in the diversification of plants zoologists often considered it detrimental and thus unimportant [1]. The debate of the relative importance of hybridization has received recent attention because the advance of molecular techniques has resulted in a surge of data suggesting that hybridization is taking place rather frequently in the animal kingdom as well. This in turn has revived questions surrounding the potential role that hybridization may play in the penetration of evolutionary novelty in animals [2,3]. A simple explanation for novel phenotypes of hybrids is available through the process of transgressive segregation. Briefly, transgressive segregation is a phenomenon specific to segregating hybrid generations and refers to individuals that exceed parental phenotypic values in any direction. This could be caused by heterosis, which is most pronounced in first generation hybrids, or alternatively by the complementary action of parental alleles dispersed among divergent parental lineages. If this is frequent, then an important evolutionary role for hybridization is more easily explained [4].
In fact, there is abundant evidence that transgressive segregation is common in both plants and animals and that the genetic architecture for it is rather commonplace than exceptional [5]. Given these findings, it is astonishing, that relatively few studies have evaluated transgressive segregation in natural systems [4]. On the one hand this a results from the paucity of study systems where sufficiently large samples are readily available and from the simple fact that quantitative genetics experiments are usually conducted in controlled environments in order to separate environmental from genetic effects. If one searches for transgressive segregation one would ideally study traits, which are determined by several genes and that display a hidden divergence of the underlying genetic network [5]. Finally, one has to study direct hybrids and not lineages of hybrid origin because otherwise secondary evolutionary processes will have reshaped any hybrid lineage and secondarily modified characters cannot be easily distinguished from transgressive traits. Despite these difficulties many evolutionary studies will ultimately have to incorporate natural populations in real ecosystems if the effects and outcome of hybridization are to be analysed.
As an example, hybrid zones among divergent lineages are viewed as natural laboratories and offer interesting study systems [6]. In these, the fitness of hybrids is a key component to understand the dynamics of the hybrid zone as a whole [7]. Transgressive segregation may affect hybrid fitness as it is a mechanism that would make hybrids different and thus produces the raw material upon which selection can act.
We have recently identified hybrid zones of European sculpins belonging to the Cottus gobio complex (Scorpaeniformes, Cottidae) that fulfill the above requirements. Sculpins are small, benthic freshwater fishes that occur in streams throughout Europe, with closely related species distributed throughout the northern hemisphere. Previous studies have revealed a high cryptic diversity of this group across the entire distribution range [8]. Our focus area is the River Rhine System, where divergent lineages of sculpins are known to occur in parapatry and have come into secondary contact [8,9]. Small tributaries to the Lower Rhine drainage are inhabited by isolated populations of 'stream' sculpins, a lineage endemic to the River Rhine [8,10]. These stream populations correspond to Cottus rhenanus [11]. Intriguingly, a new 'invasive' lineage, has recently appeared within the main channels of larger rivers that where previously free of Cottus [10]. The invasive sculpins represent a different species, Cottus perifretum [11] that differs from sculpins in streams of the Rhine area (C. rhenanus) in body shape and in that its lateral body is largely covered by modified scales vs. an almost complete absence of such modified scales [10]. Invasive sculpins come into secondary contact with populations of stream sculpins where small tributaries disembogue into the main channel of larger rivers. In these areas individuals belonging to both parental populations as well as hybrids among them occur syntopically. With respect to transgressive segregation the above prerequisites are fulfilled. First, sufficiently divergent lineages come into contact and produce recent hybrids. Secondly, these hybrids can be readily identified using genetic data. Finally, variation in body shape provides a well-suited character complex since sophisticated methods are available to study shape [12]. Furthermore, previous studies on body shape show that this character complex is usually determined by multiple genes [13-15]. Below we combine genetic and phenotypic approaches to study body shape in sculpin hybrid zones and present data suggesting that transgressive body shape phenotypes occur in natural sculpin hybrids.
In order to study variation in shape, parental groups and their hybrids were classified to establish how their phenotypes and genotypes were related. Since shape was of key interest, we relied on model based population genetic approaches [16] to independently cluster and assign specimen to their populations of origin or to determine their hybrid status. Such model based clustering is not possible in the analysis of shape because a powerful "theory of population shape" comparable to population genetic theory is lacking that would allow to independently infer population affinity.
However, simple assignment methods can contribute much in the sense of the first assignment approaches in genetics that were employed for much more basic questions [17] namely the problem that distances alone are biologically and conceptually hard to interpret. As an alternative to an abstract distance one may ask whether a given character is sufficiently informative to be diagnostic at the individual, population or at higher levels to help in assessing the significance of results. This general problem also applies to quantitative morphometric studies, especially when multivariate analyses are used. We have developed a distance based assignment approach with statistical tests that parallel population genetic approaches [18]. The method is not intended to give a measure of the absolute distance among groups but may help to interpret the differences among groups. One purpose of this paper is to introduce shape based assignment as a multivariate measure of distinctness and to employ this approach to study the relationships of genotypes and phenotypes at natural hybrid zones.
Results
Morphometric differentiation
One population of invasive sculpins (C. perifretum) and the two populations of stream sculpins (C. rhenanus) each confined to a separate stream were sampled and independently confirmed with genotypic data (see methods; see Additional file: 1). These served as basic groups for the following analyses. All of them form distinct clusters in a canonical variates analysis (CVA) along the first two axes, which display the greatest separation of the groups relative to within group variance (Figure 1). Both populations of stream sculpins separate from the invasive sculpins along the first CV axis (Lambda = 0.0678 chisq. = 777.6114 df = 72 p < 0.001). The stream Naaf and the stream Broel populations are further separated along the second CV axis (Lambda = 0.2792 chisq. = 368.6997 df = 46 p < 0.001).
Table 1 Assignment success under alternative CVA models
Assigned Group Broel Naaf Invasive BI Hybrids CVA model
Broel 90.6 1.3 2.5 25.8 Based on parental populations
Naaf 4.3 93.4 0.0 11.3
Invasive 0.9 0.0 92.5 25.8
n.s. 4.3 5.3 5.0 37.1
Broel 87.2 5.3 0.0 1.6 Including parental groups and hybrids
Naaf 2.6 89.5 0.0 1.6
Invasive 0.9 0.0 85.4 6.5
BI hybrid 6.8 1.3 12.2 83.9
n.s. 2.6 3.9 0.0 6.5
Assignment of sculpins to their population of origin based on body shape. A model based only on the differentiation of parental populations is very effective in identifying pure sculpins but assigns the majority of hybrids to pure populations as false positives. A more complex model that includes the shape components specific to hybrids correctly identifies the majority of all hybrids. The overall success of parental group assignment is decreased when hybrids are taken into account as they overlap with parental phenotypic values.
Figure 1 Differentiation of ancestral populations and hybrid intermediacy. Invasive sculpins separate from all stream sculpins along the first CV axis. Sculpin populations from Stream Broel and Stream Naaf separate along the second CV axis. BI hybrids form an intermediate group between their parental populations. Distance based assignment based on these two axes correctly identifies pure candidates while a majority of BI hybrids are wrongly assigned to one of the parental groups with which they overlap.
A fourth group comprised recent hybrids among invasive and stream Broel sculpins that were sampled from natural hybrid zones and identified based on genetic data (BI hybrids). When these are introduced into the CVA as 'unknowns', where the CVA model does not consider them as a separate group but determines their scores along the CV axes separating the parental groups, BI hybrids overlap with their ancestral populations and take somewhat intermediate positions along the first CV axis (Figure 1). If in contrast BI hybrids are used as a predefined group in the CVA, hybrids are further characterized by a third CV axis (Lambda = 0.7369 chisq. = 88.2487 df = 22 p < 0.001). They separate partially from invasive sculpins and stream sculpins along the third CV axis (Figure 2) and take, on average, more extreme phenotypic values than both parental populations along this axis.
Figure 2 Extreme phenotypic values indicate a hybrid shape component. BI Hybrids are, on average, not intermediate along the third CV axis and may occupy extreme values relative to their parental populations. The parental populations as well as stream Naaf sculpins display little differentiation along the third CV axis. An inclusion of this hybrid specific shape component in distance based assignment increases the power to correctly identify hybrids more than two fold.
The differentiation in shape as captured by CV axes can be visualized as displacement vector for each landmark on a deformation grid relative to a reference (Figure 3). Invasive sculpins differ from both populations of stream sculpins in that they have a larger head and anterior trunk as well as a shorter tail (Figure 3; first CV axis). The two populations of stream sculpins differ most in their head length and the positions of their anal and dorsal fin landmarks (Figure 3; second CV axis). While the deformation implied by the first two axes can be expressed in terms of inflation or compression of body parts, the hybrid specific shape change appears to be less balanced although this is hard to objectify (Figure 3; third CV axis).
Figure 3 Landmark configuration and displacement vectors that distinguish groups of sculpins. Fourteen Landmarks were chosen to analyse variability in sculpin body shape (top). CVA was used to identify axes along which different groups can be discriminated based on the relative position of landmarks to a reference. The shape change captured by these axes can be visualized as relative displacement vectors for each landmark on a deformation grid. CV axis 1 separates invasive sculpins from all stream sculpins and axis 2 further separates two populations of stream sculpins. CV axis 3 captures the shape component that is unique to recent hybrids. While the deformation along CV axes 1 and 2 can be expressed in terms of inflation or compression of body parts, the hybrid specific shape change appears to be less balanced.
In order to evaluate whether the observed differentiation was biased due to imbalanced sampling CVA scores were regressed on centroid size and sex for all specimens. A linear regression reveals for all axes that correlations coefficients were low at most (CV axis1 vs. size r2 = 0.03; CV axis1 vs. sex r2< 0.01; CV axis2 vs. size r2<0.01; CV axis2 vs. sex r2<0.01; CV axis3 vs. size r2 = 0.11; CV axis3 vs. sex r2 = 0.019). Therefore, neither size nor gender can explain the considerable amount of variance of the CV axes that distinguish the groups.
Assignment and cross validation
To evaluate the utility of the derived axes to discriminate among groups and to determine a given specimens group affinity, distance based assignment tests were performed. In a first approach the CVA model was based on the differentiation as observed among the pure populations but hybrids were not included as a known group. The clear differentiation of pure populations facilitates that single "unknown" specimens removed from the complete dataset can be correctly assigned to their population of origin with high success (Table 1), and the resubstitution rate of correct assignment (the assignment of the known specimens) is high also, although this resubstitution rate is known to be biased upward. Approx. 92 % of pure sculpins were assigned correctly. This number is slightly lower than the expected 95% due to false positive assignments because of a slight overlap of parental phenotypic values. The number of outliers corresponds well to the amount expected from the significance criterion. In this approach hybrids were used as "unknowns" and could only be identified as outliers relative to the pure sculpins (non significant assignment test). Only 37.1% of the BI hybrids were correctly classified while the majority was misassigned to one of the pure populations.
In an alternative approach assignment was based on a CVA model that includes the differentiation among pure populations but also takes into account the shape component specific to hybrids as captured by the third CV axis (Figures 2 &3). The assignment success of pure populations was decreased to 85.4–89.5% because of the partial overlap with the group of hybrids. In sharp contrast to the above, 83.9% of the BI hybrids were now classified correctly with only relatively few false positive assignments to the parental groups (Table 1).
A jackknife test of assignment was performed for both assignment approaches to evaluate the robustness of the CV axes and assignment model (Table 2). The cross validation procedure revealed a very consistent signal, inherent to even small partitions of the whole dataset. Roughly half of the specimens can be removed from the data without much loss of information for the CV axes. Even when 80% of the whole dataset are left out in the CVA procedure (see methods) the general outcome remains unchanged although the number of correct assignments decreases. As evident from the individual assignment tests (Table 1) the overall assignment success is lower when the more complex model including hybrids is employed.
Table 2 Jackknife estimates of assignment performance.
% left out in Jack – knife (500 replicates) 1 10 50 80 CVA model
% correct 93.6 92,7 90.3 72.1 Based on parental groups
% correct ns. 0.5 0.5 1.3 10.5
% false 5.9 6.8 8.3 14.8
% false ns. 0.0 0.0 0.1 2.6
% correct 84.7 84.3 80.4 65.3 Including parental groups and hybrids
% correct ns. 0.1 0.0 0.4 4.2
% false 15.3 15.6 19.2 28.9
% false ns. 0.0 0.0 0.1 1.6
Jackknife test of assignment. Percent correct and false assignments when fractions of 1% to 80% of the specimens are left out in the CVA procedure and then assigned to groups in the remaining dataset. Large fractions of the data can be removed without loss of the discriminatory power of CVA axes.
Discussion
Transgressive phenotypes in natural hybrids
We were able to use a microsatellite dataset with surpassing information content to classify sculpins into distinct lineages (see methods for details). With the genotypic data it is possible to unambiguously identify invasive sculpins (Cottus perifretum), which are genetically distinct from populations of stream sculpins (Cottus rhenanus). The latter are further represented here by two separate populations from the streams Naaf and Broel. In agreement with the genetic data, the CVA based on morphometric data recovers significant differentiation that separates all studied populations of sculpins with a higher amount of variance between species and a lesser amount between two conspecific samples ofstream sculpins. Cross validation confirms that these results are based on a signal inherent to the whole dataset as a removal of a large fraction of the specimens will not notably alter the axes as determined in the CVA (Table 2). This differentiation is sufficient to assign unknowns to either one of the known groups with high confidence. Thus the groups are distinctly different in their multivariate signal even though no single diagnostic morphometric character can be found.
The genotypic data served to identify a set of hybrids between the invasive and stream Broel populations (BI hybrids). In contrast to the ancestral populations, hybrids cannot be distinguished completely from all of the ancestral phenotypes (Figure 1) and are more or less intermediate in the characters that discriminate their parental populations. This is expected for a character like body shape that is most likely determined by multiple genes. Yet, there are properties of the hybrids that could not be attributed to hybrid intermediacy. The group of hybrids displays a unique shape component that distinguishes them from a their parental populations (Figure 2). Altogether it is not a strong effect thus additional evidence to evaluate the biological significance of this result are desirable. To address sampling artifacts, we have tested for possible effects of typical confounding variables in morphometric studies. Regressions show that the amount of total variance of individual CV axis scores that can be explained by size or sex is small. Therefore the influence of allometry or sexual dimorphism is most likely not important for the differentiation we observe. Despite the large overlap of the BI hybrids with the parental populations, the hybrid shape component constitutes a considerable amount of variation, which results in an increased overall assignment success when hybrid shape is considered specifically (Table 1). Moreover, cross validation has shown that all axes are robust to removal of specimens, which suggests that the signal is inherent to a majority of the recent hybrids (Table 2).
Two alternative explanations remain. One assumes an involvement of genetic factors that interact to produce novel phenotypes, in contrast, the second proposes that the genetic background is not important. According to the latter hypothesis extreme hybrid phenotypes should be determined by the environment. Our genetic data demonstrate that hybrids occur syntopically with the parental populations within the hybrid zones (Table 3). This excludes the possibility that hybrids would be exclusively subjected to environmental factors that could induce the observed phenotypes. Phenotypic plasticity cannot be fully excluded in heterogeneous environments but this process alone is not likely to explain our results. After possible confounding variables were found to play a minor role, it seems reasonable to assume the differentiation is real. In contrast to the above explanations, differentiation due to the underlying genetic background is strongly supported. This includes that the specific hybrid shape effect is coupled with the hybrid genetic background.
Table 3 Sampling sites and number of specimens in the morphometric study.
Number Sampling Site N (Genotyped) N (Genotypic Classes)
1 Stream Broel between Broel and Winterscheidt, North Rhine-Westphalia, Germany; 50°47'N 7°20'E 48 48 Stream Broel sculpins
2 Stream Broel south of Broel, North Rhine-Westphalia, Germany; 50°47'N 7°19'E 48 42 Stream Broel sculpins; 1 BI Hybrid
3 Stream Broel at Mueschmuehle, 200 m above outlet into River Sieg, North Rhine-Westphalia, Germany; 50°47'N 7°18'E 130 26 Stream Broel Sculpins; 45 BI Hybrids; 13 Invasive Sculpins
4 River Sieg at Allner, below outlet of Stream Bröl, North Rhine-Westphalia, Germany; 50°46'N 7°18'E 36 2 Stream Broel Sculpins; 16 BI Hybrids; 2 Invasive Sculpin
5 Stream Wahnbach, Outlet into River Sieg at Seligenthal, North Rhine-Westphalia, Germany; 50°47'N 7°16'E 4 1 Invasive sculpin
6 Stream Pleis, outlet into the River Sieg at Niederpleis, North Rhine-Westphalia, Germany; 50°46'N 7°12'E 5 5 Invasive sculpin
7 River Sieg at Muehlenbach, North Rhine-Westphalia, Germany; 50°47'N 7°10'E 35 19 Invasive sculpin
8 Stream Naaf, Outlet into River Agger, North Rhine-Westphalia, Germany; 50°51'N 7°14'E 48 1 Stream Naaf sculpins;
9 Stream Naaf at Kreuznaaf, 200 m above outlet into River Agger, North Rhine-Westphalia, Germany; 50°51'N 7°14'E 48 30 Stream Naaf sculpins
10 Stream Naaf southeast of Hausdorp, North Rhine-Westphalia, Germany; 50°52'N 7°16'E 48 45 Stream Naaf sculpins
Sampling Sites, total number of genotyped specimens and numbers in genotypic classes used for this study. The individual genotypic classes were inferred from microsatellite data and served to group specimens for morphometric analyses. Specimens excluded from the analysis include later generation backcrosses or those for which morphometric data could not be obtained. Note that there are sampling sites at which all genotypic classes occur syntopically.
Although there is considerable overlap of parental groups and BI hybrids along the CV axis that captures the hybrid shape component, hybrids are on average more extreme than both parental populations. Given that genetic data verify a recent hybrid status of the BI hybrids, these results can be best explained by transgressive segregation in shape traits. This is also in agreement with other studies suggesting that transgressive segregation occurs in morphometric traits [4,5]. However, to assess the evolutionary implications of the hybrid phenotypes will require functional studies and measurements of fitness to complement the mere observation of possible transgressive effects.
Information content of shape markers
A drawback as compared to population genetic model – based assignment is that our shape distance based method needs a priori defined groups as input. If such groups can be provided hypotheses regarding their differentiation and distinctness can be tested. For example, an attractive application of genotype based assignment procedures is to detect outliers that belong to source populations that were not sampled [19,16]. Unfortunately this is not straightforward in our implementation of phenotype-based assignment. If a candidate does not belong to one of the expected groups, the exclusion of source populations is not predictable because assignment based on discriminant axes is conditioned on the specific case being studied. We find this for the hybrids among stream Broel and Invasive sculpins if they are used as unknowns and not as a separate group in the CVA. Hybrids take more or less intermediate phenotypic values but largely overlap with the parental groups (Figure 1).
Similar results were already obtained by Strauss [20] in a study of phenotypic variation in hybridising North American sculpins. However, we have a sufficiently large sample of verified hybrids that could be used as an extra group in the CVA. Only this revealed significant differentiation along an extra axis that is specific to the hybrids. The differentiation specific to hybrids adds information to the group assignment. As a result the assignment success of hybrid specimen was raised notably despite the tremendous overlap of the hybrids with both parental populations (Table 1).
The assignment procedure based on morphometric data as implemented here allows to unambiguously assign sculpins to their population of origin. Morphometric differentiation of European sculpins was studied before [21] using a set of landmarks that was largely identical to the ones used here (note that these authors [21] did not study the same evolutionary lineages, see [8,11]). Groups of sculpins as defined by different tributaries to the Rhine were found to differ significantly in shape but formed largely overlapping clusters. Our system differs in that we have not compared assemblages of populations but separate more or less panmictic populations as defined by a currently shared gene pool. These form distinct clusters in the CVA (Figure 1). Such differentiation would have escaped the approach of Riffel and Schreiber [21] as they pooled specimens from different subpopulations for their analysis. This by no means negates their results but demonstrates an even higher information content of shape data, namely the power to discriminate separate populations. Although a comparison among genetic markers and body shape seems arbitrary the resolution as compared population genetic markers goes beyond recognition of ancient lineages as resolved by mitochondrial haplotypes [8] and species [11] but parallels that of microsatellites in that genetically well separated populations are also distinctly differentiated in body shape. Apparently, shape represents a character with a fast evolutionary divergence that occurs and becomes fixed even among closely related populations. Thus, in our example morphometric data resolve to the lowest possible level above the individual.
Methods
Implementation of shape based assignment
Landmark-based geometric morphometric methods were used to capture information about shape, by obtaining the x and y coordinates of homologous landmarks in the configuration shown in Figure 3. Differences among specimen in the sets of coordinates due to scaling, rotation and translation were removed using the typical geometric morphometric approach [22-25,12] of placing the specimens in Partial Procrustes Superimposition [24-26] on the iteratively estimated mean reference form, using the Generalized Procrustes Analysis procedure. This procedure places the shapes of specimens in a linear tangent space to Kendall's shape space [27], allowing the use of linear multivariate statistical methods [23,28,24,12]. After superimposition, the data were converted from Cartesian coordinate form into components along the eigenvectors of the bending energy matrix (Principal Warp axes) of the thin-plate spline model of deformations of the reference [29,22] and along the uniform axes of deformation due to shear and dilation [30]. Use of these linearly transformed variables (referred to as Partial Warp plus Uniform Component scores), produces a convenient set of variables (using a basis set called the Principal Warp axes) for use with standard multivariate statistical methods, since the Partial Warp and Uniform component scores have the same number of variables per specimen as degrees of freedom. No information is lost during this linear transformation of variables.
A canonical variates analysis (CVA) is then used to determine the set of axes which best discrimate among pre-defined groups of specimens, by determining the linear combinations of the original variables which display the greatest variance between groups relative to the variance within the groups [31,12]. Fisher's linear discriminant function was used, which makes no particular assumption about the parametric form of the distribution of the data used, but simply determines the linear combination of the original variables that results in the greatest ratio of the between groups sum of squares to the within groups sum of squares [32]. A simple distance-based approach is then used to determine which group each specimen belongs to, based on the canonical variate scores. The predicted group membership of each specimen based on the CVA scores is determined by assigning each specimen to the group whose mean is closest (measured as the square root of summed squared distances along the CV axes, see [32]) to the specimen. To obtain a measure of the quality of the assignment of each specimen to a group, an assignment test was developed. The CVA axes can always be used to assign any given specimen to some group, since a minimum distance can always be found but a measure of whether the quality of the assignment is similar to that expected for specimens known to be in that particular group is desirable. The assignment test presented here is modeled on the genetic distance-based assignment test [33,18]. The distribution of distances produced by a Monte Carlo simulation (see discussion in [34]) is used to determine if the observed distance of a given specimen is consistent with the null model of random variation around the mean of the group to which the specimen is assigned to. The distance from a specimen to a group mean can then be assigned a p-score which describes how likely it would be for a specimen from the original population to be as far from the mean specimen as the observed specimen is (under the null model used in the Monte-Carlo simulation). If the p score is smaller than 5%, then we can assert that there is a less than 5% chance that random variation could have produced a distance as large as that of the particular specimen from the group mean, and hence that the assignment of that specimen to the group is in doubt.
It should be noted that in a study with many specimens, a number of them will have low p-scores by chance, and so to assess the validity of the assignments of the set as a whole, the researcher should assess the number of specimens expected to have p values less than 5%. It will then be possible to determine if the observed number of low p values exceeds that expected by chance. The model used in the Monte Carlo simulation of the distribution of distances of specimens within a group around the group mean (the average specimen within the group) is based on a normal model of the distribution of the CVA scores of each group about the mean of that group. For a given group, it appears probable that the CVA scores along each CVA axes for the specimens within the group are correlated, thus there exists within each group a covariance structure to the CVA scores of specimens within the group. In carrying out the Monte-Carlo simulation of the distribution of specimens within the group about the mean, it is necessary to preserve this covariance structure in order to produce a valid model of the distribution. An eigenvalue decomposition of the variance-covariance matrix of within group CVA scores is used to find the principal component axes of the within group variance. This yields the same number of variables as the CVA scores, but now with uncorrelated axes (the eigenvectors), each of which has a variance given by the corresponding eigenvalue. The model used for the distribution of the CVA scores of the specimens assumes the group has an independent random normal distribution along each of these eigenvectors (principle component axes), with amplitudes given by the square roots of the eigenvalues (the eigenvalues are the variances of the group along the corresponding eigenvectors), so that the square root of the eigenvalue is the standard deviation of the population along that eigenvector.
An independent, normal distribution with known amplitude (the square root of the eigenvalue) is assumed along each eigenvector. This allows generation of a Monte Carlo population of specimens, assuming the independent normal distribution along these principal component axes. Each simulated specimen is generated using a random number generator to compute locations along the eigenvectors, which are then translated back into CVA axes scores (a simple linear translation of basis vectors). The Monte Carlo generated CVA axis scores will have the same mean and variance-covariance structure as the original population did. Using an independent multinomial distribution model of the CVA axes scores in the Monte-Carlo simulation (by using the random number generater to directly generate the CVA scores) would not preserve any covariation structure in the data. If there is no significant covariation structure to the CVA scores, the use of the PCA axes is not necessary, but will not induce covariations.
The distance from the group mean is then calculated for each simulated specimen. The Monte Carlo distribution of distances of specimens about the group mean under the null model of random variation can then be used as an estimate of the distribution of distances about the group mean in the original data. Based on the estimated distributions of distances about the group means expected for specimens in the group, p-values may be determined for the assignment of specimens, with either initially known or unknown group affinities. Based on an alpha level of p = 0.05, all assignments of specimens to groups can be scored as either statistically significant or not.
As a test of the performance of the assignment test, a cross validation or jackknife procedure [34,35] was implemented. Unlike a standard jackknife where only one specimen at a time is omitted, a "delete-d" jackknife [35] was used in which d specimens at a time were omitted. Under the delete-d jackknife, a variable percentage of individuals from a dataset are left out during the CVA procedure, and then assigned to groups as "unknown" specimens. The specimens treated as unknowns during the jackknife procedure are also assigned an assignment p-value during this procedure. High success rates under the delete-d jackknife resampling indicate that the differentiation among the involved groups is sufficient to be diagnostic. This implies that the discriminant axes capture enough information to assign individuals of the given groups, and form a reasonable estimate of the distribution of distances based on the Monte Carlo procedure. The jackknife procedure also allows estimation of the number of individuals needed to obtain meaningful CVA axes, and distance distribution estimates.
The sculpin data set
Here we employ the methods outlined above to study the differentiation in shape that occurs among divergent populations and their naturally occurring hybrids. Population affinity and hybrid status are independently derived from genetic markers. Note that the specimen are taken from natural populations and occur syntopically in the studied hybrid zones (Table 3). Sculpins were sampled across an area of secondary contact of invasive and stream sculpins (C. perifretum and C. rhenanus) in the Lower Rhine basin, which is situated at the confluence of the Stream Broel with the River Sieg (Table 3). An extra population of stream sculpins was sampled from the stream Naaf (also a tributary to the River Sieg drainage).
All specimens were genotyped for 45 microsatellite loci [36]. The loci were chosen for their particularly high information content for our study system following the approach of Shriver et al. [37] using Whichloci [38]. We have used a preliminary genetic map of Cottus [39], to verify that our set of microsatellite markers does not contain pairs of loci that are tightly physically linked. The genotypic data allow to unambiguously classify individuals to belong to pure populations or to identify them as hybrids with a mixed ancestry using the methods outlined in Falush et al. [16]. The program Structure 2.1 [40] yielded consistent results in independent runs (burnin: 20000; sampling iterations 100000, correlated allele frequency model allowing for an individual alpha and different FST for each subpopulation) according to which the genetic ancestry of individual could be determined (see Additional file: 1, for genotypic data of those specimen included here). The classification based on genotypic data was highly congruent with data from distribution and morphology.
Two independent populations of stream sculpins confined to separate streams (Stream Naaf, Stream Broel) both being devoid of skin prickling were recovered. A third population can be identified, which represents the invasive sculpin. Invasive sculpins generally occur within the main channel of the River Sieg and display pronounced skin prickling [10]. Hybrids among Stream Broel sculpins with the invasive sculpins were only found at the confluence where the Broel merges with the River Sieg (Table 3). A detailed study of these hybrid zones, particularly on the geographic extension, is currently in preparation (Nolte et al. in prep). Of particular relevance in this context is the fact that hybrid sculpins occur syntopically with their parental populations within the hybrid ones (Table 3; Sites 2, 3, 4). For the morphometric analysis we grouped specimens that were found to belong to pure populations from the genotypic data into those corresponding to the invasive sculpins (invasive) and to the two stream sculpin populations (Streams Naaf and Broel). To allow for some uncertainty in the estimates we used those specimens that were found to be at least 97% pure in the structure analysis. These populations are represented here by 117 Stream Broel sculpins, 76 Stream Naaf sculpins and 40 Invasive sculpins (Table 3). In contrast, hybrids represent a somewhat inhomogeneous group consisting of various degrees of ancestry. To restrict this analysis to those specimens that have a pronounced hybrid genotype and to exclude later generation backcrosses that might have been subject to repeated rounds of natural selection (as this could blur possible transgressive effects) we decided to exclude hybrids with less than 25% ancestry in one of the ancestral populations. Based on genotypic data, we were able to identify 62 BI hybrids (mixed Stream Broel/Invasive ancestries, less than 75% pure ancestry).
Images of specimens were taken with a digital camera fixed on a stage so that the midsagittal body plane was as much as possible perpendicular to the image plane. Fourteen morphological landmarks were used to capture the shape of each individual (suppl. Table 2) using TPSdig [41]. The positions of the tip of the nasal (#1), nares (#2), interorbital pores (#3), dorsal fin I origin (#4), dorsal fin II origin (#5), dorsal fin II end (#6), upper caudal fin origin (#7), lower caudal fin origin (#8), anal fin end (#9), anal fin origin (#10), ventral fin origin (#11), upper origin of the gill opening (#12), opercular spine (#13) and posterior end of the maxilla (#14) were used as landmarks (Figure 3). The morphometric analysis was conducted using the IMP package according to the methods outlined above [42]. Shape based assignment tests were conducted with CVAgen6N (part of IMP). In order to estimate possible confounding effects of allometric growth and sexual dimorphism these variables were determined individually. A scale bar was photographed besides each specimen as a size reference and sex was determined for individuals larger than 45 mm by examination of the genital papilla (see Additional file: 2).
Authors' contributions
AN identified the sculpin hybrid zones and carried out the molecular genetic studies, morphometric analyses and drafted the manuscript. HDS participated in the morphometric analysis, developed the assignment procedure based on morphometric data and helped to draft the manuscript. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
Inferred group affinity and individual genotypic data. Genotypes of all specimens for 45 microsatellite loci (0 = missing data, alleles numbered according to size, but not necessarily repeat size) with group affinity and sampling site as of Table 3.
Click here for file
Additional File 2
Individual landmark data, centroid size and sex. Cartesian coordinates (X – Y format) for fourteen landmarks, with individual group affinities and sampling site as of Table 3 as well as sex (0 = female; 1 = male) and centroid size.
Click here for file
Acknowledgements
We thank D. Tautz for critical comments on the manuscript. AN thanks D. Tautz for his support of the Cottus project. The Cottus project was funded by the DFG (Ta99/20). We have obtained permissions to sample and take specimens from A. Mellin, T. Heilbronner and W. Fettweis, whom we thank for their benevolent support.
==== Refs
Burke JM Arnold ML Genetics and the fitness of hybrids Annu Rev Genet 2001 35 31 52 11700276 10.1146/annurev.genet.35.102401.085719
Seehausen O Hybridization and adaptive radiation Trends Ecol 3 vol 2004 19 198 207 10.1016/j.tree.2004.01.003
Schliewen UK Klee B Reticulate sympatric speciation in Cameroonian crater lake cichlids Frontiers in Zoology 2004 1 5 15679917 10.1186/1742-9994-1-5
Rieseberg LH Archer MA Wayne RK Transgressive segregation, adaptation, and speciation Heredity 1999 83 363 372 10583537 10.1038/sj.hdy.6886170
Rieseberg LH Widmer A Arntz AM Burke JM The genetic architecture necessary for transgressive segregation is common in both natural and domesticated populations Phil Trans R Soc Lond B 2003 58 1141 1147 12831480 10.1098/rstb.2003.1283
Harrison RG Harrison RG Hybrids and hybrid zones Hybrid zones and the Evolutionary process 1993 Oxford University Press New York
Barton NH The role of hybridization in evolution Mol Ecol 2001 10 551 568 11298968 10.1046/j.1365-294x.2001.01216.x
Englbrecht CC Freyhof J Nolte A Rassmann K Schliewen U Tautz D Phylogeography of the bullhead Cottus gobio (Pisces: Teleostei: Cottidae) suggests a pre-Pleistocene origin of the major central European populations Mol Ecol 2000 9 709 722 10849287 10.1046/j.1365-294x.2000.00912.x
Volckaert FAM Hänfling B Hellemans B Carvalho GR Timing of the population dynamics of bullhead Cottus gobio (Teleostei: Cottidae) during the Pleistocene J Evol Biol 2002 15 930 944 10.1046/j.1420-9101.2002.00469.x
Nolte AW Freyhof J Stemshorn KC Tautz D An invasive lineage of sculpins, Cottus sp. (Pisces, Teleostei) in the Rhine with new habitat adaptations has originated from hybridization between old phylogeographic groups Proc Roy Soc B
Freyhof J Kottelat M Nolte A Taxonomic diversity of European Cottus with description of eight new species (Teleostei: Cottidae) Ichthyological Exploration of Freshwaters 2005 16 107 172
Zelditch ML Swiderski DL Sheets HD Fink WL Geometric Morphometrics: A primer 2004 Academic Press, London
Klingenberg CP Leamy LJ Quantitative genetics of geometric shape in the mouse mandible Evolution 2002 55 2342 2352 11794792
Albertson RC Streelman JT Kocher TD Genetic Basis of Adaptive Shape Differences in the Cichlid Head Journal of Heredity 2003 A4 291 301 12920100 10.1093/jhered/esg071
Moraes EM Manfrin MH Laus AC Rosada RS Bomfin SC Sene FM Wing shape heritability and morphological divergence of the sibling species Drosophila mercatorum and Drosophila paranaensis Heredity 2004 92 466 473 15107807 10.1038/sj.hdy.6800442
Falush D Stephens M Pritchard JK Inference of Population Structure Using Multilocus Genotype Data: Linked Loci and Correlated Allele Frequencies Genetics 2003 164 1567 1587 12930761
Paetkau D Calvert W Stirling I Strobeck C Microsatellite analysis of population structure in Canadian polar bears Mol Ecol 1995 4 347 354 7663752
Cornuet J-M Piry S Luikart G Estoup A Solignac M New Methods Employing Multilocus Genotype to Select or Exclude Populations as Origins of Individuals Genetics 1999 153 1989 2000 10581301
Primmer CR Koskinen MT Piironen J The one that did not get away: individual assignment using microsatellite data detects a case of fishing competition fraud Proceedings of the Royal Society of London Series B Biological Sciences 267 1699 1704
Strauss RE Natural hybrids of the freshwater sculpins Cottus bairdi and Cottus cognatus (Pisces: Cottidae): Electrophoretic and morphometric evidence Am Midl Nat 1986 115 87 105
Riffel M Schreiber A Morphometric differentiation of sculpin (Cottus gobio), a fish with deeply divergent genetic lineages Can J Zool 1998 76 876 885 10.1139/cjz-76-5-876
Bookstein FL Morphometric Tools for Landmark Data: Geometry and Biology 1991 Cambridge University Press
Rohlf FJ Marcus LF A Revolution in Morphometrics Trends in Ecology and Evolution 1993 8 129 132 10.1016/0169-5347(93)90024-J
Dryden IL Mardia KV Statistical shape analysis 1998 John Wiley & Sons, New York
Rohlf FJ Shape Statistics: Procrustes Superimpositions and Tangent Spaces Journal of Classification 1999 16 197 223 10.1007/s003579900054
Slice DE Landmark coordinates aligned by Procrustes analysis do not lie in Kendall's shape space Systematic Biology 2001 50 141 149 12116591
Kendall DG The diffusion of shape Advances in Applied Probability 1977 9 428 430
Bookstein FL Marcus LF, Corti M, Loy A Combining the tools of geometric morphometrics Advances in Morphometrics 1996 Plenum Press 131 152
Bookstein FL Principal warps: thin-plate splines and the decomposition of deformations IEEE Transactions on Pattern Analysis and Machine Intelligence 1989 11 567 585 10.1109/34.24792
Bookstein FL Marcus LF, Corti M, la oy A Standard formula for the uniform shape component in Landmark data Advances in Morphometrics 1996 Plenum Press 153 168
Krzanowski WJ Principles of Multivariate Analysis 1988 Oxford University Press
Mardia KV Kent JT Bibby JM Multivariate Analysis 1979 Academic Press, London
Piry S Cornuet J-M Gene-Class Users Manual
Manly BF Randomization, Bootstrap and Monte Carlo Methods in Biology 1997 Chapman & Hall
Efron B Tibshirani RJ An Introduction to the Bootstrap 1993 Chapman & Hall
Nolte AW Stemshorn KC Tautz D Direct cloning of microsatellite loci from Cottus gobio through a simplified enrichment procedure Mol Ecol 3 Notes
Shriver MD Smith MW Jin L Marcini A Akey JM Deka R Ferrell RE Ethnic-affiliation estimation by use of population-specific DNA markers American Journal of Human Genetics 1997 60 957 964 9106543
Whichloci
Stemshorn KC Nolte AW Tautz D A Genetic Map of Cottus gobio (Pisces, Teleostei) based on microsatellites can be linked to the Physical Map of Tetraodon nigroviridis J Evol Biol
Structure 2.1
Rohlf FJ TpsDig, digitize landmarks and outlines, version 1.39 2003 Department of Ecology and Evolution, State University of New York at Stony Brook
Sheets HD IMP-Integrated Morphometrics Package 2002 Department of Physics, Canisius College, Buffalo, New York
|
15987531
|
PMC1198242
|
CC BY
|
2021-01-04 16:38:34
|
no
|
Front Zool. 2005 Jun 29; 2:11
|
utf-8
|
Front Zool
| 2,005 |
10.1186/1742-9994-2-11
|
oa_comm
|
==== Front
Global HealthGlobalization and Health1744-8603BioMed Central London 1744-8603-1-91601879910.1186/1744-8603-1-9EditorialTo quell obesity, who should regulate food marketing to children? Kelly Ben [email protected] Public Health Advocacy Institute in Boston, MA, USA2005 14 7 2005 1 9 9 25 6 2005 14 7 2005 Copyright © 2005 Kelly; licensee BioMed Central Ltd.2005Kelly; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
The global hegemony of the United States in the production and marketing of food, while a marvel of economic success, has contributed to the epidemic of obesity that is particularly afflicting children. So far the U.S. government has declined to regulate the aggressive ways in which food producers market high-energy, low-nutrition foods to young people. That public-health responsibility has been left to an industry-created scheme of self-regulation that is deeply flawed; there is a compelling need for government involvement. The issue is certain to be raised by health advocates at a U.S. Federal Trade Commission meeting in mid-July to discuss the self-regulatory approach, but the outlook for remedies to emerge from the meeting is not encouraging.
==== Body
United States businesses have been unsurpassed leaders in proliferating the availability, marketing and consumption of high-energy, low-nutrition foods at home and around the world. Their aggressive, high-priced marketing of those foods has especially targeted children – the adult consumers of the future in whom the early creation of brand and product commitment is, in the eyes and ledgers of some huge corporations, essential to profitability.
But the leadership in so-called "junk food" promotion has not been matched in the U.S. by effective regulatory controls to prevent marketing abuses of children. Advertising of such products has been misleading, overblown, and seemingly bent on undermining the ability of parents to moderate their children's eating practices. Although still concentrated in television viewing hours directed at child entertainment and publications read by children, it is spreading rapidly to other means of messaging, among them online games ("advergames"), product placements in films and television shows, and product-linked websites.
The position of the driving-force industries – food marketers and advertising agencies – presents an interesting internal contradiction. On one hand they assert valid evidence is lacking that exposure to such messages actually influences the consumption practices of children, let alone contributes to health problems such as obesity. On the other they claim that abusive child-directed marketing is being effectively and adequately controlled by an industry-sponsored system of self-regulation. Health advocates, meanwhile, are increasingly urging that the government intervene by legislating and enforcing objective standards over and above any that the industry imposes or claims to impose.
The regulatory agency that would set and implement such standards is the U.S. Federal Trade Commission (FTC), but at present it contends it cannot do so without additional statutory authority. Some in Congress, prominent among them Senator Tom Harkin, have proposed legislation to create such authority, but Harkin and like-minded lawmakers are members of the minority Democratic Party. President Bush and the majority members of Congress are on record as opposing strictures on the scope of food marketing; the view was reflected in efforts by U.S. spokespersons last year, made on behalf of American food companies, to rein in the World Health Organization's proposed nutrition guidelines.
FTC now plans to hold a public workshop, "Perspectives on Marketing, Self-Regulation, and Childhood Obesity," in Washington, D.C. on July 14–15, jointly with the Department of Health and Human Services. The description of the workshop – an "open discussion of self-regulation and the marketing of food and beverages to children" – suggests that the option of government controls will not be considered.
If that is the case, the proceeding will be limited to discussions of the adequacy of the existing self-regulatory scheme. The locus of that scheme, operated by the food and advertising industries, is the 20-year-old Children's Advertising Review Unit (CARU). It comprises a five-person staff within the Council of Better Business Bureaus and is linked organizationally to a network of trade groups representing manufacturers and advertising agencies. Documents found on CARU's website, , describe the history, governance and mission of the activity. They do so with scant reference to public-health needs.
CARU claims for itself a high level of effectiveness, but does so by using measures that are seriously flawed. The measures assume that the guidelines and principles that CARU claims to impose on child-directed food marketing messages are meaningful. They imply that CARU in fact proactively sees, screens, and acts on a large body of such messages. And they suggest that CARU has powers to enforce its decisions against messages found to violate its guidelines and principles. In fact, none of these assumptions is accurate.
A recent article in the Wall Street Journal, the leading pro-business newspaper in the United States, put into perspective the reality of CARU's relationship to the marketing activities it claims to regulate. Under the headline "General Mills Touts Sugary Cereal as Healthy Kids Breakfast," the June 22 article reported that the giant food packager "plans to launch a national ad campaign targeted at children that will tout the health benefits of eating breakfast cereal – including Trix, Cocoa Puffs and other sugary ones it sells."
After noting the controversial nature of the General Mills plan and the opposition it is provoking among obesity control proponents, the article noted that CARU "is endorsing the General Mills campaign after the company sought the organization's input. 'I think it is responsible advertising,' said Elizabeth Lascoutx, director of the group, which is partly funded by children's advertisers, including General Mills. 'They're encouraging a behavior that is healthful' as opposed to not eating breakfast." Subsequently Lascoutx was quoted – in a news story originating in Golden Valley, Minnesota, General Mills's home base – as saying, "This is exactly what a leader in the food industry should be doing. Ensuring that positive, nonbranded health messages like 'Choose Breakfast' are being delivered to children is not only responsible, but commendable."
CARU's level of comfort with the pitching of sugared cereals to kids in the midst of an obesity epidemic is consistent with the intimacy of its connections with the very companies whose activities it is supposed to regulate. For instance, its funding comes from such industry giants, to name a few, as Burger King Corp., Frito-Lay, Inc., Grocery Manufacturers of America, Inc., Hershey Foods Corp., Kellogg Company, Kraft Foods, Inc., Masterfoods USA, McDonald's Corporation, National Confectioners Association, Nestle USA, Inc., Pepsico Beverages & Food, Sara Lee Corporation, and – last but clearly not least – General Mills itself.
At its July public meeting FTC is sure to hear from health advocates about the child-targeting marketing behavior of General Mills and other volume purveyors of high-energy, low-nutrition food to children. Some of those advocates will be urging the agency to use its existing authority to regulate that behavior, and to seek additional statutory authority if it is needed. It is likely, though, that the agency will turn a deaf ear to such entreaties and instead will emphasize the alleged benefits of the CARU scheme while asking for suggestions to strengthen it.
In a paper submitted to the FTC meeting docket, the Public Health Advocacy Institute , a Boston-based organization working to foster greater support for public health goals by the law community, has sharply disputed the premise that the CARU scheme works, or can be made to work, effectively. "Industry Controls Over Food Marketing: Are They Effective?" reviews a range of global assessments of self-regulation in general. It presents an extensive review and summary of leading commentaries addressing the world-wide state of regulation directed at food advertising that targets children. It concludes that the U.S. self-regulatory system is ineffective when measured against available criteria for gauging the adequacy of self-regulation, and also ineffective in the context of the worsening obesity epidemic and its damaging impact on children.
Using benchmarks drawn from a number of studies and commentaries concerned with self-regulation that have been published in the U.S. and abroad, the PHAI paper finds the CARU scheme and its potential for effectiveness to be seriously flawed and not remediable. It notes among other things that CARU fails to "provide an adequate public interest response" to public health needs; lacks "strong independent input, well-resourced monitoring and tough sanctions for breaches of the rules"; applies subjective criteria in assessing advertisements; does not review advertising prior to dissemination; lacks third-party review of its decision, and cannot enforce its decisions, which can be ignored by advertisers
Further, according to the paper, CARU fails to meet self-regulatory criteria indicated for the global business community (of which leading food marketers are a major component) in a study published by the United Nations Research Institute for Social Development. 1 Its failures include:
• Lack of independent monitoring, which is "crucial" to effective self-regulation. "...implementation can only be guaranteed where there is an element of independent monitoring of codes of conduct."
• Unrealistic performance claims, which have sometimes "led to a worsening of the situation of those whom they purport to benefit."
• Adoption of self-regulatory schemes "simply to pre-empt external pressure."
• Weaknesses in implementation and compliance, such as a lack of "clear methods of implementation and means to ensure that it is being complied with..."
• Discouragement of stakeholder involvement: "It is in this area that the contrast between rhetoric and reality is particularly jarring. In the absence of independent monitoring and verification, it is difficult to evaluate whether company codes are applied extensively in practice or remain mere expressions of good intentions."
• Absence of sanctions.
How will these conclusions influence the FTC at its July public meeting? Although the outlook for health advocates is not encouraging given the agency's long-avowed support for CARU's self-regulatory scheme and the reflexive aversion to regulation of the party in power in Washington, that has not deterred groups such as Campaign for a Commercial-Free Childhood, The Center for Informed Food Choices, and the Strategic Alliance for Healthy Food and Activity Environments from raising the issue in their comments to the agency.
In contrast, the Grocery Manufacturers Association, speaking for the most powerful food corporations in America, has made clear that it expects the agency to be fully supportive of the CARU self-regulation approach. The issues it proposes to see raised at the July meeting emphasize "the role of self-regulation in promoting responsible advertising, including to children," and "how advertising will be part of the solution to the problem of obesity." If nothing else comes of the FTC meeting, it will at least be worth watching to see whether the Association will be challenged to reconcile its notion of advertising as "part of the solution to the obesity problem" with the new push to sell sweetened cereals being launched by one of its biggest members, General Mills.
Note
1Rhys Jenkins, Corporate Codes of Conduct Self-Regulation in a Global Economy, Technology, Business and Society. Programme Paper Number 2, April 2001
|
16018799
|
PMC1198243
|
CC BY
|
2021-01-04 16:39:02
|
no
|
Global Health. 2005 Jul 14; 1:9
|
utf-8
|
Global Health
| 2,005 |
10.1186/1744-8603-1-9
|
oa_comm
|
==== Front
Health Qual Life OutcomesHealth and Quality of Life Outcomes1477-7525BioMed Central London 1477-7525-3-491612020610.1186/1477-7525-3-49ResearchHealth-related quality of life and physical well-being among a 63-year-old cohort of women with androgenetic alopecia; a Finnish population-based study Hirsso Päivi [email protected] Ulla [email protected] Mauri [email protected] Liisa [email protected]ärkönen Pirjo [email protected]änen-Kiukaanniemi Sirkka [email protected] Department of Public Health Science and General Practice, Box 5000, FIN-90014 University of Oulu, Finland2 Unit of General Practice, Oulu University Hospital, FIN-90029 OYS, Finland3 Oulu Health Center, Box 8, FIN-90015 City of Oulu, Finland4 Oulu Deaconess Institute Department of Sports Medicine, Finland2005 24 8 2005 3 49 49 25 4 2005 24 8 2005 Copyright © 2005 Hirsso et al; licensee BioMed Central Ltd.2005Hirsso et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The aim of this study was to assess the possible associations between female androgenetic alopecia (AGA), insulin resistance and health-related quality of life (HRQOL)-linked factors in women. We hypothesized that not only the mental aspects but also certain physical aspect of women's health, such as insulin resistance, have an important role in the determination of HRQOL among women with hair loss.
Methods
A population-based cohort of 330 healthy women aged 63 years, who participated in this study in the City of Oulu in Northern Finland, underwent a medical check-up including assessment of hair status on Ludwig's scale. Background data were collected with a standard questionnaire including a validated RAND 36-Item Health Survey (RAND-36) questionnaire.
Results
105 (31%) women with AGA and 225 (69%) controls completed the RAND-36 questionnaire. The women with AGA were more insulin-resistant than the women with normal hair (QUICKI 0.337 vs. 0.346, p = 0.012). Impaired glucose regulation (IGR) was more prevalent among the former than the latter group (39% vs. 25%). The mean RAND-36 scores were significantly lower on the dimensions of physical functioning, role limitation due to physical health and general health, but not on the mental or social dimensions, among the women with AGA compared with the controls. In multivariate logistic regression analyses with the lowest quintiles of the HRQOL dimensions as the dependent variables and AGA, depression, marital status, education and IGR or QUICKI as independent variables, AGA was independently associated with role limitations due to physical health (2.2, 95% CI 1.20–4.05, 2.45 95% CI 1.32–4.55, respectively).
Conclusion
In women aged 63 years, AGA was associated with role limitations due to physical health. Furthermore, the prevalence rates of IGR and insulin resistance measured by QUICKI were higher among the women with hair loss than those with normal hair.
androgenic alopeciahair lossinsulin resistancehealth related quality of life
==== Body
Background
Androgenetic alopecia (AGA) has been observed in women, with an increasing prevalence particularly after the age of 50 years [1]. Moreover, it has been suggested that hair loss could have an adverse effect on psychosocial life and self-esteem among both genders [2-7]. A number of studies have shown that, in selected clinical populations, androgenic alopecia is associated with poorer quality of life scores on disease-specific quality of life questionnaires [8,9]. Women with AGA describe their self-perceived health and psychosocial situation more negatively (e.g. low self-esteem, emotional distress due to AGA, social anxiety) than those with normal hair [2,5,8,9]. In addition, one study suggested that hair loss in women could be associated with psychiatric disorders, e.g. depression [3].
It is unclear why women with AGA would have a poorer quality of life. As most studies have looked at these associations in subgroups of women with underlying illness, it is plausible that the women with AGA may have an underlying subclinical disease. In males, an association between AGA and insulin resistance has been demonstrated [10,11], but the data on women are less conclusive [12].
We explored the association between AGA and various dimensions of HRQOL among community-dwelling women. We further explored whether any differences in the quality of life between those with and without AGA could be explained by an underlying metabolic abnormality. The aim of this study was to assess the potential associations between AGA, insulin resistance (IR) and the different HRQOL dimensions among women at the population level.
Methods
A population-based study was carried out in northern Finland in 1990–1992 to assess, among other things, the prevalence of diabetes mellitus (DM) and impaired glucose tolerance (IGT). Altogether 831 subjects (435 women) born in 1935 and living in Oulu, a city with 100,000 inhabitants, participated in the study. They were invited to attend a follow-up study in 1996–1998, and 593(71%) subjects (347 women, 83%) attended. A detailed description of the data has been published earlier [13].
As part of the clinical examination, hair status (330 women, 93% of those participated) was assessed by the same trained nurse, using Ludwig's scale [14], and re-classified into two classes for the analysis. The original classes 0 and I (no or minimal hair loss) were combined as a normal hair group and the original classes II and III (moderate or marked hair loss) were combined as a hair loss group.
We measured health-related quality of life (HRQOL) using the Finnish version of the RAND 36-Item Health Survey 1.0 (RAND-36) [15]. This self-reported measure is composed of eight separate scales assessing physical functioning (10 items), role limitations due to physical health (4 items), role limitations due to emotional problems (3 items), energy/fatigue (4 items), emotional well-being (5 items), social functioning (2 items), pain (2 items) and general health (5 items). All scale scores range from 0 to 100, with 100 representing the most favourable functioning/well-being, and the minimal clinically important difference is cautiously suggested to be 3–5 points [16,17]. RAND-36 [18] includes the same items as the Item Short-Form Health Survey SF-36 [19], but the scoring algorithm has been slightly modified [18].
From the questionnaires, we assessed various sociodemographic variables, including the respondents' current marital status (married, unmarried/divorced or widowed) and educational attainment (elementary, secondary or university). Self-perceived health and the participant's own opinion of his or her overall physical fitness and life satisfaction were also asked (good, moderate or poor). A sum score of overall physical capacity was constructed based on their ability to walk up one flight of steps, a few flights of steps, half a kilometre or two kilometres or to run one hundred metres. The total score ranged from 0 to 20, with a higher score indicating poorer physical capacity. Depressive symptoms were measured using the Beck Depression Inventory (BDI) [20] with scores ≥ 10 defined as indicative of depressive symptoms.
Anthropometric measurements (weight, height, waist circumference and hip circumference) were obtained and used to calculate body mass index (weight/height2) (BMI) and waist-to-hip ratio (WHR). Blood pressure (BP) was measured by a physician from both arms in a sitting position. The mean value of these two measurements was used in the analyses. Hypertension was defined as either a systolic blood pressure ≥ 160 mmHg or a diastolic blood pressure ≥ 90 mmHg or being on antihypertensive medication regardless of the blood pressure values. Prevalent chronic disease (hypertension, ischemic heart disease, diabetes, stroke, intermittent claudication, arthritis) was based on a medical diagnosis as reported by the participant on the questionnaire or the use of medication for any of these diseases as reported during the clinical examination.
After an overnight fast, blood samples were collected using a standardized procedure and assayed for blood glucose (fB-gluc), two-hour blood glucose concentration (2 h-gluc) and serum insulin (excluding diabetic patients with insulin treatment). A standardized 75-g oral glucose tolerance test (OGTT) was performed. We defined the participants as having impaired glucose regulation (IGR) if they met the criteria for having impaired fasting glucose (IFG), IGT and DM based on the WHO 1997 criteria [21]. To measure insulin sensitivity, a quantitative insulin sensitivity check index (QUICKI) was used [22,23]. QUICKI was determined from the fasting insulin and glucose values according to the equation: QUICKI = 1/ [log(fasting insulin) + log(fasting glucose)]. Diabetic subjects with antidiabetic medication were excluded from the analysis.
Statistical analyses
For these analyses, we used data from 347 women with complete hair status scores (n = 330). Descriptive comparisons of the groups defined by hair status were presented as cross-tabulations and percentages for the categorical variables and assessed with the T-test when the distribution was normal or with the Mann-Whitney U-test in the case of a non-normal distribution of the continuous variables. The statistical differences between the scores of the HRQOL dimensions were tested by the Mann-Whitney U-test [24]. The HRQOL dimensions on which the women with significant hair loss scored lower than the women with normal hair were analyzed in more detail. After bivariate comparisons, multivariate logistic regression analyses with the lowest quintile of HRQOL as the dependent variables and AGA, BDI, IGR/QUICKI, marital status and education as independent variables were made. The possible interactions of hair status with depression (AGA*BDI) and impaired glucose regulation with depression (IGR*BDI) were tested by interaction terms.
Results
The prevalence of moderate to marked hair loss (grade II-III on Ludwig's scale) was 31% among the study population. Table 1 gives the percentages and Table 2 gives the means of the background characteristics of the women stratified into the categories of normal hair (grade 0 and I on Ludwig's scale) and hair loss (grades II and III on Ludwig's scale). In bivariate analysis, the women with hair loss had statistically significantly lower self-perceived health, self-perceived physical fitness and life satisfaction compared to the women with normal hair. (Table 1) Waist circumference and WHR were significantly higher and BMI tended to be higher (p = 0.094) among the women with hair loss compared with those with normal hair status. The mean QUICKI was lower, indicating higher IR, in the women with hair loss compared with the women with normal hair. (Table 2) The sum score of overall physical capacity was significantly higher (describing the limitation in physical functions) among the women with hair loss compared with the women with normal hair (mean sum score 7.9 ± 3.1 vs. 7.2 ± 2.8, p = 0.015).
Table 1 Percentages of some background characteristics among the study groups
Variable Hair loss (n = 105) Normal hair (n = 225)
% (n) % (n) p
Marital status 0.163
married 59 (61) 69 (154)
not married/divorced 27 (28) 18 (41)
widow 14 (14) 13 (29)
Educational level 0.115
elementary 89 (92) 80 (181)
secondary 5 (5) 11 (25)
tertiary 6 (6) 8 (19)
Self-perceived health 0.030
good 41 (43) 54 (121)
moderate 48 (50) 42 (94)
poor 11 (11) 4 (10)
Self-perceived physical fitness 0.001
good 26 (27) 46 (104)
moderate 54 (56) 44 (99)
poor 20 (21) 10 (22)
Life satisfaction 0.044
satisfied 66 (69) 79 (176)
moderately satisfied 32 (34) 20 (45)
dissatisfied 2 (2) 1 (3)
BDI1) 0.311
no depression 76 (78) 83 (173)
depressive symptoms 22 (21) 17 (35)
1) BDI = Beck's Depression Inventory; cut point ≥ 10
Table 2 Mean and standard deviations (SD) of background characteristics among women with hair loss and normal hair
Variable Hair loss n = 105 Normal hair n = 225
Mean Sd Mean Sd p
Waist circ. (cm) 86.6 11.4 83.0 10.9 0.010
BMI 1) (kg/m2) 28.5 4.5 27.5 4.0 0.094
WHR2) (waist/hip) 0.84 0.06 0.81 0.06 0.002
fB-gluc3) (mmol/l) 5.2 1.0 5.0 0.8 0.064
2-h gluc4) (mmol/l) 7.2 1.8 7.1 1.8 0.321
QUICKI5) 0.337 0.03 0.346 0.03 0.012
1) BMI = body mass index (kg/m2); 2) WHR = waist to hip ratio; 3) fB-gluc = fasting glucose level (mmol/l); 4) 2-h gluc = 2-h blood glucose level (mmol/l); 5) QUICKI = Qualitative Insulin Sensitivity Check Index
There were no significant differences between the hair status categories of the women reporting heart failure, hypertension or musculoskeletal symptoms, but the women with hair loss self-reported more DM than those with normal hair (14% vs. 6%). The corresponding prevalences for IFG and IGT were as follows (2% vs. 5%, 24% vs.16%, respectively). The prevalence of IGR (cluster of self-reported DM and IFG, IGT and DM according to OGTT) defined by WHO was different between the groups (43% vs. 30%, p = 0.015).
The women with hair loss scored significantly lower on the physical functioning, role limitations due to physical health and general health perceptions dimensions of HRQOL compared with the women having normal hair. Also, a nearly statistical significant difference was seen in the pain score between the groups. (Table 3).
Table 3 Means/medians and standard deviation (SD) / interquartile range of HRQOL dimensions of women with hair loss and normal hair
Scale Hair loss (n = 105) Normal hair (n = 225)
Mean SD Mean SD p
Physical functioning 71,2 20,7 77,9 19,4 0,004
Energy/fatigue 69,0 19,3 70,4 17,9 0.431
Emotional well-being 78,9 15,5 78,7 16,4 0.557
Social functioning 85,7 19,3 88,7 18,5 0.170
Pain 68,6 24,7 73,2 24,9 0,085
General health 55,9 17,3 60,8 16,8 0,007
Median Q1–Q3 Median Q1–Q3
Role limitations due to physical health 75 25–100 100 50–100 0,032
Role limitations due to emotional problems 100 33–100 100 67–100 0,350
The interaction terms of depression with hair loss and depression with IGR were not statistically significant, suggesting that the effects of hair loss and IGR on the dimensions of HRQOL were not different in depressive and non-depressive subjects. The multivariate logistic regression analyses were made to analyze the independent association of AGA with the HRQOL dimensions. Two different adjusted models are presented in Table 4. In model a), glucose status is presented as IGR, and in model b), glucose metabolism is described with QUICKI. Because depressive symptoms, marital status and education are associated with HRQOL, they were included in the models as confounding factors. In both models AGA was independently associated with role limitations due to the physical health dimension of HRQOL, the odd rations being 2.2 (1.20–4.05) in model a) and 2.45 (1.32–4.55) in model b). In the dimensions of physical functioning and general health, no such association between AGA and HRQOL was seen. In addition, depressive symptoms were independently associated with role limitations due to physical health and general health.
Table 4 Multivariate logistic regression analyses for impaired quality of life. Odds ratios for risk of belonging to the lowest quintiles in the dimensions of quality of life
Variable Physical functioning
Model a Model b
OR (% CI) OR (% CI)
BDI1) 1.96 (0.98–3.92) 2.21(1.10–4.44)
IGR2) 2.30 (1.29–4.25)
QUICKI3) 2.60 (1.33–5.09)
AGA4) 1.47 (0.79–2.71) 1.58 (0.85–2.95)
Marital status 0.78 (0.28–2.17) 0.78 (0.28–2.21)
Education level 2.39 (0.88–6.53) 0.44 (0.59–1.20)
Role limitations due to physical health
Model a Model b
BDI1) 4.81 (1.20–4.05) 5.36 (2.75–10.6)
IGR2) 1.79 (0.99–3.24)
QUICKI3) 1.15 (0.55–2.38)
AGA4) 2.20 (1.20–4.05) 2.45 (1.32–4.55)
Marital status 1.24 (0.45–3.48) 1.20 (0.42–3.44)
Education level 0.63 (0.30–1.33) 1.58 (0.74–3.35)
General health
Model a Model b
BDI1) 4.74 (2.42–9.31) 5.67 (2.84–11.3)
IGR2) 1.68 (0.89–3.17)
QUICKI3) 1.62 (0.76–3.45)
AGA4) 1.58 (0.82–3.03) 1.83 (0.93–5.59)
Marital status 1.22 (0.55–3.41) 1.20 (0.34–3.79)
Education level 1.37 (0.55–3.41) 0.75 (0.29–1.20)
Model a; adjusting without QUICKI. Model b; adjusting without IRG.
1) BDI = Beck's Depression Inventory; depressive symptoms ≥ 10.
2) Impaired glucose tolerance according to WHO 1997 criteria.
3) Insulin sensitivity check index; the lowest quintile.
4) Androgenetic alopecia; classes II–III
Discussion
The main finding of this study was that women with AGA had impaired self-perceived health. They were more insulin-resistant when measured by QUICKI as a marker of insulin sensitivity, and impaired glucose regulation was more prevalent among them compared with the women with normal hair. The women with AGA had significantly higher waist circumference and WHR values as markers of abdominal obesity than the women with normal hair. Compared with the women having normal hair, the women with AGA scored significantly lower on three of the eight RAND-36 dimensions: physical functioning, role limitations due to physical health and general health items compared with the women having normal hair. In addition, the scores on all the other RAND-36 dimensions except emotional well-being tended to be lower.
This is the first study of an unselected and representative population of women to report an association of AGA with the general measures of HRQOL, which are recommended for epidemiological studies. The results of the previous studies concerning the quality of life of AGA subjects are somewhat difficult to compare with our results because of methodological and other differences, such as age, limited criteria of quality of life and the selection of study populations [5,8,9]. Actually, the previous studies by Schmidt and co-workers, in which quality of life was measured with an instrument specific for patients with hair loss (Hairdex), showed reduced values in the social and emotional domains in female patients with alopecia [8,9]. Only experiences of negative psychological effects on their quality of life have been reported by women with hair loss [5].
These results are opposite to our result, which might be explained by population differences and differences in the evaluation of HRQOL. Moreover, the women in our study were not seeking treatment for hair loss. The scaling properties and the validity of the Finnish RAND-36 have been tested among randomly selected Finns (3000 subjects aged 18–79 years and 400 subjects aged 65–79 years) [15]. The results were comparable with the results obtained for RAND-36 in international studies, but in the Finnish study, the general health scale correlated more strongly with physical health. In our study, all scores of dimensions were at the same level in this age-group as in the Finnish RAND-36 study, but differences were seen in comparisons of hair status.
The women with AGA were more insulin resistant, and IGR was more prevalent among them compared to those with normal hair. Moreover, central obesity, which was prevalent among women with AGA, is known to be associated with IR, type 2 diabetes mellitus, hypertension, dyslipidemia and coronary heart disease [25]. In our earlier studies, hair loss in 63-year-old women was suggested to be associated with insulin-linked disturbances, such as overweight, hyperinsulinemia and microalbuminuria [12]. A similar association has also been reported in men [10,11,26,27]. Furthermore, an association between depression and insulin resistance has been suggested among this study population [28]. This raises a question about the role of somatic diseases at the background of impaired HRQOL in women with AGA.
It is hence not self-evident that women with AGA have only emotional problems [2,3,5], but they may also have physical problems as well as chronic diseases and risk factors that need to be evaluated. According to our results, the physical dimensions, but not the mental health dimensions, have an important role in the determination of HRQOL in women with hair loss. In this study, the prevalence of depressive symptoms was not significantly higher in women with hair loss compared with women having normal hair. However, depression seemed to be a significant risk for lower HRQOL on the dimensions of physical health, role limitations due to physical health and general health in this cohort, which is in accordance with earlier studies [18,29]. This finding might be explained by the similarities in the measurement of psychological dimensions of HRQOL and BDI.
One limitation of our study may be that the hair classification was done by combining grade 0 and grade I (minimal hair loss). The reason is that they are difficult to differentiate from each other, especially in Finnish subjects with light and thin hair. Moreover, the classes II and III were combined because of the low number of women in class III (n = 4). The strengths of our study include the large and representative population sample of one female age group, giving the study the relevance of an epidemiologic survey, and the standardized scale used by the same trained nurse as part of the clinical examination, which means that inter-rater variation was lacking. We used well-documented questionnaires (RAND-36) to measure quality of life, which is a multidimensional phenomenon that includes several aspects of well-being. RAND-36 has well-documented reliability and validity and is useful in describing HRQOL in epidemiological studies of unselected populations [30].
Conclusion
It can be concluded that AGA among 63-year-old women was associated with the physical aspects of HRQOL, but not with the social or mental health aspects. Hair loss was independently associated with role limitations due to physical health. In this study, AGA was not associated with the psychosocial aspects of HRQOL or depression. Instead, AGA was associated with IR, IGT and central obesity. Because these conditions are linked with somatic diseases, they can explain this finding of an association of AGA with impaired physical function. Hair loss could be a marker of poorer physical health, and women developing female AGA might benefit from attention to cardiovascular diseases and their risk factors in medical check-ups.
Authors' contributions
PH conceived the study, reviewed the literature, analyzed the data and wrote the initial and subsequent drafts. UR, ML, LH and PiH collected the data and contributed to the study design, interpretation and revisions of the manuscript. SKK helped to conceive the study and revise the initial and subsequent drafts and supervised the research group. All authors read and approved the final manuscript.
==== Refs
Norwood OT Incidence of female androgenetic alopecia (female pattern alopecia) Dermatol Surg 2001 27 53 54 11231244 10.1046/j.1524-4725.2001.00124.x
Cash TF Price VH Savin RC Psychological effects of androgenetic alopecia on women: comparisons with balding men and with female control subjects J Am Acad Dermatol 1993 29 568 575 8408792
Maffei C Fossati A Rinaldi F Riva E Personality disorders and psychopathologic symptoms in patients with androgenetic alopecia Arch Dermatol 1994 130 868 872 8024274 10.1001/archderm.130.7.868
van der Donk J Passchier J Knegt-Junk C van der Wegen-Keijser Nieboer C Stolz E Verhage F Psychological characteristics of women with androgenetic alopecia: a controlled study Br J Dermatol 1991 125 248 252 1911317
van der Donk J Hunfeld JA Passchier J Knegt-Junk KJ Nieboer C Quality of life and maladjustment associated with hair loss in women with alopecia androgenetica Soc Sci Med 1994 38 159 163 8146707 10.1016/0277-9536(94)90311-5
Girman CJ Hartmaier S Roberts J Bergfeld W Waldstreicher J Patient-perceived importance of negative effects of androgenetic alopecia in women J Womens Health Gend Based Med 1999 8 1091 1095 10565667
Budd D Himmelberger D Rhodes T Cash TE Girman CJ The effects of hair loss in European men: a survey in four countries Eur J Dermatol 2000 10 122 127 10694311
Schmidt S Fischer TW Chren MM Strauss BM Elsner P Strategies of coping and quality of life in women with alopecia Br J Dermatol 2001 144 1038 1043 11359394 10.1046/j.1365-2133.2001.04195.x
Schmidt S Female alopecia: the mediating effect of attachment patterns on changes in subjective health indicators Br J Dermatol 2003 148 1205 1211 12828750 10.1046/j.1365-2133.2003.05327.x
Matilainen V Koskela P Keinanen-Kiukaanniemi S Early androgenetic alopecia as a marker of insulin resistance Lancet 2000 356 1165 1166 11030300 10.1016/S0140-6736(00)02763-X
Hirsso P Rajala U Hiltunen L Laakso M Koskela P Härkönen P Keinänen-Kiukaanniemi S Association of low insulin sensitivity measured by QUICKI with hair loss in 55-year-old men. A Finnish population-based study Diabetes Obes Metab
Matilainen V Laakso M Hirsso P Koskela P Rajala U Keinanen-Kiukaanniemi S Hair loss, insulin resistance, and heredity in middle-aged women. A population-based study J Cardiovasc Risk 2003 10 227 231 12775957 10.1097/00043798-200306000-00012
Rajala U Laakso M Päivänsalo M Pelkonen O Suramo I Keinänen-Kiukaanniemi S Low Insulin sensitivity Measured by Both Quantitative Insulin Sensitivity Check Index and Homeostasis Model Assessment Method as a Risk Factor of Increased Intima-Media Thickness of the Carotid Artery J Clin Endocrinol Metab 2002 87 5092 5097 12414877 10.1210/jc.2002-020703
Ludwig E Classification of the types of androgenetic alopecia (common baldness) occurring in the female sex Br J Dermatol 1977 97 237 254 144518
Aalto A-M Aro AR Teperi J RAND-36 terveyteen liittyvän elämänlaadun mittarina – Mittarin luotettavuus ja suomalaiset väestöarvot Tutkimuksia 1999 Helsinki Stakes 101
Samsa G Edelman D Rothman ML Williams GR Lipscomb J Matchar D Determining clinically important differences in health status measures: a general approach with illustration to the Health Utilities Index Mark II Pharmacoeconomics 1999 15 141 155 10351188
Hays RD Morales LS The RAND-36 measure of health-related quality of life Ann Med 2001 33 350 357 11491194
Hays RD Sherbourne CD Mazel RM The RAND 36-Item Health Survey 1.0 Health Econ 1993 2 217 227 8275167
Ware JE JrSherbourne CD The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection Med Care 1992 30 473 481 1593914
Beck AT Ward CH Mendelson M Mock J Erbaugh J An inventory for measuring depression Arch Gen Psychiatry 1961 4 561 571 13688369
Alberti KG Zimmet PZ Definition, diagnosis and classification of diabetes mellitus and its complications. Part 1: diagnosis and classification of diabetes mellitus provisional report of a WHO consultation Diabet Med 1998 15 539 553 9686693 10.1002/(SICI)1096-9136(199807)15:7<539::AID-DIA668>3.0.CO;2-S
Katz A Nambi SS Mather K Baron AD Follmann DA Sullivan G Quon ML Quantitative insulin sensitivity check index: a simple, accurate method for assessing insulin sensitivity in humans J Clin Endocrinol Metab 2000 85 2402 2410 10902785 10.1210/jc.85.7.2402
Hrebicek J Janout V Malincikova J Horakova D Cizek L Detection of insulin resistance by simple quantitative insulin sensitivity check index QUICKI for epidemiological assessment and prevention J Clin Endocrinol Metab 2002 87 144 147 11788638 10.1210/jc.87.1.144
Walters SJ Campbell MJ The use of bootstrap methods for analysing health-related quality of life outcomes (particularly the SF-36) Health Qual Life Outcomes 2004 2 70 15588308 10.1186/1477-7525-2-70
Pi-Sunyer FX The obesity epidemic: pathophysiology and consequences of obesity Obes Res 2002 10 S97 S104
Lesko SM Rosenberg L Shapiro S A case-control study of baldness in relation to myocardial infarction in men JAMA 1993 269 998 1003 8429606 10.1001/jama.269.8.998
Lotufo PA Chae CU Ajani UA Hennekens CH Manson JE Male pattern baldness and coronary heart disease: the Physicians' Health Study Arch Intern Med 2000 160 165 171 10647754 10.1001/archinte.160.2.165
Timonen M Laakso M Jokelainen J Rajala U Meyer-Rochow Benno V Keinänen-Kiukaanniemi S Insulin resistance and depression: cross sectional study BMJ 2005 330 17 18 15604155 10.1136/bmj.38313.513310.F71
Saarijarvi S Salminen JK Toikka T Raitasalo R Health-related quality of life among patients with major depression Nord J Psychiatry 2002 56 261 264 12470316 10.1080/08039480260242741
Hemingway H Stafford M Stansfeld A Shipley M Marmot M Is the SF-36 a valid measure of change in population health? Results from the Whitehall II study BMJ 1997 315 1273 1279 9390054
|
16120206
|
PMC1198244
|
CC BY
|
2021-01-04 16:38:15
|
no
|
Health Qual Life Outcomes. 2005 Aug 24; 3:49
|
utf-8
|
Health Qual Life Outcomes
| 2,005 |
10.1186/1477-7525-3-49
|
oa_comm
|
==== Front
Hum Resour HealthHuman Resources for Health1478-4491BioMed Central London 1478-4491-3-61609296410.1186/1478-4491-3-6CommentaryWastage in the health workforce: some perspectives from African countries Dovlo Delanyo [email protected] Population Council, Accra, Ghana2005 10 8 2005 3 6 6 13 5 2004 10 8 2005 Copyright © 2005 Dovlo; licensee BioMed Central Ltd.2005Dovlo; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Sub-Saharan Africa faces a human resources crisis in the health sector. Over the past two decades its population has increased substantially, with a significant rise in the disease burden due to HIV/AIDS and recurrent communicable diseases and an increased incidence of noncommunicable diseases. This increased demand for health services is met with a rather low supply of health workers, but this notwithstanding, sub-Saharan African countries also experience significant wastage of their human resources stock.
Methods
This paper is a desk review to illustrate suggestions that the way human resources for health (HRH) are trained and deployed in Africa does not enhance productivity and that countries are unable to realize the full potential expected from the working life of their health workers. The paper suggests data types for use in measuring various forms of "wastage".
Results
"Direct" wastage – or avoidable increases in loss of staff through factors such as emigration and death – is on the rise, perhaps as a result of the HIV/AIDS epidemic. "Indirect" wastage – which is the result of losses in output and productivity from health professionals' misapplied skills, absenteeism, poor support and lack of supervision – is also common.
HIV/AIDS represents a special cause of wastage in Africa. Deaths of health workers, fear of infection, burnout, absenteeism, heavy workloads and stress affect productivity.
Conclusion
The paper reviews strategies that have been proposed and/or implemented. It suggests areas needing further attention, including: developing and using indicators for monitoring and managing wastage; enhancing motivation and morale of health workers; protecting and valuing the health worker with enhanced occupational safety and welfare systems; and establishing the moral leadership to effectively tackle HIV/AIDS and the brain drain.
==== Body
Introduction
Africa, unlike the other continents, faces a severe human resources crisis in the health sector. The continent's economic performance has been poor, which has affected the ability of countries in sub-Saharan Africa (with few exceptions) to sustain credible health services and to train, employ and use health workers most efficiently. Economic growth has been low or negative in many countries, with investment in health that has generally been inadequate, both as proportions of GDP and in gross terms. Motivation, incentives and productivity and retention of health workers have been severely affected. Furthermore, over the past two decades the population of countries in the SSA region has increased significantly, with a major expansion in the disease burden due to HIV/AIDS, recurring high levels of communicable diseases and recent rises in the incidence of noncommunicable diseases and other diseases related to diet and lifestyle changes. However, in the face of the high demand for health services that the foregoing entailed, sub-Saharan Africa has had a low supply of health workers, and this notwithstanding, also experiences significant wastage of its human resources.
While recognizing the paucity of health workers in Africa and the retention and motivation difficulties, this paper suggests that the way human resources in health are trained, deployed and managed by many countries in the region reduces their productivity. Thus these countries are unable to realize the full potential that could be reasonably expected from the working life of their health workers. The potential of health workers to produce health, even within the constraints alluded to, is often shortened by severe attrition and other more indirect forms of "wastage".
For the purposes of this paper, the author uses the term "wastage" to refer to the loss in utility of health workers/health professionals due to attrition or poor productivity that can be prevented or managed and that is over and above what is expected in normal work situations.
Wastage of human resources may be seen from a variety of viewpoints. In some countries, wastage may result from underuse or non-use of trained personnel, resulting in unemployment caused by overproduction, retrenchment or an inability to absorb certain skill types. Others experience wastage as when a health system is unable to realize the full potential and skills of its health workers even when they are fully employed; this second area of wastage is felt to be a problem for a number of health systems in sub-Saharan Africa. Wastage thus goes beyond mere attrition or the losses, normal or otherwise, that occur within a workforce.
For the purposes of this review and based on the foregoing definition, "wastage" has been classified into two main forms: "direct" and "indirect".
"Direct" wastage occurs when avoidable loss of health personnel arises from factors such as emigration and death (i.e. complete losses to the health sector) and reflects attrition of people from the health workforce. "Indirect" wastage is the result of losses in output and productivity of health professionals, such as those arising from absenteeism and poor performance.
HIV/AIDS is discussed in this paper as a special cause of wastage with combined effects of both direct and indirect wastage. Though various aspects of the impact of HIV/AIDS on health workers reflect either direct or indirect losses, the severity of its effects merits separate attention. Deaths of health workers have risen exponentially in some countries in recent years, and many health workers may be leaving the workforce from fear of infection. Burnout, absenteeism and stress among staff are other effects of HIV/AIDS.
The premise of our discussion is therefore that the health workforce in health in Africa faces many challenges and is in a crisis; countries face the following:
• Preventable exit of professionals from the workforce is a major wastage. Deaths, early retirement, emigration and retrenchment have also shortened the optimal working life of health workers.
• Excess production of some types of health workers has occurred without adequate use or with unemployment or underemployment.
• The supply of health workers may at times not match required skills and scopes of practice, which in turn may not match service delivery needs.
• Poor human resources management results in suboptimal deployment and use of professionals.
• Staff time may be inappropriately applied: for example, in some countries a heavy load of in-service training activities and general administrative duties for technical staff reduces the time available for service delivery.
The method used was to review publications, conference reports, presentations at meetings and other data/information on human resources in health in Africa. The wastage issues raised that were related to the five types of losses mentioned above were then categorized and discussed. Because of the paucity of data from countries on the continent, peer-reviewed papers as well as unpublished country reports and communications were reviewed and analysed.
This paper is intended to help identify and clarify causes of wastage and discuss possible indicators that may assist health care managers to monitor, manage and reduce the various forms of wastage. Using experiences from sub-Saharan Africa, wastage of health workers is illustrated by factors such as increased pre-retirement mortality, early and premature retirement, increased emigration and high levels of workplace accidents and injury. While the review did not determine any current standards or benchmarks that exist to show routinely expected levels for such wastage, trends and comparisons between similar countries point out some problem areas. In the following sections, the concept and evidence of different forms of wastage are discussed, using experiences and information from African countries.
Discussion
In this section, the paper examines the various modes of the two types of wastage defined above and illustrates some of the situations in which these occur. In addition, the paper further reviews, as an exceptional case, the impact of HIV/AIDS on wastage of human resources for health.
Direct wastage
Attrition of workers from any form of employment is an expected factor in human resource management, as workers change jobs, retire or die. However, if for any reason the rate of attrition is higher than normally anticipated, this may reflect a problem.
"Direct wastage" as discussed here represents those losses from the stock of health professionals that are considered to be over and above the norm. For example, the total workforce or a component of it (such as nurses) may be expected to lose members at about 1% of stock per annum. If this rate of losses rises suddenly or changes significantly over time, a retention problem may be indicated. In the case of deaths, a health problem may be indicated.
These changes (increased attrition) may occur over time within the same country or as differences between the attrition rates of two similar countries and their populations of professionals. Direct wastages can present in many forms, and from this review some of the wastage types encountered are represented below.
Movement from health into non-health professions
Health professionals may leave health work altogether and do something completely unrelated. While those who do this have not been fully studied in the literature reviewed, they appear to be a small proportion of the loss of professionals. Dovlo and Nyonator (1999), studying a cohort of 192 doctors in Ghana, found that only two had changed professions completely (had became full-time ministers of religion) [1]. The Mozambique Health Ministry (2003) said 18 nurses changed jobs in 2002, of whom five retrained as doctors and others went into psychology, law, biology, international affairs and geography [2]. A 1999 study of health worker migration showed that the proportion of nurses leaving the workforce who chose to leave before their working life ended ranged from 23% to 78% of all leavers depending on the country [3]. Table 1 shows the proportion of staff departing from the health workforce who leave public sector employment prematurely in some African countries. There is no indication as to whether these leavers remain in health work or within the country. Work done for the Joint Learning Initiative on Human Resources for Health by Dare et al. [4] (also indicated that 7.9% of doctors in Nigeria worked outside the health sector.
Table 1 Voluntary leavers: example of direct wastage in selected countries, 1999
Categories Ghana Lesotho Namibia Malawi
Total leavers % voluntary Total leavers % voluntary Total leavers % voluntary Total leavers % voluntary
Nurses 744 58.7% NA NA 47 78% 43 23%
Doctors 315 84.4% NA NA 16 93.8% 18 83.3%
Others/All - - 50 62% - - 17 76.5%
Source: Dovlo D: Issues Affecting the Mobility and Retention of Health Workers/Professionals in Commonwealth African States. A consultancy report prepared for the Commonwealth Secretariat. London; 1999.
Emigration/brain drain of health professionals
Significant increases in the migration of health professionals have occurred in recent years but monitoring of emigration flows is difficult, as few countries keep adequate statistics. Dovlo and Nyonator (1999) estimated that between 1986 and 1995, 61% of doctors who qualified from one medical school in Ghana left the country [1]. Of these, 6.2% had migrated to another African country (South Africa), but most went to the United Kingdom (55%) or the United States of America (35%). Huddart and Picazo (2003) indicate that 840 out of 1200 doctors trained in Zimbabwe in the 1990s left the country and 17% of locally trained physicians and dentists left the Sudan in the 1980s and 1990s [5]. In the case of Ghana, the physicians had left within 10 years of qualification, after working less than a third of the expected duration of their services.
While the numbers emigrating are in themselves problematic, they also hide serious qualitative consequences that occur when losses of even small numbers of specialists and tutors create a much wider effect on the training of new health workers and in sustaining quality. For example, Martineau and Decker (2002) report that the recruitment of just two specialized anaesthetists from the Boxburg Centre for Spinal Injuries in South Africa by a Canadian Institution led to permanent closure of the unit [6].
In smaller countries, even minimal numbers of migrants represent significant losses. In relatively wealthy Mauritius, the Ministry of Health estimated that 327 nurses (about 12.9% of the nurse workforce) migrated from an average annual nurse workforce level of 2534 between 1998 and 2001. 89% went to the United Kingdom, 8% to New Zealand and 3% to Canada and the United Arab Emirates [7]. While the impact of these losses has not been thoroughly studied, it has been suggested that migration means a loss of the investment made in health workforce development, which creates equity and distribution problems within the country's health systems and causes high workloads, poor quality of care and low morale among health workers [8,9].
Work-induced injury, accidents, deaths as causes of premature loss from the workforce
The past decade has shown marked increases in death rates of health workers in some African countries. Huddart and Picazo (2003) and Tawfik and Kinoti (2003), among others, have ascribed these deaths to the silent impact of HIV and AIDS on the health workforce [5,10]. However, African health systems can also be blamed for work environments that induce high levels of occupational accidents or create a strong perception of high risk [11]. The Chief Nursing Officer of the Ghana Health Service has suggested at a national health workforce forum that high workload and stress may have contributed to higher than usual trends in nurses' deaths (an approximate 26% rise in nurse deaths between 2001 and 2003) in a country with comparatively low HIV prevalence rates.
A study of nurses and teachers in Ghana (Clarke, 2003) showed cervical spondylosis as the second most common cause of morbidity among nurses, with nurses being 21.5 times more likely to develop lower back pain than teachers and 1.4 times more high back pain [11]. Despite the lack of benchmarks on the expected levels of morbidity and mortality, the rising trends in death and disability over the years depicts a serious problem.
HIV/AIDS will be discussed later in this review, but it is mentioned here as a major contributor to "direct wastage" due to high death rates being reported among health workers. A 1999 study found that among workers leaving public health services in Malawi, 25% of clinical officers and 51% of nurses leaving had died, compared to deaths constituting 1.1% deaths among all staff leaving Ghana's Ministry of Health. Recent Government of Malawi/United Nations Development Programme work has been quoted in Aitken and Kemp (2003) that shows deaths as the main cause of losses from the Malawi health workforce [12].
Inefficient personnel administration
Health workforce management in many African countries is part of civil service administration. For example, it was found that in Lesotho, recruitment delays meant new health workers at times spent a year before being appointed and sometimes an entire batch of new nurses is lost to the public service due to delays in processing their appointment [3]. Moreover, structural adjustment policies aimed at reducing public sector expenditures have led to retrenchments and a recruitment freeze in countries such as Cameroon and Uganda, even in the face of low availability of health workers and poor coverage of health services [13].
Ghana and Zambia have tried to address management deficiencies by "de-linking" health services from the Civil Service and creating new autonomous agencies (Ghana Health Service, Zambia Central Board of Health). In Ghana, the de-linked agencies (two tertiary hospitals and a Ghana Health Service) have been established and are gradually extending their autonomy. In Zambia anecdotal evidence suggests this experiment has not worked out well for several reasons related to transfer of workers' benefits and pensions to the new agency. The literature is not available to indicate whether implementing these structures has improved health workforce management or not, and given the available migration data, they have not improved retention and motivation of staff.
Indirect wastage
The concept of indirect wastage as contrasted to direct wastage implies losses that arise from inefficient productivity or use of health workers. Such waste of human resources is the result of the inefficient or poor use of staff already employed and providing services. Other forms of indirect waste include the inappropriate use of skills, and "ghosts" that plague payrolls while restricting room for new employment.
Almost all the issues discussed in this section border on effective management systems for human resources, but the contentions raised below elaborate on some aspects of indirect wastage of staff derived from examples in African countries. For example, absenteeism rates in similar groups of health workers (within the same country or in different countries) may vary significantly, indicating a wastage problem in one group. On the other hand, using available professionals for work not related to their skills represents another form of wastage, especially if shortages of those skills exist in the country.
Wastage as unemployment of available staff
In sub-Saharan Africa, even with its well-acknowledged shortages of health workers, unemployment occurs. Ngufor (1999) in Cameroon suggested that structural adjustment policies and related fiscal limits on governments have meant that new health graduates are not employed even when the demand exists; instead, retrenchments from the public sector continued to be encouraged [13]. Personal communications from a Deputy Commissioner of Health for Human Resources Development in Uganda also confirmed this problem.
A second factor is the blockage of health worker positions by "ghost workers" – persons who fill the payrolls but do not actually exist at workplaces. A recent report on case studies of African countries commissioned by the High Level Forum on the Millennium Development Goals found, for example, that Kenya had some 5000 ghost workers on its payroll [14].
Wastage as ineffective staff use
Underuse often occurs when staffing norms and established posts in the public service do not relate to actual workload needs but are standardized by facility type. Underemployment and underuse may result in facilities with widely varying patient loads having the same staff strength, with redundant staff in some areas and overworked ones in others.
Along with underuse is some level of misuse. Delegates from Malawi attending a migration conference in South Africa suggested that trained midwives may be avoiding postings into labour and delivery wards for fear of possible risk of exposure to HIV-infected blood. Thus less-qualified staff members are left to offer the services instead [15]. The Ghana Health Service, for example, employs 5.3% of all its doctors in mainly managerial functions at its administrative headquarters, while the deprived Upper-West Region, with a population of about 600 000, has only 1.5%, or 10 doctors [16]. Thus the use of health professionals in administrative roles or in inappropriate duties represents a form of wastage of the available resources.
Wastage resulting from poor skills/cadre mix
Ghana, Kenya, Lesotho, Malawi and Zambia have banned enrolled nurse training, even when significantly increased migration of registered nurses has severely reduced the nursing workforce. Enrolled nurses usually received two years' or less formal training and entered nursing training with lower qualifications, while registered nurses normally completed 12 years' basic education and received 3 or 4 years' professional training. Enrolled nurses usually served as auxiliaries to registered nurses. The result is a workforce that is highly internationally mobile and costs more to remunerate, while being made to carry out some tasks that enrolled nurses could readily do. The use of other mid-level health workers has often been limited by restrictions in scope of practice and resistance from the more established professional groups.
Again, despite the high need for rural health services and the high migration rate among its doctors, the health sector in Ghana produces six times more doctors annually than medical assistants, who are better retained and are more likely to be found in rural areas [17].
Wastage arising from low health worker performance and outputs
The volume and quality of work expected from otherwise competent staff are not always forthcoming, and many reports exist of absenteeism and low productivity among health workers. The main referral hospital in Ghana (Korle-Bu Teaching Hospital) in its 2002 annual report, recorded 1334 days of sick leave in 2002 by 556 nurses from a total nurse workforce of 809. Thus 70% of the nursing workforce reported ill at some time during the year, an average of 2.4 days off per nurse [18]. Staff time is also lost writing reports and carrying out basic administrative tasks. Absenteeism from HIV/AIDS is discussed separately.
The large amount of in-service training carried out by various programmes is recognized as one of the sources of wastage through sanctioned absences of staff. Though training provides skills and may arguably enhance productivity, it has sometimes provided an inverse incentive in poor countries due to generous allowances received as participants. Key staff members spend a lot of time in training courses organized by various agencies and programmes.
We had earlier under direct wastage recognized the role that the migration of trainers and specialists plays in mitigating the effectiveness of the remaining workforce's productivity. Achieving viable productivity requires a good mix of professions, with adequate numbers providing good supervision.
Wastage from misadministration of human resources
The poor management of human resources for health found in many sub-Saharan African countries is likely to contribute to widespread misdeployment and maldistribution. It contributes directly to wastage in specific ways. "Ghost" workers are a problem in many African countries, as health workforce administrators lack good staff databases and payroll systems are poorly managed. Poorly regulated or unregulated dual practice carried out by public sector doctors and nurses working in the private sector may lead to neglect of their government duties. Poor support and supervision of health workers was cited as a problem in many African countries at the Commonwealth Workshop on Developing Strategies for Attracting and Retaining Health Workers in January 2003.
Distribution problems in Ghana, for example, have resulted in the country's three most deprived regions – all with serious maternal mortality problems – having only a single gynaecologist and two surgeons serving one third of the country's land area and one sixth of its population, while 35% of all health staff are found in just two teaching hospitals [19]. Distribution problems are common, with rural and periurban slum communities probably the most deprived of trained professionals.
HIV/AIDS and HRH – a special case of wastage
The impact of HIV/AIDS on the workforce, as alluded to previously, creates both direct and indirect forms of wastage. Its complex and self-reinforcing negative impact on the health workforce merits specific mention as a major emerging source of HRH wastage.
A study by Buve et al. in Zambia showed that mortality rate among female nurses in two hospitals rose from between 2 per 1000 in 1980–1985 to 26.7 per 1000 in 1989–1991 [20]. World Bank projections quoted by Kinoti (2003) are that a country with 15% adult seroprevalence rate for HIV can expect to lose between 1.6 and 3.3% of its health care providers from AIDS annually, a direct wastage [10].
However, indirect wastage from HIV/AIDS can have as bad an effect as the direct wastage noted above. Kinoti and Tawfik (2003) estimate that absenteeism can consume up to 50% of staff time in the final year of life for health workers with AIDS. Calculations from Botswana showed that if the average infected health worker lost 60 working days in his or her final year of life, this would translate into a loss by the public health sector of 23 000 workdays in 2003 alone [21]. This excludes absenteeism arising from workers needing to attend numerous funerals of relatives and co-workers and other forms of indirect wastage.
Few countries in sub-Saharan Africa appear to have instituted programmes to cater for the counseling, support and antiretroviral treatment needs of health workers. A recent press release reporting collaboration between International Council of Nurses, Zambian Nursing Association and the pharmaceutical firm Boehringer Ingelheim to supply nevirapine to health workers is one of the new initiatives that need to be expanded quickly [22].
Managing the impact of HIV/AIDS on the health workforce in high-prevalence countries must necessarily be an important aspect of reducing both direct and indirect wastage and improving productivity from health professionals. The huge need that will be generated by the global initiative to treat 3 million persons with antiretroviral drugs by 2005 may further encumber the existing workforce away from routine duties that are not tied to a specific project.
Reducing wastage and improving staff retention
Table 2 proposes indicators for health worker wastage as a framework that country health workforce managers may use in monitoring the extent of various forms of wastage. Indicators for direct wastage depend on having a fairly robust human resources information system or the ability to carry out surveys from time to time to determine trends.
Table 2 Wastage monitoring framework
DIRECT WASTAGE
Factor Examples of contribution to wastage Possible indicators
Movement from health to non-health sector Probably small: 2 – 20 staff per year (Ghana. Mozambique, Namibia) % of job leavers exiting health work completely (exit interviews)
Emigration to health sector outside country 10% of Mauritian nurses, 61% of Ghanaian doctors Certificate verification rates
Routine leaving data, e.g. resignations
Deaths, injuries and premature removal from the workforce High significance of HIV/AIDS; Ghana 1.1% deaths compared with Malawi (<55%) of leavers Mortality rates as % of workforce leavers, or Mortality rate in workforce
Inappropriate Administrative systems and policies Affects other losses. Delays lose work input and may increase likelihood of emigration. Average recruitment duration
Staff recruitment rate versus vacancies
INDIRECT WASTAGE
Wastage as unemployment Not well documented in Africa. Estimates of "ghost workers"? Unemployed health workers as % of total workforce (for each category)
Wastage as underemployment Data is not routinely collated but staff/workload indicators may help. Staff workload Indicators, e.g. outpatient and inpatient staff per cadre
Wastage as a misuse Significant in countries with senior medics and nurses as managers. % staff: technical or professional in full-time managerial/administrative function
Wastage as inappropriate categories 4–6 categories to deliver package of services in Ghana. Workforce composition of skilled and semi-skilled staff
Absenteeism, low outputs 2.3 days' sick leave per staff member versus 1.65 days off for all staff (Ghana) Number of days off per staff member, per annum.
Misdeployment and maldistribution Distribution differential: Doctors (Ghana): best 1:16201, worst 1:66071 Doctor/nurse population ratios in different parts of country.
Wastage from misadministration of HRH Difficult to assess quantitatively: e.g. 100% of new Lesotho nurses not recruited in 1998 Recruitment and retention rates of new graduates of health training schools.
In our study, numerator difficulties have sometimes made the use of data for indicators difficult. While we could determine losses from the workforce through civil service statistics, the numbers of workers actually in the workforce, for example, was more difficult to determine with accuracy. However, the number of deaths as a proportion of the total number of people leaving public service is clearly rising; this paper suggests this could be a fairly simple system of monitoring such changes. In some countries, the reasons for leaving the workforce are not recorded in much detail; using exit interviews or forms is recommended as a way to collect data.
"Ghost" workers are a problem in some SSA countries, where the names of nonexistent workers are maintained on health payrolls by dishonest managers. It was difficult to think of routine ways to monitor "ghost" workers, apart from conducting snap censuses at workplaces, by means of the payroll. These snap censuses will probably not work as tools for regular routine implementation.
Indicators for indirect forms of wastage are much less categorical and more complex than those for direct wastage. For example, given the difficulties with data on employed health workers it might be difficult to determine how many are unemployed. This information could be collected from census data, but censuses take place at long intervals. Skill mix also represents a challenge, as the standards vary widely between countries and not many standardized benchmarks exist. Again, this may best be served by surveys showing trends and changes rather than measurements against a particular standard.
Workload standards are important to determine underemployment and appropriate distribution and deployment of health workers. Indicators can then be prepared to match staffing levels with workload. A difficulty here is that workload may vary according to seasons or with other factors, and thus workload data must be observed and analysed for a period before major changes are made. For example, attendance at health facilities may be very high during harvest periods when migrant labour comes to assist with harvest and accidents occur frequently. It may also vary over time with changes in population and economic activity in the area.
Using data to manage wastage is useful only if management systems are coherent and countries attempt to put in place strategies to cope with wastage in a comprehensive way. A number of coping strategies have been tried and many more proposed. Strategies implemented have included improving incentives and motivation of health workers through various mechanisms. In Ghana, incentives include new extra duty allowances, vehicle loans, cash incentives for rural based health workers, and local specialist training opportunities. On the other hand, Eritrea exacts a 2% income tax on its citizens living abroad; such remittances are at the level of 85.8% of development aid received. Other countries, such as Nigeria, receive remittances from their émigré community as significant sources of foreign exchange, far outstripping official development aid [23].
Using clinical officers and medical assistants to deputize for doctors may be disputed, but these categories are less internationally mobile and do mitigate shortages caused by emigrating doctors, provide comparable quality of service and are more likely to serve and remain in rural areas [24]. Some countries have recruited doctors from Cuba and some others recruit significant numbers of health workers from other African countries.
This paper has not dealt with the movements between the public and private health sectors within countries, as these are deemed to be part of the health system and are not included in wastage. Concerns do exist, however, about the possible neglect of rural areas in siting of private health facilities and hence the pull of health workers to urban areas away from other geographical zones with more needs. Moreover, dual practice by public sector health workers in the private sector may well induce wastage, as neglect of their public sector duties may ensue as a result. Private sector health professionals were most significant in Kenya, Nigeria and South Africa, among other countries.
Conclusion
Caught in a vicious cycle, the poor African economies are unable to fund systems to manage and control wastage adequately, even as new international investment in the health sector has increased demands on staff while restricting investment in incentives. The HIV/AIDS epidemic, combined with the economic crises, threatens even the few coping mechanisms that are being attempted [25].
What can be done to alleviate the problem?
The capacity of countries' HRH departments must be strengthened, and development partners and governments must invest significant portions of health budgets in building capacity, not only through training, tools and technology, but with incentives to retain staff. Currently, most health sector human resources departments are managed as part of the general civil service and have little influence on policy development; they may lack specialists in health workforce planning and management.
Awareness of the problems of wastage can be increased by integrating wastage indicators into human resources information systems used by country HRH managers. Some indicators have been suggested in Table 2; structuring data systems to collect and analyse these indicators may provide evidence to compare productivity of different health service delivery units and also to advocate changes in how human resources for health are planned for and managed.
Motivation and morale are key factors in wastage. How can this change?
Governance and leadership in health must now be expressed as tangible actions that result in senior managers and policy-makers valuing and respecting health workers. New career and incentives systems must be developed, along with better social and technical support for health workers. Real or perceived occupational risk from the health workplace appears to contribute significantly to low morale and consequent wastage. The public sector must establish occupational health services that assure prevention and treatment for workplace incidents. An essential component of this service should include voluntary counseling and testing, as health workers also need education on HIV/AIDS as well as the antiretroviral treatment policies that are becoming more available in other industries.
Is moral leadership needed?
There is almost a sense of helplessness in dealing with the HRH crisis in Africa. Because health managers anticipate that development partners will avoid support for HRH incentive issues, they now rarely include them in their proposals to the global funding agencies. Emigration is another area where international action has mainly been in the form of voluntary codes of conduct that have had little effect. Economic, labour market and human rights arguments are made by the developed countries as the basis for their reluctance to assist developing countries to manage emigration more effectively. However, the HRH crisis in Africa requires that countries on both sides also create a moral discourse to take actions that will improve the health of Africans.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
Delanyo Dovlo was the sole author.
Acknowledgements
This paper was encouraged and supported by the Rockefeller/WHO/World Bank Joint Learning Initiative on Human Resources for Health, Working Group on HRH Demand.
==== Refs
Dovlo D Nyonator F Migration of graduates of the University of Ghana Medical School: a preliminary rapid appraisal Human Resources for Health Development Journal 1999 3 45
Davuca JA Staff loss in Mozambique Commonwealth Workshop on Development Strategies for Attracting and Retaining Health Workers, 20–22 January 2003 Johannesburg 2003
Dovlo D Issues affecting the mobility and retention of health workers/professionals in Commonwealth African states 1999 London: Commonwealth Secretariat
Dare O Brain drain and human resource development in Nigeria accessed 4 January 2005
Huddart J Picazo O The health sector human resources crisis in Africa: an issues paper 2003 Washington DC:USAID Bureau for Africa, Office of Sustainable Development
Martineau T Decker K Briefing note on international migration of health professionals: leveling the playing field for developing country health systems 2002 Liverpool: Liverpool School of Tropical Medicine
Seebah M Migration of nurses from the public sector: causes/reasons for leaving and proposals for their retention – the Mauritian context Presentation at HRD Seminar: Presentation of research projects 2002 Cape Town: University of Western Cape
Dovlo D Retention and deployment of health workers & professionals in Africa Consultative Meeting on Collaboration Between Health Professions and Governments in Policy Formulation and Implementation of Health Sector Reform 2002 Addis Ababa: World Bank/WHO
Meeuss W "Pull" factors in international migration of health professionals. An analysis of developed countries' policies influencing migration of health professionals Master's thesis 2003 University of Western Cape, School of Public Health
Tawfik L Kinoti SN The impact Of HIV/AIDS on health systems and the health workforce in sub-Saharan Africa 2003 Washington DC: Support for Analysis and Research in Africa (SARA) Project, USAID, Bureau For Africa, Office Of Sustainable Development
Clarke E Contribution of occupational health and safety factors to the brain drain in the health sector African Newsletter on Occupational Health and Safety 2003 13 2 Aug 2003
Aitken J-M Kemp J HIV/AIDS, equity and health sector personnel in southern Africa Harare: Regional Network for Equity in Health in Southern Africa (EQUINET) in cooperation with Oxfam GB, Equinet Discussion Paper Number 12 2003 accessed 3rd March 2005
Ngufor GF Public service reforms and their impact on health sector personnel in Cameroon ILO/WHO Public service reforms and their impact on health sector personnel: Case studies on Cameroon 1999 Geneva: International Labour Organization/World Health Organization
WHO HLF Secretariat Health workforce challenges: lessons from country experiences Paper prepared for the High Level Forum on Health MDGs Abuja 2004
Bandazi SN Deputy Director of Nursing, Ministry of Health & Population, Malawi Personal communication at Commonwealth Workshop on Developing Strategies for Attracting and Retaining Health Workers, 20–22 January 2003 Johannesburg 2003
Ministry of Health, Ghana Program of Work 2002 2003 Report of External Review Team. Accra
Dovlo D Using mid-level cadres as substitutes for internationally mobile health professionals in Africa. A desk review Human Resources for Health 2004 2 7 accessed 18 June 2004 15207010 10.1186/1478-4491-2-7
Ministry of Health, Ghana Korle Bu Teaching Hospital Annual Report, 2002 Accra 2003
Ministry of Health, Ghana Hospital Strategy – A Document for Hospital Reforms in the Health Sector Accra 2003
Buve A Foaster SD Mbwili C Mungo E Tollenare N Zeko M Mortality among female nurses in the face of the AIDS epidemic: a pilot study in Zambia AIDS 1994 8 396 8031526
Abt Associates South Africa Inc The Impact of HIV/AIDS on the Health Sector in Botswana 2000 Gaberone: Ministry of Finance and Development Planning
International Council of Nurses, Zambian Nurses Association, Boehringer Ingelheim Novel AIDS Treatment Programme for Health Care Workers in Zambia accessed 4 January 2003
African Union The Role of the African Diaspora in the Development of Their Countries of Origin First Ordinary Session Port Louis, Mauritius 10–15 April 2003 LSC/9 (XXVI) 2003 Port Louis: Labour and Social Affairs Commission of Mauritius
Vaz F Bergstrom S da Luz Vaz M Langa J Bugalho A Training medical assistants for surgery Bulletin of the World Health Organization 1999 77 688 691 10516791
Sanders D Dovlo D Meeuss W Lehmann U Beaglehole R Public health in Africa Global Public Health: A New Era 2003 Oxford: Oxford University Press 135
|
16092964
|
PMC1198245
|
CC BY
|
2021-01-04 16:37:43
|
no
|
Hum Resour Health. 2005 Aug 10; 3:6
|
utf-8
|
Hum Resour Health
| 2,005 |
10.1186/1478-4491-3-6
|
oa_comm
|
==== Front
Int Semin Surg OncolInternational seminars in surgical oncology : ISSO1477-7800BioMed Central London 1477-7800-2-151611531410.1186/1477-7800-2-15ResearchIs preoperative core biopsy accurate in determining the hormone receptor status in women with invasive breast cancer? Al Sarakbi W [email protected] M [email protected] V [email protected] K [email protected] St George's and The Princess Grace Hospitals, London, UK2005 22 8 2005 2 15 15 14 7 2005 22 8 2005 Copyright © 2005 Al Sarakbi et al; licensee BioMed Central Ltd.2005Al Sarakbi et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The objective of this study was to determine the concordance rate between core needle biopsy (CNB) and surgical excision of invasive breast cancer regarding the oestrogen receptor (ER) and Progesterone receptor (PgR) status as determined by Immunohistochemistry (IHC).
Methods
Hormone receptor status was established using IHC (using quickscore system 0–8) on preoperative CNB and subsequent surgical excision in 93 patients with invasive breast cancer. Results were compared taking into account tumour's size, grade, and patient's age.
Results
The ER concordance rate between CNB and surgical excisions was 95%. The PgR concordance rate was 89%. This shows that CNB has a sensitivity of 97% for ER and 95% for PgR.
There is a positive correlation of ER and PgR between CNB and surgical excision (p < 0.000001). There was no significant difference in the number of core biopsies between concordant and discordant cases.
Conclusion
Preoperative core biopsy is highly sensitive for the IHC detection of ER and PgR in invasive breast cancer. The concordance rate is higher for ER than PgR, which could be due to the fact that ER is more homogeneously distributed.
core needle biopsyoestrogen receptorsprogesterone receptors
==== Body
Introduction
The core needle biopsy (CNB) is technique increasingly used for the preoperative assessment of breast lesions [1]. Image guidance increases accuracy and reduces the number of false negative cases [2]. The presence of malignancy and tumour's type and grade are routinely reported.
There is an increasing need to provide prognostic data on CNB in order to improve treatment outcome [3,4].
Hormone receptor status, and especially ER, provides valuable prognostic information and predicts the response to adjuvant and neo-adjuvant systemic treatment [5].
Traditionally, the hormone receptor status was determined by enzyme immunoassay of ER and PgR proteins. However, this has been replaced gradually by immunohistochemistry (IHC), which has shown an equal reliability [6].
Previous studies that examined the reliability of preoperative CNB using enzyme immunoassay showed conflicting results [7-9].
This retrospective study examines the correlation between CNB and surgical excision in regards to the ER ad PgR status of invasive breast cancer using IHC.
Patients and Methods
In this retrospective study we looked at consecutive 95 cases of invasive breast carcinoma in 93 patients. All patients underwent CNB at their clinic appointment and proceeded for breast surgery subsequently 2–3 weeks later. Preoperative CNBs and surgical excision specimens were analysed for ER and PgR status using IHC (DAKO mab) after antigen retrieval at high temperature. All specimens were analysed using semi quantitative IHC "quick score" system (0 – 8) by the same breast pathologist. With this method, the intensity of the immunohistochemical reaction as viewed under the light microscope was recorded 0–4 (0 indicated no staining of any nuclei even at high magnification). The proportion of cells staining positively at any intensity was scored as 0 (no cell staining), 1 (1–25% cells stained), 2 (26–50% cells stained), 3 (51–75% cells stained) or 4 (when >75% cells stained). The proportion and intensity scores were added together to obtain a total score ranging from 0 to 8.
We also examined other parameters including: tumour size, grade, and patient's age.
ER/PgR status was considered positive if quick score was 2 – 8. Results were re-assessed when quick score was raised to 4 – 8 to label ER/PgR status as positive. The number of biopsies taken on each occasion was also recorded.
Results
The mean and median age of this study group was 59.2 and 59 respectively (range 32 – 92 years). The mean tumour size was 20 mm (3.2 – 110 mm), and the median grade was 2 (1 – 3). The median number of CNBs taken at preoperative assessment was 2.6 (1 – 9). The mean IHC score for ER was 6.7 (range 0 – 8).
All patients had either mastectomy or lumpectomy as their definitive surgical treatment.
Firstly, ER/PgR status was considered positive if quick score was 2 – 8. The concordance for tumour grade was 65%. The concordance rate for ER was 95% between CNB and surgical excision. There were 2 false negative and 3 false positive cases. PgR concordance rate was 89% with 4 false negative and 6 false positive cases. According to the above results the sensitivity of CNB for ER and PgR was 97% and 95% respectively. Furthermore, to re-examine our findings, we analysed the results considering ER/PgR to be positive if quick score was 4 – 8. We found a concordance rate for ER to be 93% (3 false negative and 5 false positive cases). The concordance rate for PgR was 92% (3 false negative and 6 false positive cases) (figure 1). This gives CNB sensitivity of 98% for positive ER status and 96% for positive PgR status (table 1).
Figure 1 Graphic Results, 1: ER in CNB, 2: ER in surgical excision.
Table 1 Sensitivity and specificity of CNB for ER and PgR Status.
Parameter ER status N = 95 PgR status N = 93
Positive CNB 89 85
Negative CNB 6 8
Positive SE 86 82
Negative SE 9 11
Sensitivity of CNB 97.6% 96.3%
Specificity of CNB 96.6% 96.4%
Concordance rate 92.6% 92.4%
CNB: Core Needle Biopsy
SE: Surgical Excision
Positive: Quickscore 4 – 8
Negative: Quickscore 0 – 3
ER in CNB positively correlated with ER in surgical excision (r: 0.61, p < 0.000001). PgR in CNB also positively correlated with PgR in surgical excision (r: 0.66, p < 0.000001)
There was no significant difference in the number of CNBs between concordant and discordant cases.
Discussion
Core needle biopsies offer a reliable and accurate assessment of hormone receptor status.
Previous studies on a smaller case sample have suggested similar findings regarding ER [10-12]. However, results on PgR were less consistent.
The concordance rate in our study for ER was higher than for PgR. Homogenous distribution of ER through out the tumour is a possible explanation. Heterogeneity of the ER in tumour cell populations may have important implications for analytic cell selection and for prognosis in ER-positive carcinomas. Previous studies have shown homogenous geographic distribution of ER in the tumour cell population [13,14].
These results indicate that the hormone receptor status as determined by CNB can be reliably used to guide neo-adjuvant and adjuvant systemic therapy in patients with invasive breast cancer.
In summary, preoperative CNB is highly sensitive for the IHC detection of ER and PgR in invasive breast cancer.
==== Refs
Sharifi S Peterson MK Baum JK Assessment of pathologic prognostic factors in breast core needle biopsies Mod Pathol 1999 12 941 945 10530557
Shah VI Raju U Chitale D False-negative core needle biopsies of the breast: an analysis of clinical, radiologic, and pathologic findings in 27 concecutive cases of missed breast cancer Cancer 2003 97 1824 31 15 12673707 10.1002/cncr.11278
Harris GC Denley HE Pinder SE Correlation of histologic prognostic factors in core biopsies and therapeutic excisions of invasive breast carcinoma Am J Surg Pathol 2003 27 11 5 12502923 10.1097/00000478-200301000-00002
King TA Cederbom GJ Champaign JL A core breast biopsy diagnosis of invasive carcinoma allows for definitive surgical treatment planning Am J Surg 1998 176 497 501 9926778 10.1016/S0002-9610(98)00250-5
Colditz GA Rosner BA Chen WY Risk factors for breast cancer according to estrogen and progesterone receptor status J Natl Cancer Inst 2004 96 218 28 4 14759989
Ferrero-Pous M Trassard M Le Doussal V Comparison of enzyme immunoassay and immunohistochemical measurements of estrogen and progesterone receptors in breast cancer patients Appl Immunohistochem Mol Morphol 2001 9 267 75 11556756 10.1097/00022744-200109000-00012
Bridges KG Keshgegian AA Kumar H Influence of surgical technique on estrogen and progesterone receptor determination in breast cancer Cancer 1983 51 2317 20 6850511
Sharoni T Feldman B Inhar I Estrogen and progesterone receptor levels are lower in specimens taken from previously biopsied breast tumour tissue J Surg Oncol 35 197 200 3599994
Day T Yeoman RR Nelson G Influence of mastectomy technique on sex steroid receptor analysis Am J Surg 1988 156 446 9 3202254
Connor CS Tawfik OW Joyce AJ comparison of prognostic tumor markers obtained on image-guided breast biopsies and final surgical specimens Am J Surg 2002 184 322 4 12383893 10.1016/S0002-9610(02)00953-4
Jacobs TW Siziopikou KP Prioleau J Do prognostic marker studies on core needle biopsy specimens of breast carcinoma accurately reflect the marker status of the tumor? Mod Pathol 1998 11 259 64 9521472
Railo M Nordling S Krogerus L Preoperative assessment of proliferative activity and hormonal receptor status in carcinoma of the breast: a comparison of needle aspiration and needle-core biopsies to the surgical specimen Diagn Cytopathol 1996 15 205 10 8955602 10.1002/(SICI)1097-0339(199609)15:3<205::AID-DC6>3.0.CO;2-F
Layfield LJ Saria E Mooney EE Tissue heterogeneity of immunohistochemically detected estrogen receptor. Implications for image analysis quantification Am J Clin Pathol 1998 110 758 64 9844588
Charpin C Martin PM De Victor B Multiparametric study (SAMBA 200) of estrogen receptor immunocytochemical assay in 400 human breast carcinomas: analysis of estrogen receptor distribution heterogeneity in tissues and correlations with dextran coated charcoal assays and morphological data Cancer Res 1988 48 1578 86 15 2449956
|
16115314
|
PMC1198246
|
CC BY
|
2021-01-04 16:38:37
|
no
|
Int Semin Surg Oncol. 2005 Aug 22; 2:15
|
utf-8
|
Int Semin Surg Oncol
| 2,005 |
10.1186/1477-7800-2-15
|
oa_comm
|
==== Front
J Exp Clin Assist ReprodJournal of Experimental & Clinical Assisted Reproduction1743-1050BioMed Central London 1743-1050-2-91598752810.1186/1743-1050-2-9ResearchProspective assessment of Y-chromosome microdeletions and reproductive outcomes among infertile couples of Japanese and African origin Kihaile Paul E [email protected] Atsushi [email protected] Yoshihiro [email protected] Oita Medical University, Oita City, 879-5593 Japan2 St. Luke IVF Center, Oita City, 870-0497 Japan3 Muhimbili University College of Health Science, Box 65001, Dar Es Salaam, Tanzania2005 29 6 2005 2 9 9 15 9 2004 29 6 2005 Copyright © 2005 Kihaile et al; licensee BioMed Central Ltd.2005Kihaile et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
To compare the frequency of Y-chromosome microdeletions in Japanese and African azoospermic and oligozoospermic men and describe embryo characteristics and reproductive outcome following in vitro fertilization (IVF) with intracytoplasmic sperm injection (ICSI).
Methods
Our study was performed prospectively at two centers, a private IVF clinic and a university hospital. Japanese and African (Tanzanian) men with nonobstructive azoospermia (NOA) and oligozoospermia (concentration < 5 × 106 /ml) were evaluated for Y-chromosome microdeletions (n = 162). Of the 47 men with NOA, 26 were Japanese and 21 were Africans. Of the 115 men with oligozoospermia, 87 were Japanese and 28 were Africans. Reproductive outcomes of patients with Y-chromosome microdeletions were then compared with those of 19 IVF+ICSI cycles performed on couples with Y-chromosome intact males/tubal factor infertility which served as a control group.
Results
Seven azoospermic and oligozoospermic patients had Y-chromosome deletions; the total number of deletions in the AZFc region was five. There was only one deletion in the AZFa region and one complete deletion involving all three regions (AZFa, b, and c) within AZF. In our study population, microdeletion frequency among Japanese men was 6.2% (95% CI, 4.25% – 14.45%), whereas no deletions were identified in the African group (95% CI, 0.0% – 7.27%). The difference between the two groups was not statistically significant, however. Embryos derived from ICSI utilizing sperm with Y-chromosome microdeletion showed reduced rates of fertilization, blastocyst development, implantation, and pregnancy compared to the Y-chromosome intact group, although these observed differences were not statistically significant.
Conclusion
The observed frequency of Y-chromosome microdeletion was 6.2% among Japanese azoospermic and oligozoospermic males; no microdeletions were identified among our African study patients. In this population of couples undergoing IVF+ICSI, there was no statistically significant difference in embryo characteristics or pregnancy outcome between patients with Y-chromosome microdeletion and those with an intact Y-chromosome.
==== Body
Introduction
Approximately15% of the couples worldwide cannot conceive after one year of regular sexual intercourse, with the male factor accounting for ~40% of all infertility; however, in up to 30% of cases the etiology is unexplained [1,2]. Among these unknown cases of infertility, a genetic etiology for male infertility was long suspected. This was due to cytogenic evidence showing that 0.2% of azoospermic men, who were otherwise phenotypically normal, exhibited Y-chromosome microdeletions [3]. This evidence was supported by karyotyping which revealed autosomal translocations in 1.3% of infertile couples [4-7]. Researchers realized that many cases of male infertility might be genetic because of the failure of most clinical treatments to correct abnormal sperm parameters [8,9]. Observations of sperm counts in various species, including studies of naturally occurring deletions in drosophila via molecular analysis, also led to an intense search for human spermatogenesis gene(s) which might be deficient in some infertile males [10-13]. Recent advancements in molecular methodology have permitted careful mapping of Y-chromosome microdeletions in men with azoospermia and oligozoospermia; in Western populations this frequency varies between 1–35%, depending on inclusion criteria [14]. For Japanese males, a Y-chromosome microdeletion frequency range of 7.6–17% has been reported [15-22]. Such studies have identified three "azoospermic factor" regions (AZF) where deletions occur on the Y-chromosome long arm: AZFa, AZFb, and AZFc. AZFc was shown to contain the most frequently deleted gene cluster, known as the DAZ gene [23]. Several studies found AZFc deletions to be associated with successful retrieval of sperm during testicular sperm extraction (TESE), whereas deletions in AZFa and AZFb were not [23-25]. Histologically, these deletions are associated with various spermatogenetic alterations including Sertoli cell-only syndrome (SCOS), maturation arrest, and hypospermatogenesis.
Recently, several investigators have shown that embryo characteristics following intracytoplasmic sperm injection (ICSI) using sperm obtained from men with Y-chromosome microdeletions were not adversely affected by the deletion [26-31]. The central concern was that vertical transmission of the microdeletion via ICSI might be passed from father to son [32] or by natural (unassisted) conception [30,33]. Since very few studies concerning Y-chromosome microdeletion have been undertaken in Japanese males (and none in African males), we aimed to investigate the frequency of Y-chromosome microdeletion as well as selected embryo features and reproductive outcome in Japanese and African azoospermic and oligozoospermic men who underwent IVF+ICSI.
Materials and Methods
Between January 1998 and January 2003, male volunteers (n = 162) presenting for infertility evaluation and treatment at two centers were evaluated for Y-chromosome microdeletions via peripheral venipuncture. The study population consisted exclusively of Japanese (n = 113) and African (n = 49) males who, with their partners, sought infertility treatment either at St. Luke IVF Center (Japan) or Muhimbili National Hospital (Tanzania). Written informed consent was obtained from all study patients and the investigation was approved by the hospital's ethical committee. Subjects were partitioned into two groups based on sperm concentration: 1) nonobstructive azoospermia (NOA), and 2) oligozoospermia, defined as sperm concentration < 5 × 106 ml.
The first group consisted of 47 males (26 Japanese and 21 Africans); the second group consisted of 115 males (87 Japanese and 28 Africans). None of the study patients were diagnosed with obstructive azoospermia. Six couples from the Muhimbili center were excluded from the study due to active STD. The GFX Genomic Blood DNA Purification Kit (Amersham Biosciences, Buckinghamshire, United Kingdom) was used to extract DNA from peripheral venipuncture samples as previously described [34]. Y-chromosome microdeletions were detected using a polymerase chain reaction (PCR) amplification with a specific sequence tag site (STS) using 24 sets of primers which allowed evaluation of the following sites: sY14, sY18, sY78, sY81, sY83, sY85, sY84, sY90, sY100, sY131, sY134, sY139, sY145, sY143, sY153, sY147, sY156, sY149, sY254, sY157, sY202, sY243, sY158, and sY159. Deletion of the loci was confirmed if the product of the expected size was not obtained after three single STS PCR experiments.
Four patients with NOA and Y-chromosome microdeletion underwent TESE. The samples were microscopically examined to search for sperm, which was cryopreserved as described previously [24]. TESE was successful in 2 of 4 azoospermic cases; after cryopreservation these couples subsequently underwent two ICSI cycles. The three Japanese oligozoospermic patients with Y-chromosome microdeletion produced fresh ejaculated sperm which was subsequently used for 6 ICSI cycles. Sperm morphology in each laboratory was examined by 2 observers [35]; classification of normal sperm morphology at our centers is: <4% = severe teratozoospermia, 4–14% = moderate teratozoospermia and >14% = normal. No fathers or brothers of our male patients were available for testing. All Japanese couples who had a multifactoral diagnosis of oligozoospermia and tubal factor infertility were also assessed for Y-chromosome microdeletions and those found to have an intact Y-chromosome were assigned to the control group. Fourteen Japanese couples were initially considered for this group, but three were excluded because of endometriosis (n = 2) and leiomyoma (n = 1). These remaining couples (n = 11) constituted the control group and they underwent 19 ICSI cycles. Positive and negative controls were used for all AZF microdeletion tests. Thirteen fertile men with a sperm concentration of >20 × 106 /ml were used as positive controls, and twelve females served as negative controls.
Ovarian stimulation protocol, oocyte handling, laboratory procedures for insemination, measurement of sperm parameters, hormones, ICSI, and embryo and blastocyst grading were performed as previously described [35]. Using this protocol, only types I, II and III embryos were considered suitable for transfer and ≤ 3 embryos were transferred on day 3 after microinjection. For 6 cycles, embryos were cultured until day 5–6 and were transferred at the blastocyst stage. Clinical pregnancy was confirmed at 6 weeks via transvaginal ultrasound to establish embryonic cardiac activity.
Data were analyzed for equality of variance using the Levene's test. When p > 0.05 the variances were considered equal, a Student's t-test was performed, and p < 0.05 was considered significant. When p < 0.05, then the variances were considered unequal and a Wilcoxon signed ranked test was done, and p < 0.05 was considered significant. All computations were conducted using the SPSS statistical package (SPSS Inc., Chicago, USA).
Results
Of 162 azoospermic and oligozoospermic Japanese and African males who participated in the study, 47 were diagnosed with NOA and 115 as oligozoospermic. Of 47 NOA cases, 26 were Japanese and 21 were Africans. Of 115 oligozoospermic cases, 87 were Japanese and 28 were African. We identified seven cases with microdeletions and they were all within the azoospermia and oligozoospermia Japanese group. Five of these deletions were identified in the AZFc region, whereas only one deletion was identified in the AZFa region. There was also one complete deletion involving all three regions of AZF (AZFa, b, c). Therefore, the microdeletion frequency within the Japanese group was 6.2% (95% CI, 4.25% – 14.45%). No deletions were identified within the African group (95% CI, 0.0% – 7.27%); this difference was not statistically significant. The deletion frequency among azoospermic Japanese males was 15.4% (95% CI, 11.0% – 42.0%) whereas there were none among the African males (95%CI, 0.0% – 15.4%). Again, the difference did not reach statistical significance. The deletion frequency in oligozoospermic Japanese was 3.4% (95% CI, 1.18% – 9.6%) and there were none in oligozoospermic African males (95% CI, 0.0% – 12.0%); also with no statistical difference.
Two patients were diagnosed with SCOS and both exhibited one microdeletion in the AZFa and AZFabc regions. (Table 2). These two patients had characteristically high serum FSH levels (mean = 30.4 U/L, normal is < 10.0 U/L in our laboratory). Testicular volumes of males with Y-chromosome microdeletion and those with an intact Y-chromosome showed no statistically significant difference (average ± SD volumes were 11.14 ± 2.41 and 12.4 ± 3.40 ml, respectively).
Table 3 summarizes comparisons among various parameters in patients with and without Y-chromosome microdeletion. As expected, sperm concentration was significantly lower in the Y-chromosome microdeletion group than in the intact Y-chromosome group (p < 0.05).
Table 4 presents a comparison of embryo characteristics observed in AZFc microdeletion and Y-chromosome intact patients. Although the Y-chromosome microdeletion group showed a trend towards reduced rates of fertilization, implantation, and pregnancy when compared to the Y-chromosome intact group, this difference did not reach statistical significance. Screening for Y-chromosome microdeletions among our study patients' male offspring was not conducted in this investigation.
Discussion
This study shows the frequency of microdeletion in the AZF region of the Y-chromosome to be 6.2% among Japanese males with azoospermia and oligozoospermia. Our findings are consistent with prior reports which found a microdeletion frequency of 1% to 35%, depending on the male subfertility definition used for inclusion and on the choice of sequence tagged sites used for screening [6,14,28,36,37]. In our populaton, the majority of Y-chromosome microdeletions (~70%) occurred in the AZFc region of the AZF region and is in agreement with other investigations [27,38]. Previous studies on Y-chromosome microdeletion frequency in Japanese males [15-22] suggested a range of 7.6% – 17.0%. Separating these Japanese patients with deletions into azoospermic and oligozoospermic cases revealed frequencies of 15.4 % and 3.4%, respectively.
We excluded from analysis two study patients with Y-chromosome microdeletions at sY202 and sY243. Indeed, no known location has been ascertained so far for sY202 and sY243 is found in several locations both inside and outside DAZ. Nevertheless, the frequency observed in our population is consistent with previous data describing a 15%-20% microdeletion frequency in men with idiopathic azoospermia and a 7%-10% frequency in men with severe idiopathic oligozoospermia [15,23,38,39]. Interestingly, no microdeletions were documented among the 49 African males with azoospermia or severe oligozoospermia. Since ours is the first study of Y-chromosome microdeletion frequency to be undertaken on an African population, a comparison with earlier data sets from this group was not possible.
The absence of Y-chromosome microdeletions in our African study population may be due to limited sampling. However, this finding is in general agreement with the few studies conducted in other parts of Africa that found frequencies of male infertility secondary to oligozoospermia or azoospermia to be much lower (~20%) in Africa than elsewhere [40-48]. Unlike some other regions, the most common cause of male infertility in Africa was found to be infection. Yeboah et al. reported on 595 infertile African males and found ~70% of them to have inflammatory testicular lesion due to STD [40]. A multi-center study by Cates et al. demonstrated that >50% of African couples had secondary infertility due to STD [41], which was a rate much higher than in non-African countries (i.e., <30%).
In our study all active STD cases were excluded, although the association between male infertility and STD remains controversial. Some investigators have shown that treatment of infection directly improved the sperm quality in oligozoospermia [42,43] while others did not see any improvement in sperm quality after treatment [44,45]. It is difficult to know the precise frequency of azoospermia and oligozoospermia in Africa as certain cultural factors (e.g., polygamy) are more common than in other parts of the world [46]. Therefore, an azoospermic man may have children whose actual biological father was another man if the wife had extramarital intersourse with a fertile male [47,48]. Financial constraints did not enable testing to confirm paternity when babies were born after infertility treatments at our centers.
All 5 cases with Y-chromosome microdeletion in the AZFc region had successful outcomes following IVF+ ICSI. A comparison of reproductive outcomes between couples with AZFc microdeletion and couples with an intact Y-chromosome showed no statistically significant difference. Our results confirm previous studies showing that Y-chromosome microdeletions do not appear to adversely affect fertilization and pregnancy rates (either in azoospermic or severe oligozoospermic men) when sperm are successfully retrieved [26-33,36]. The concerns that IVF+ICSI might yield poorer results in the setting of Y-chromosome microdeletions were not seen in previous reports [26-30]. However, Van Volde et al. [29] found fertilization and embryo quality to be significantly lower in couples with Y-chromosome microdeletions compared to couples without Y-chromosome microdeletions; pregnancy and take home baby rates were not statistically different.
In conclusion, our investigation did not detect any Y-chromosome microdeletions in azoospermic or severely oligozoospermic men of African origin. This was in contrast to seven cases of Y-chromosome microdeletion identified in Japanese males. Furthermore, comparison of embryo characteristics in Japanese couples with Y-chromosome microdeletion and control couples (those with no Y-chromosome microdeletion) revealed no statistically significant difference. We regard it as premature to conclude definitively that Y-chromosome microdeletion in azoospermic and oligozoospermic African males is not as common as in other races, due to limited sampling. We hope to continue this investigations with a larger patient population to provide additional information on the overall frequency of azoospermia/oligozoospermia in African males, as well as the association of these conditions with Y-chromosome microdeletions.
Table 1 STS markers of the 7 infertile men with Y chromosome deletions.
STS Region
Markers 1.1 1.2 1.5 1.6 1.7 1.8 1.9
sY14 + + + + + + +
sY18 + + + + + + +
sY78 + + + + + + +
sY81 AZFa - - + + + + +
sY83 - - + + + + +
sY85 - - + + + + +
sY84 - - + + + + +
sY90 - - + + + + +
sY100 AZFb + - + + + + +
sY131 + - + + + + +
sY134 + - + + + + +
sY139 + - + + + + +
sY145 + - + + + + +
sY143 + + + + + +
sY153 AZFc + - - - - - -
sY147 + - - - - - -
sY156 + - - - - - -
sY149 + - - - - - -
sY254 + - - - - - -
sY157 + - - - - - -
sY202 + - - - - - -
sY243 + - - - - - -
sY158 + - + + + + +
sY159 + - + + + + +
Table 2 Findings in 7 infertile men with Y chromosome microdeletion
Patient Nos.
1.1 1.2 1.5 1.6 1.7 1.8 1.9
Left/right testis volume (ml) 9&14 12&14 14&15 13&15 9&8 13&14 8&7
sperm conc (× 106 /ml) 0 0 0 0.9 0.6 1.5 0
Deleted AZF regions a a, b, c c c c c c
FSH (U/ml) 29.5 31.3 10.3 11.6 9.1 12.7 14.3
Testicular histology SCOS SCOS MA ND ND ND MA
Sperm testicular recovery 0 0 + ND ND ND +
ND = Not done; MA = Maturation Arrest; SCOS = Sertoli cell only syndrome
Table 3 Results of the various parameters of ejaculate sperm cycles in patients with AZFc region Y-deleted and Y-Intact chromosomes.
AZFc region
Y-deleted Chromosome Y-Intact Chromosome Significance
Total No. Patients 5 11
Total No. Cycles 8 19
Female Age (mean) 30.9 ± 6.2 30.4 ± 5.7 NS
Duration of infertility in years (mean) 4.9 ± 3.1(2–10) 4.4 ± 2.7(2–6)
Peak E2 pg/dl (mean) 3613 ± 1876 3785 ± 1543 NS
Embryos Transferred (average)) 2.7(2–3) 2.8 (2–4) NS
Sperm Normal Morphology (%) 2.3 ± 0.5% 3.4 ± 0.7 NS
Sperm Concentration (× 106/mL) 1.5 ± 1.1 4.3 ± 1.2 p < 0.05
Motility (%) 35 ± 14.7 41 ± 11.4 NS
Table 4 ICSI attempts in 8 cycles with Y-chromosome microdeletions and 19 cycles without microdeletions.
Y-deleted Y-Intact
Chromosome Chromosome Significance
No cycles with ejaculated sperm 6 19 NS
No. cycles with Testicular sperm 2 0
Total No. oocytes 108 269 NS
Average No.oocytes (n) 13.5 ± 2.7 12.8 ± 3.5 NS
Fertilization rate(%) 60.1 ± 17.9% 71.6 ± 15.7% NS
Cleavage rate (%) 87.6 ± 14.5 85.1 ± 9.4 NS
Embryo grade 1 & II on day 3/cleaved embryos 51.70% 59.60% NS
≥6 cells embryos on day 3/cleaved embryos (%) 75.00% 72.9 NS
Blastocyts on day 5&6 /cleaved embryos (%) 50.00% 53.40% NS
Canceled cycle(s) 1 0 NS
Average mixed embryo & blastocyst ET 2.7 ± 0.5 2.8 ± 0.9 NS
Pregnancy 3 (37.5%) 9 (42.9) NS
Implantation rate 13.5 16.9 NS
Acknowledgements
We wish to thank Dr. Takafumi Utsunomiya, Yoko Kumasako and Keiko Hirotsuri of St. Luke hospital for their advice and technical help ; We also wish to thank Prof. Kensuke Yamamoto, Dr. Kazuo Aoki, Prof. Junichi Misumi and Akira Kono of Oita University school of Medicine for their help in study design and constant examination of our data; Drs. Ramzy E. Kisanga and Godfrey Lema of Muhimbili University of college science, Dar-es-salaam, Tanzania for their technical assistance while in Tanzania and San Francisco Edit for editing this manuscript.
==== Refs
Thonneau P Marchand S Tallec A Incidence of main causes of infertility in a resident population (1 850 000 inhabitants) of three French regions (1988–1989) Hum Reprod 1991 6 811 81 1757519
Abma JC Chandra A Mosher WD Peterson LS Fertility, Family Planning and Women's Health: New data from the 1995 national Survey of Family Growth Vital heath statistics 1997 23 1 114
Tiepolo L Zuffardi O Localization of factors controlling spermatogenesis in nonfluorescent portion of the human chromosome Y long arm Hum Genet 1976 34 119 124 1002136 10.1007/BF00278879
Van Assche E Bonduelle M Tournaye H Joris H Verheyen G Devroey P Van Steirteghem A Liebaers I Cytogenetics of infertile men Hum Reprod 1996 11 1 24 9147109
Bonduelle M Aytoz A Van Assche E Incidence of chromosomal aberrations in children born after assisted reproduction through intracytoplasmic sperm injection Hum Reprod 1998 13 781 2 9619521 10.1093/humrep/13.4.781
Bonduelle M Camus M De Vos A Staessen C Tournaye H Van Assche E Seven Years of Intracytoplasmic Sperm injection and follow up of 1987 subsequent children Hum Reprod 1999 14 243 264 10573038
Egozcue S Blanco J Vendrell JM García F Veiga A Aran B Barri PN Vidal F Egozcue J Human male infertility: chromosome anomalies, meiotic disorders, abnormal spermatozoa and recurrent abortion Hum Reprod Update 2000 6 93 105 10711834 10.1093/humupd/6.1.93
Devroey P Vandervorst M Nagy P Van Steirteghem AC Devroey P, Tarlatzis B, Van Steirteghem A Do we treat the male or his gamete? In "Current Theory and Practice of ICSI" Hum Reprod 1998 13 Oxford, UK, Oxford University Press 178 185 9663782 10.1093/humrep/13.1.178
Silber SJ Van Sterteghem AC Liu J Nagy Z Tournaye H Devroey P High fertilization and pregnacy rates after ICSI with spermatozoa obtained from testicle biopsy Hum Reprod 1999 10 148 152 7745045
Obrien SJ Wildt DE Bush M The Cheetah in genetic peril Sci Am 1986 254 84 92 3704622
Ma K Sharkey A Kirsch S Towards the molecular localisation of the AZF locus: mapping of microdeletions in azoospermic men within 14 subintervals of interval 6 of the human Y-chromosome Hum Mol Genet 1992 1 29 33 1301132
Ma K Inglis JD Sharkey A Bickmore WA Hill RE Prosser EJ A Y chromosome gene family with RNA binding protein homology; candidate for the azoospermia factor AZF controlling human spermatogenesis Cell 1993 75 1287 1295 8269511 10.1016/0092-8674(93)90616-X
Eberhart CG Maines JZ Wasserman SA Meiotic cell requirement for a fly homologue of human deleted in azoospermia Nature 1996 381 783 5 8657280 10.1038/381783a0
Simoni M Bakker E Eurlings MC Laboratory guidelines for molecular diagnosis of Y-chromosomal microdeletions Int J Androl 1999 22 292 299 10509229 10.1046/j.1365-2605.1999.00193.x
Tsujimura A Matsumiya K Takao T Miyagawa Y Koga M Takeyama M Fujioka H Okuyama A Clinical analysis of patients with azoospermia factor deletions by microdissection testicular sperm extraction Int J Androl 2004 27 76 81 15149464 10.1046/j.0105-6263.2003.00450.x
Carvalho CM Fujisawa M Shirakawa T Kamidono GA Freitas S Paulo T Santos SE Rocha J Pena SD Santos F Lack of association between Y chromosome haplogroups and male infertility in Japanese men American J Med Genet 2003 116A 152 8 10.1002/ajmg.a.10827
Ewis AA Lee J Shinka T Nakahori Y Microdeletions of a Y-specific marker, Yfm1, and implications for a role in spermatogenesis Journal of Hum Genet 2002 47 257 61 12032594 10.1007/s100380200035
Nagafuchi S Namiki M Nakahori Y Kondoh N Okuyama A Nakagome Y A minute deletion of the Y chromosome in men with azoospermia J Urol 1993 150 1155 7 8371377
Takao T Akira T Sada M Goto R Koga M Miyagawa Y Susceptibility gene for non azoospermia in the HLA 11 region ; Correlations with Y chromosome microdeletions and spermatogenesis Inter J Androl 2004 27 37 41 10.1046/j.0105-6263.2003.00445.x
Kobayashi K Mizuno K Hida A Komaki R Tomita K Matsushita I Namiki M Iwamoto T Tamura S Minowada S PCR analysis of the Y chromosome long arm in azoospermic patients: evidence for a second locus required for spermatogenesis published erratum appears in Hum Mol Genet 1995 4 974 7633461
Kato H Komori S Nakata Y Sakata K Kanazawa R Handa M Kobayashi S Koyama K Isojima S Screening for deletions in interval D16-22 of the Y chromosome in azoospermic and oligozoospermic Japanese men J Hum Genet 2001 46 110 4 11310577 10.1007/s100380170097
Nakahori Y Kuroki Y Komaki R Kondoh N Namiki M Iwamoto T Toda T Kobayashi K The Y chromosome region essential for spermatogenesis Hormone Research 1996 46 20 3 8864744
Reijo R Alagappan RK Patrizio P Page DC Severe oligozoospermia resulting from deletions of azoospermia factor gene on Y chromosome Lancet 1996 347 1290 1293 8622504 10.1016/S0140-6736(96)90938-1
Silber SJ Alagappan R Brown LG Page DC Y chromosome deletions in azoospermic and severely oligozoospermic men undergoing intracytoplasmic sperm injection after testicular sperm extraction Hum Reprod 13 3332 3337 9886509 10.1093/humrep/13.12.3332
Hopps CV Mielnik A Goldstein M Palermo GD Rosenwaks Z Schlegel PN Detection of sperm in men with Y chromosome microdeletions of the AZFa, AZFb and AZFc regions Hum Reprod 2003 18 1660 1665 12871878 10.1093/humrep/deg348
Mulhall JP Reijo R Alagappan Azoospermic men with deletion of the DAZ gene cluster are capable of completing spermatogenesis: fertilization, normal embryonic development and pregnancy occur when retrieved testicular spermatozoa are used for intracytoplasmic sperm injection Hum Reprod 1997 12 503 508 9130751 10.1093/humrep/12.3.503
Oates RD Silber S Brown L Page DC Clinical characterization of 42 oligospermic or azoospermic men with microdeletion of the AZFc region of the Y chromosome, and of 18 children conceived via ICSI Hum Reprod 2002 17 2813 2824 12407032 10.1093/humrep/17.11.2813
Choi JM Chung P Veeck LA Mielnik LA Palermo GD Schlegel PN AZF microdeletions of the Y chromosome and in vitro fertilization outcome Fertil Steril 2004 81 337 341 14967370 10.1016/j.fertnstert.2003.06.030
Van Golde RJT Wetzels AMM de Graaf R Decreased fertilization rate and embryo quality after ICSI in oligozoospermic men with microdeletions in the azoospermia factor c region of the Y chromosome Hum Reprod 2001 16 289 292 11157822 10.1093/humrep/16.2.289
Silber SJ pontaneous pregnancy in couples with very severe oligospermia (<0.5 × 10.6 sperm): implications for transmission of Y chromosome deletions Fertil Steril 2001 76 S10 10.1016/S0015-0282(01)02044-1
Kihaile PE Kisanga RE Aoki K Kumasako Y Misumi J Utsunomiya T Embryo Outcome in Y-chromosome Microdeleted Infertile Males after ICSI: Mo Reprod Dev 2004 68 176 181 10.1002/mrd.20074
Page DC Silber S Brown LG Men with infertility caused by AZFc deletion can produce sons by intracytoplasmic sperm injection but are likely to transmit the deletion and infertility Hum Reprod 1999 14 1722 1726 10402375 10.1093/humrep/14.7.1722
Rolf C Gromoll J Simoni M Nieschlag1 E Natural transmission of a partial AZFb deletion of the Y chromosome over three generations: Case report Hum Reprod 2002 17 2267 2271 12202412 10.1093/humrep/17.9.2267
Henegariu O Hirschmann P Kilian K Rapid screening of the Y chromosome in idiopathic sterile men, diagnostic for deletions in AZF, a genetic Y factor expressed during spermatogenesis Andrologia 1994 26 97 106 8042776
Utsunomiya T Naitou T Nagaki M A prospective trial of blastocyst culture and transfer Hum Reprod 2002 17 1846 1851 12093850 10.1093/humrep/17.7.1846
Peterlin B Kunej T Sinkovec J Gligorievska N Zorn B Screening for Y chromosome microdeletions in 226 Slovenian subfertile men Hum Reprod 2000 17 17 24 11756356 10.1093/humrep/17.1.17
Foresta C Moro E Ferlin A Y chromosome microdeletions and alterations of spermatogenesis Endocrine Reviews 2001 22 226 239 11294825 10.1210/er.22.2.226
Reijo R Lee TY Salo P Alagappan R Brown LG Rosenberg M Diverse spermatogenic defects in humans caused by Y chromosome deletions encompassing a novel RNA-binding protein gene Nature Genetics 1995 10 383 393 7670487 10.1038/ng0895-383
Foresta C Ferlin A Garolla A Moro E Pistorello M Barbaux S Rossato M High frequency of well-defined Y-chromosome deletions in idiopathic Sertoli cell-only syndrome Human Reproduction 1998 13 302 307 9557827 10.1093/humrep/13.2.302
Yeboah ED Wadhwani JM Wilson JB Etiological factors of male infertility in Africa International Journal of Fertility 1992 37 300 7 1358843
Cates W Farley TM Rowe PJ Worldwide patterns of infertility: is Africa different? Lancet 1985 2 596 8 2863605 10.1016/S0140-6736(85)90594-X
Imade GE Towobola OA Sagay AS Otubu JA Sexually transmitted diseases and medico-social factors associated with male infertility in Nigeria Archives of AIDS Research 1993 7 245 52 12288483
Bornman MS Crewe-brown HH Reif S Mahomed MF Broomker D Schulenburg GW Sexually transmitted diseases (STD) in infertile males attending the andrology clinic at Ga-Rankuwa Hospital Archives of AIDS Research 1993 145 150
Collet M Reniers J Frost E Gass R Yvert F Leclerc A Roth-Meyer C Ivanoff B Meheus A Infertility in Central Africa: infection is the cause International Journal of Gynaecol & Obstet 1988 26 423 8 10.1016/0020-7292(88)90340-2
Vigil P Morales P Tapia A Riquelme R Salgado AM Chlamydia trachomatis infection in male partners of infertile couples: incidence and sperm function Andrologia 2002 34 155 61 12059810 10.1046/j.1439-0272.2002.00472.x
Ness R Markovic N Carlson CL Coughlin MT Do men become infertile after having sexually transmitted urethritis? An epidemiologic examination Fertil & Steril 1997 68 205 13 9240243 10.1016/S0015-0282(97)81502-6
Bambra CS Current status of reproductive behaviour in Africa Human Reproduction Update 1999 5 1 20 10333365 10.1093/humupd/5.1.1
Gausset Q AIDS and cultural practices in Africa: the case of the Tonga (Zambia) Social Science & Medicine 2001 52 509 18 11206649 10.1016/S0277-9536(00)00156-8
|
15987528
|
PMC1198247
|
CC BY
|
2021-01-04 16:40:17
|
no
|
J Exp Clin Assist Reprod. 2005 Jun 29; 2:9
|
utf-8
|
J Exp Clin Assist Reprod
| 2,005 |
10.1186/1743-1050-2-9
|
oa_comm
|
==== Front
J Inflamm (Lond)Journal of Inflammation (London, England)1476-9255BioMed Central London 1476-9255-2-91609114010.1186/1476-9255-2-9ResearchProanthocyanidins, from Ribes nigrum leaves, reduce endothelial adhesion molecules ICAM-1 and VCAM-1 Garbacki N [email protected] M [email protected] B [email protected] D [email protected] J [email protected] Laboratoire de Physiologie humaine, CHU, Tour 3, Université de Liège, Avenue de l'Hôpital, 3, B-4000 Sart Tilman, Belgium2 Laboratoire de Biologie des Tissus Conjonctifs, Tour 3, Université de Liège, Avenue de l'Hôpital, 3, B-4000 Sart Tilman, Belgium3 Département de Morphologie et de Pathologie, Pathologie générale, Faculté de Médecine vétérinaire, Université de Liège, bvd de Colonster, 20, B-4000 Sart Tilman, Belgium2005 9 8 2005 2 9 9 25 1 2005 9 8 2005 Copyright © 2005 Garbacki et al; licensee BioMed Central Ltd.2005Garbacki et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The effects of proanthocyanidins (PACs), isolated from blackcurrant (Ribes nigrum L.) leaves, on neutrophil accumulation during inflammatory processes were investigated in vivo and in vitro.
Methods
In vivo studies were performed using carrageenin-induced pleurisy in rats pre-treated with PACs. Exudate volume and PMNs accumulation were measured. Leukocyte cell adhesion molecules (LFA-1, Mac-1 and VLA-4) mobilization in circulating granulocytes were analysed by flow cytometry and endothelial cell adhesion molecules (ICAM-1 and VCAM-1) were detected by immunohistochemistry on lung sections.
In vitro studies were conducted on endothelial LT2 cells, stimulated with TNF-α, to evaluate ICAM-1, IL-8 and VEGF mRNA expression upon PACs treatment.
Data sets were examined by one-way analysis of variance (ANOVA) followed by a Scheffe post-hoc test.
Results
Pretreatment of the animals with PACs (10, 30 and 60 mg/kg) inhibited dose-dependently carrageenin-induced pleurisy in rats by reducing pleural exudate formation and PMNs infliltration. Leukocyte cell adhesion molecules mobilization was not down-regulated on granulocytes by PACs. Immunohistochemistry on lung sections showed a decreased production of endothelial cell adhesion molecules.
In vitro experiments demonstrated that PACs were able to significantly inhibit ICAM-1 but not IL-8 and VEGF165 mRNA expression. Moreover, VEGF121 mRNA expression was dose-dependently enhanced.
Conclusion
This study provides evidence to support the anti-inflammatory activity of proanthocyanidins is related to an inhibition of leukocyte infiltration which can be explained at least in part by a down-regulation of endothelial adhesion molecules, ICAM-1 and VCAM-1 and that these compounds are capable of modulating TNF-α-induced VEGF transcription.
==== Body
Introduction
Cell-cell and cell-extracellular matrix interactions play a central role in cell migration and leukocyte activation during the inflammatory process. Leukocyte adhesion to the endothelial cells and recruitment to extravascular tissue are fundamental events in the pathogenesis of the inflammatory disease. All the steps in the recruitment cascade are orchestrated by cell-adhesion molecules (CAMs) expressed on both leukocytes and endothelial cells, and different subsets of CAMs are responsible for the different steps in leukocyte extravasation [1]. This cascade includes the selectins, the integrins and members of the immunoglobulin superfamily. The expression of these adhesion molecules is induced by various stimuli including cytokines such as TNF-α, IL-1β, chemokines such as IL-8, lipid mediators and peptides [2].
ICAM-1 (Intercellular Adhesion Molecule-1, also designated as CD54) and VCAM-1 (Vascular Cell Adhesion Molecule-1, also designated as CD106) are members of immunoglobulin superfamily involved in cell adhesion. ICAM-1 is normally found on the surface of endothelial cells, but its expression can be significantly increased upon endothelial cell activation with cytokines or endotoxin. ICAM-1 binds to integrin members of the beta-1 subfamily such as LFA-1 (Lymphocyte Function-associated Antigen-1, also designated as CD11a/CD18) and Mac-1 (Macrophage antigen-1, also designated as CD11b/CD18). Leukocyte adhesion to ICAM-1 is performed by LFA-1 upregulation and conformational change and by Mac -1 mobilisation to the cell surface by fusion of storage granules to the cell membrane [3]. VCAM-1, a transmembrane endothelial protein, is the counter-receptor of VLA-4 (Very Late Antigen-4, also designated as α4β1 or CD49d/CD29). This latter beta-1 integrin mediates the adhesion of lymphocytes, monocytes, eosinophils and natural killer cells to activated endothelial cells [4].
VEGF (Vascular Endothelial Growth Factor) and IL-8 are two major mediators involved in inflammatory processes associated with angiogenesis [5,6]. VEGF functions as an endothelial growth and survival factor, it also increases endothelial permeability [5] and is particularly relevant to neutrophils emigration into the lung [7]. It causes vasodilatation through the nitric oxide synthase pathway in endothelial cells [8] and can activate migration of neutrophils [7,9,10]. IL-8 stimulates the ability of neutrophils to invade injured tissues. Besides its chemotactic influence, IL-8 also stimulates a rapid Mac-1 as well as LFA-1 expression on neutrophils which enables the adherence of these cells to activated vascular endothelial cells [11].
Proanthocyanidins (PACs) are a group of biologically active polyphenols, synthesized by many plants. PACs isolated from blackcurrant (Ribes nigrum L.) leaves are monomers and oligomers of flavan-3-ol units, the prodelphinidins and the procyanidins [12,13]. These compounds have been reported to have anti-inflammatory properties in vivo and in vitro [14-17]. We have previously shown that PACs decrease the accumulation of circulating leukocytes and plasma exudation during carrageenin acute inflammatory reaction induced in rats. This anti-inflammatory effect was associated with a reduction of pro-inflammatory cytokines [18]. The current study was undertaken to build on that previous finding and investigate the effects of the PACs on adhesion molecules during inflammatory processes.
Materials and methods
Proanthocyanidins
Proanthocyanidins from Ribes nigrum leaves were extracted and isolated according to a previously described method [12]. A voucher sample (RN 210590) has been deposited in the Pharmaceutical Institute of Liège, Belgium. Briefly, leaves were powdered separately and then extracted at room temperature with acetone (70 % v/v in water). The acetone was removed under vacuum at 40°C. The resulting aqueous solution was freeze-dried. Isolation of monomers and oligomers was carried out by MPLC on reversed-phase RP8 with water-acetone (9:1) to obtain a total proanthocyanidin-enriched fraction (PACs).
Carrageenin-induced pleurisy in rats
Animals
Mae Wistar rats, weighing 250 – 300 g were used. The animals were maintained on a standard laboratory diet with free access to water. The experiments were conducted as approved by the Animal Ethics Committee of the University of Liège, Belgium.
Carrageenin-induced pleurisy
Rats were pretreated with an intraperitoneal (i.p.) injection of saline or PACs (10, 30 or 60 mg/kg) 30 min before the intrapleural injection of carrageenin. They were then anaesthetized with ketamine HCl (75 mg/kg, i.p.) and carrageenin (0.2 ml, 10 mg/ml) or saline (0.2 ml) was administered into the right pleural cavity. Each experimental group contained 6 animals. Four hours later, the animals were anaesthetized with sodium pentobarbital (80 mg/kg, i.p.). The peritoneal cavity was opened and blood (3–5 ml) was withdrawn into a heparinized syringe from the inferior vena cava.
The chest was carefully opened and the pleural cavity rinsed with 2.0 ml saline solution containing heparin (5 U/ml). Exudates and washing solutions were removed by aspiration. Exudates with blood were rejected. The volume of the exudates was calculated by subtracting the volume of the washing solution (2.0 ml) from the total volume recovered. A sample of each exudate was diluted in phosphate buffer and polymorphonuclear leukocytes count was performed using a hemocytometer.
After removal of the exudates, lungs were withdrawn and fixed for one day under 30 cm pressure with 10% formaldehyde aqueous solution containing 0.480 M Na2HPO4 and 0.187 M KH2PO4 (pH 7.2) at room temperature. They were then dehydrated by graded ethanol and embedded in Paraplast.
Production of LFA-1, Mac-1 and VLA-4
The relative expression by circulating granulocytes of LFA-1, Mac-1 and VLA-4 were measured using flow cytometry. Granulocytes (200 000 cells) were stained by saturating concentrations of monoclonal antibodies: R-PE-conjugated WT.1 against rat CD11a (0.1 μg per million cells), FITC-conjugated WT.5 against CD11b (1 μg per million cells) or FITC-conjugated MRα4-1 against CD49d (0.1 μg per million cells).
After incubation of whole blood (40 μl), withdrawn into a heparinized syringe from the inferior vena cava, for 20 min in the dark at room temperature, erythrocytes were lysed with ammonium chloride buffer (30 min, in the dark at room temperature). The cells were finally washed three times with phosphate buffer saline containing 0.2% bovine serum albumin and 0.09% sodium azide (pH 7.4), fixed with 4% paraformaldehyde in PBS and then analyzed on a flow cytometer.
For this experiment, aspirin was used as a reference drug and was administrated intraperitoneally to rats at 200 mg/kg.
The antibodies were supplied by BD Pharmingen (Erembodegem, Belgium).
Immunohistochemical localization of ICAM-1 and VCAM-1
Indirect immunohistochemical staining was performed on 7 μm thick lung sections (2 sections per animal). After deparaffinization with UltraClear, endogenous peroxidase was quenched with 3% H2O2 in H2O for 10 min. The sections were permeabilized with 0.1% Tween 20 in PBS for 10 min and then incubated overnight with a 1/100 primary mouse anti-rat ICAM-1 monoclonal antibody or a 1/500 primary mouse anti-rat VCAM-1 monoclonal antibody or with control solution (buffer alone). Specific labelling was detected by incubating the sections 30 min at room temperature with a 1/100 biotin-conjugated goat anti-mouse IgG. Streptavidin-HRP complex was then added to cover the tissue sections and incubated 30 min at room temperature. DAB substrate was applied to the sections, incubated for 5 min and finally washed in water. Tissue sections were then counterstained with hematoxylin-eosin, dehydrated and examined using light microscopy.
The antibodies, streptavidin-HRP complex pre-diluted and DAB substrate kit were supplied by BD Pharmingen (Erembodegem, Belgium).
ICAM-1, IL-8 and VEGF mRNA expression in LT2 endothelial cells
Cell culture
LT2 cell line originates from human umbilical vein endothelium cells (HUVEC) which have been immortalized by transfection with Large T SV40 antigen (a kind gift of E. Dejana, Milan, to one of us BN). Cells were grown at 37°C in a humidified 95% air-5% CO2 atmosphere, using 2% pork skin gelatin-coated petri dish and RPMI 1640 medium supplemented with 10% FCS, 45 U/ml penicillin and 45 μg/ml streptomycin.
Cytotoxicity of PACs on LT2 cells was previously tested using the MTT assay (Sigma, Bornem, Belgium), based on conversion by mitochondrial dehydrogenases of the substrate (MTT) containing tetrazolium ring into the blue formazan detectable spectrophotometrically [19]. The level of blue formazan is then used as indirect index of cell density. The optical density of each sample was measured with a microplate spectrophotometer reader at 560 nm. Three replicates were used for each sample.
mRNA determination using RT-PCR
LT2 cells (100 000 cells), pretreated 24 h with PACs (10, 30 and 60 μg/ml) or with PBS (controls), were activated in triplicate with TNF-α (5 ng/ml, 12 h). The medium was then discarded and the cells frozen (-70°C). The level of mRNA was determined using RT-PCR as described below.
RNA extraction
The total RNA was isolated using High Pure RNA Isolation kit (Roche Applied Science, Penzberg, Germany). Briefly, endothelial cells were treated by a lysis buffer which contains guanidine hydrochloride. The lysates were laid on filter tubes, where RNA was bound, and contaminating DNA eliminated with DNase containing buffer. The bound RNA was purified from salts, proteins and other cellular impurities by washing steps and finally eluted in water. Quantification of RNA was assayed by Ribo Green RNA Quantification kit (Molecular Probes Europe BV, Leiden, The Netherlands) and RNA samples were diluted to 4 ng/μl.
One step RT-PCR
The one-step RT-PCR reactions were performed in duplicate using the GeneAmp Thermostable rTth Reverse Transcriptase RNA PCR kit (Applied Biosystems, Roche Molecular Systems, New Jersey, USA), specific pairs of primers (Table 1), 10 ng of total RNA and, when available, a known copy number of an original synthetic internal standard of RNA, in 25 μl reaction mixture [20].
Table 1 Specific pairs of primers used for RT-PCR analysis.
Reverse (5'-3') Forward (5'-3')
28S -GATTCTGACTTAGAGGCGTTCAGT- -GTTCACCCACTAATAGGGAACGTGA-
ICAM-1 -TCCAGTTCAGTGCGGCACGAGAA- -CTGATGGGCAGTCAACAGCTAAAA-
IL-8 -GAATTCTCAGCCCTCTTCAAAAAC- -GCCAAGGAGTGCTAAAGAACTTAG-
VEGF -CTCACCGCCTCGGCTTGTCACA- -CCTGGTGGACATCTTCCAGGAGTA-
The RT step (70°C, 15 min) was followed, after an initial denaturation for 2 min at 95°C, by PCR amplification for the adequate number of cycles and terminated by a final elongation step of 2 min at 72°C. The PCR conditions for amplification of ICAM-1 (30 cycles) and the 28S RNA (17 cycles) were: 94°C for 15 sec; 66°C for 20 sec; 72°C for 10 sec. For IL-8 (25 cycles), the conditions were: 94°C for 15 sec, 60°C for 30 sec, 72°C for 10 sec. For VEGF (32 cycles), the conditions were: 94°C for 20 sec, 66°C for 30 sec, 72°C for 1 min.
A 10 μl aliquot from each PCR reaction were subjected to electrophoresis in 10% polyacrylamide gel and analysed using a Fluor-S-MultiImager (BioRad, Hercules, Ca) after staining with GelStar (FMC BioProducts, Rockland, ME, USA) dye. The results were expressed in arbitrary units per unit of 28S RNA. The 28S, ICAM-1 and VEGF primers were commercially synthesized by Invitrogen (Merelbeke, Belgium) and IL-8 primers by Eurogentec (Liège, Belgium).
Statistical evaluation
Results are given as mean ± standard error of the mean (s.e. mean) of N observations. Data sets were examined by one-way analysis of variance (ANOVA) followed by a Scheffe post-hoc test. A P-value of less than 0.05 was considered significant.
Results
Carrageenin-induced pleurisy
In control rats, the volume of the exudate collected 4 h after carrageenin injection reached 0.98 ± 0.06 ml per rat (n = 6) (Figure 1). This exudate contained a large number of cells, mostly (> 95%) polymorphonuclear leukocytes (PMNs). The total leukocyte number in the exudate was 88.53 ± 5.4 × 106 per rat (Figure 2). PACs (10, 30 and 60 mg/kg) significantly reduced the volume of the exudate in a dose-dependent relationship (by, respectively, 31, 37 and 55%). Moreover, PMNs infiltration was also significantly inhibited by PACs in a dose-dependent way (by, respectively, 64, 73 and 75%).
Figure 1 Effect of PACs on exudate volume during carrageenin-induced pleurisy. At 4 h after intrapleural injection of carrageenin, the volume of the exudate was reduced by PACs (10, 30 and 60 mg/kg) administration. Each value is the mean ± s.e. mean of n = 6 experiments. °P < 0.01 versus saline. *P < 0.05 versus carrageenin. **P < 0.01 versus carrageenin.
Figure 2 Effect of PACs on PMNs infiltration during carrageenin-induced pleurisy. At 4 h after intrapleural injection of carrageenin, the accumulation of PMNs in the pleural cavity was dramatically inhibited by PACs (10, 30 and 60 mg/kg). Each value is the mean ± s.e. mean of n = 6 experiments. °P < 0.01 versus saline. **P < 0.01 versus carrageenin.
Production of LFA-1, Mac-1 and VLA-4
In order to examine whether PACs affect the expression of adhesion molecules on leukocytes during carrageenin-induced pleurisy, the expression of three different molecules: LFA-1 (CD11a/CD18), Mac-1 (CD11b/CD18) and VLA-4 (CD49d/CD29) was measured by flow cytometry.
None of these molecules was significantly modulated during carrageenin-induced pleurisy (Figure 3). Aspirin reduced the level of VLA-4, as also did PACs at 10 mg/kg (Figure 3C) but at 60 mg/kg PACs increased the expression of Mac-1 (Figure 3B) and VLA-4 (Figure 3C).
Figure 3 Effect of PACs and aspirin on leukocyte cell adhesion molecules during carrageenin-induced pleurisy. The relative expression by circulating granulocytes of LFA-1, Mac-1 and VLA-4 were measured using flow cytometry. Granulocytes were incubated with saturating concentrations of monoclonal antibodies (mAb): R-PE-conjugated WT.1 against rat CD11a (LFA-1) (A), FITC-conjugated WT.5 against CD11b (Mac-1) (B) or FITC-conjugated MRα4-1 against CD49d (VLA-4) (C). mAb against CD11a gave negative results for all tested groups. PACs induced up-regulation of CD11b and CD49d with the highest dose tested and down-regulation of CD49d with the lowest dose. Aspirin also down-regulated CD49d at 200 mg/kg. Each value is the mean ± s.e. mean of n = 6 experiments. *P < 0.05 versus carrageenin.
Immunohistochemical localization of ICAM-1 and VCAM-1
Immunohistochemical analysis of lung sections obtained from animals, treated with carrageenin alone, revealed positive staining for ICAM-1 (Figure 4) and VCAM-1 (Figure 5). In contrast, sham group showed poor staining. In lung tissue obtained from PACs pre-treated rats, immunohistochemical staining for ICAM-1 and VCAM-1 was lower than those of positive control and disappeared with the highest doses tested to reach negative control level at 60 mg/kg.
Figure 4 Immunohistochemical localization of ICAM-1 in the lung. When compared to lung sections taken from sham group (A), lung sections from carrageenin-treated rats (B) demonstrated intense positive brown staining for ICAM-1 (black arrow). The degree of tissue staining for ICAM-1 from carrageenin-treated rats that had received PACs (10, 30 and 60 mg/kg, respectively C, D and E) was markedly reduced. Original magnification: × 125. This figure is representative of six experiments performed on different animals.
Figure 5 Immunohistochemical localization of VCAM-1 in the lung. When compared to lung sections taken from sham group (A), lung sections from carrageenin-treated rats (B) demonstrated intense positive brown staining for VCAM-1 (black arrow). The degree of tissue staining for VCAM-1 from carrageenin-treated rats that had received PACs (10, 30 and 60 mg/kg, respectively C, D and E) was markedly reduced. Original magnification: × 125. This figure is representative of six experiments performed on different animals.
ICAM-1, VEGF and IL-8 mRNA expression
PACs were tested from 5 to 100 μg/ml on LT2 cells during 24 h and were not found to be significantly cytotoxic for all the used doses. The effect of PACs on TNF-α-induced ICAM-1, VEGF and IL-8 expression in LT2 cells was then investigated. Basal ICAM-1, IL-8, VEGF121 and VEGF165 expression levels were low in unstimulated cells (Figure 6). Treatment of cells with TNF-α for 12 h significantly increased the expression of ICAM-1, IL-8, VEGF121 and VEGF165. Treatment of cells with PACs alone had no effect on the basal expression of ICAM-1, VEGF and IL-8 genes by LT2 cells (data not shown). A significant inhibition in inducible ICAM-1 expression was observed in a dose-dependent way in TNF-α stimulated cells (Figure 6A). PACs did not significantly modify IL-8 (Figure 6B) nor VEGF165 (Figure 6D) expression in TNF-α stimulated cells while it increased dose-dependently the expression of VEGF121 (Figure 6C).
Figure 6 Effect of PACs on TNF-α-induced ICAM-1, IL-8, VEGF121 and VEGF165 mRNA expression in human endothelial LT2 cells. LT2 cells were pretreated with PACs (10, 30 and 60 μg/ml) or PBS for 24 h and then activated with 5 ng/ml TNF-α for 12 h. PACs dose-dependently inhibited mRNA expression of ICAM-1 (A), maintained IL-8 (B) and VEGF165 (D) and up-regulated dose-dependently VEGF121 mRNA expression (C). Gel electrophoresis of PCR products is shown under X-axis. Each value is the mean ± s.e. mean of three experiments analysed in duplicate each. °P < 0.01 versus negative control. *P < 0.05 versus positive control. **P < 0.01 versus positive control.
The downregulation effects of PACs on inducible ICAM-1 expression were not a result of cytotoxicity, because no significant difference of viability was observed between non-treated or PACs-treated cells (up to 100 μg/ml) using the MTT assay.
Discussion
Our results confirm and extend previous data [14-16,18] showing that proanthocyanidins (PACs) from Ribes nigrum leaves reduce carrageenin-induced pleurisy. The volume of the exudate was dose-dependently diminished by PACs. In this fluid, PACs decreased IL-1β, TNF-α, CINC-1 and nitrite/nitrate levels [18]. Simultaneously, the levels of leukocytes, mostly polymorphonuclear cells, were decreased. The transvasation of leukocytes from blood to the inflammatory exudate requires activation of leukocytes and expression of adhesion molecules on the surface of endothelial cells. We have examined both processes. PACs did not significantly reduce the expression of adhesion molecules on the surface of leukocytes but greatly reduced the production of adhesion molecules on the surface of endothelial cells in vivo during carrageenin-induced pleurisy.
Three adhesion molecules expressed by leukocytes were examined. LFA-1 (CD11a/CD18) is not stored within the leukocytes, its upregulation most likely results from a conformational change that allows the leukocyte to interact with ICAM-1 [3]. All the animals treated with saline or carrageenin in the pleural cavity or PACs showed the same level of LFA-1 expression. This lack of difference could depend on the fact that our antibody cannot differentiate quiescent LFA-1 from activated LFA-1. On the other hand, PACs did not reduce Mac-1, an inducible β2-integrin, but increased its expression at high doses. While low doses of PACs reduced VLA-4 expression, higher doses of PACs increased its mobilization to the membrane. The influence of PACs on adhesion molecules on leukocytes was thus variable and cannot explain the reduction of leukocyte accumulation in the pleural fluid.
In pulmonary tissue, our immunohistochemical studies showed a dose-dependent decrease in the expression of VCAM-1 and ICAM-1. These results were confirmed by the examination of ICAM-1 production by LT2 endothelial cells in vitro. PACs dose-dependently reduced ICAM-1 mRNA production, suggesting that PACs inhibited ICAM-1 production at the mRNA level. Previously, it was shown that other compounds of the proanthocyanidin family inhibit the expression of endothelial adhesion molecules, in systemic sclerosis patients [21], in keratinocytes [22] and in HUVEC [23]. In LT2 endothelial cells, the level of ICAM-1 mRNA was increased 25-fold by TNF-α. This overproduction would depend on the activation of NF-κB or JNK/AP-1 transduction pathway. Previous results from our laboratory (data not shown) have revealed that PACs are able to inhibit NF-κB activation but at concentrations 10 times higher than the dose used to inhibit ICAM-1 mRNA expression in LT2 cells. On the other hand, compounds from the proanthocyanidin family have been suggested to inhibit the activation of JNK-1 and c-JUN [23]. Thus the inhibition of the expression of adhesion molecules by PACs could be related to an interference of JNK/c-JUN/AP-1 pathway activation.
The inhibition of ICAM-1 mRNA production by PACs was rather specific as PACs did not reduce IL-8 mRNA and VEGF mRNA in LT2 cells. Although, we have observed that in vivo PACs inhibited CINC-1 production during carrageenin-induced pleurisy, these compounds did not affect the production of similar molecules such as IL-8 by LT2 endothelial cells. On the other hand, PACs interact with VEGF mRNA expression in vitro. VEGF functions as an endothelial growth and survival factors and also increases vascular permeability [5]. It may play a relevant role in neutrophil migration. Five human VEGF mRNA species encoding VEGF isoforms of, among them, the most abundant 121 and 165 amino acids are produced by alternative splicing of VEGF mRNA [24]. The deletion in VEGF121 begins at codon 116 of the mature form of VEGF165 [24]. PACs did not influence expression of VEGF165 mRNA but provoked an increase in VEGF121 expression. VEGF production is stimulated by cytokines such as IL-6 [25], IL-1β [26] and TNF-α [27]. Up-regulation of inducible VEGF in human HaCaT keratinocytes by compounds from the proanthocyanidin family was also observed by Khanna et al. [28,29], who correlated this property to redox activity of these products. Proanthocyanidins are, in fact, known to display anti-oxidant and free radical scavenging properties [30-34]. This effect of VEGF up-regulation may have a beneficial role in regulating and facilitating wound healing.
In conclusion, our current study provides evidence to support that the anti-inflammatory activity displayed by proanthocyanidins from blackcurrant leaves is related to down-regulation of endothelial adhesion molecules, ICAM-1 and VCAM-1, probably through the transcription factor AP-1 regulation pathway. It would be of great interest in the near future to determine the effect of PACs and of similar compounds on the JNK/c-Jun/AP-1 regulatory pathway. Moreover, PACs are capable of modulating inducible VEGF transcription. The role of PACs in wound healing still remains to be determined.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
NG carried out PACs isolation, immunohistochemistry, cell culture, RT-PCR, statistical analysis, participated in animal experimentation, and drafted the manuscript. MK carried out flow cytometry and participated in animal experimentation. BN coordinated cell culture and RT-PCR techniques. DD coordinated and helped in immunohistochemistry. JD participated in animal experimentation, in the concept, design and coordination of the study and helped to draft the manuscript. All authors read and approved the final manuscript.
==== Refs
Carlos TM Harlan JM Leukocyte-endothelial adhesion molecules Blood 1994 84 2068 2101 7522621
Panés J Perry M Granger DN Leukocyte-endothelial cell adhesion: avenues for therapeutic intervention Br J Pharmacol 1999 126 537 550 10188959 10.1038/sj.bjp.0702328
Panés J Granger DN Leukocyte-endothelial cell interactions: molecular mechanisms and implications in gastro-intestinal disease Gastroenterology 1998 114 1066 1090 9558298
Elices MJ Osborn L Takada Y Crouse C Luhowskyj S Hemler M Lobb RR VCAM-1 on activated endothelium interacts with the leukocyte integrin VLA-4 at a site distinct from the VLA-4/fibronectin binding site Cell 1990 60 577 584 1689216 10.1016/0092-8674(90)90661-W
Neufeld G Cohen T Gengrinovitch S Poltorak Z Vascular endothelial growth factor (VEGF) and its receptors FASEB J 1999 13 9 22 9872925
Baggiolini M Dewald B Moser B Interleukin-8 and related chemotactic cytokines-CXC CC chemokines Adv Immunol 1994 55 97 119 8304236
Burns AR Smith CW Walker DC Unique structural features that influence neutrophil emigration into the lung Physiol Rev 2003 83 309 336 12663861
Hood JD Meininger CJ Ziche M Granger HJ VEGF upregulates ecNOS message, protein, and NO production in human endothelial cells Am J Physiol 1998 274 H1054 H1058 9530221
Gaudry M Bregerie O Andrieu V El Benna J Pocidalo MA Hakim J Intracellular pool of vascular endothelial growth factor in human neutrophils Blood 1997 90 4153 4161 9354686
Taichman NS Young S Cruchley AT Taylor P Paleolog E Human neutrophils secrete vascular endothelial growth factor J Leukoc Biol 1997 62 397 400 9307080
Seo SM McIntire LV Smith CW Effects of IL-8, Gro-alpha, and LTB4 on the adhesive kinetics of LFA-1 and Mac-1 on human neutrophils Am J Physiol Cell Physiol 2001 281 C1568 1578 11600420
Tits M Angenot L Poukens P Warin R Dierckxsens Y Prodelphinidins from Ribes nigrum Phytochemistry 1992 31 971 973
Tits M Poukens P Angenot L Diercksens Y Thin-layer chromatographic analysis of proanthocyanidins from Ribes nigrum leaves J Pharm Biomed Anal 1992 10 1097 1100 1298369 10.1016/0731-7085(91)80128-V
Tits M Angenot L Damas J Dierckxsens Y Poukens P Anti-inflammatory prodelphinidins from black currant (Ribes nigrum) leaves Planta Med 1991 57 A134
Garbacki N Angenot L Bassleer C Damas J Tits M Effects of prodelphinidins isolated from Ribes nigrum on chondrocytes metabolism and COX activity Naunyn-Schmiedeberg's Arch Pharmacol 2002 365 434 441 12070756 10.1007/s00210-002-0553-y
Garbacki N Tits M Damas J Anti-inflammatory effect of natural proanthocyanidins : pharmacological evaluation on in vivo models Pflügers Archiv Eur J Physiol 2003 446 R5
Garbacki N Damas J Some effects of proanthocyanidins isolated from Ribes nigrum on the cardiovascular system of the rat Fund Clin Pharmacol 2004 18 270
Garbacki N Tits M Angenot L Damas J Inhibitory effects of proanthocyanidins from Ribes nigrum leaves on carrageenin acute inflammatory reactions induced in rats BMC Pharmacol 2004 4 25 15498105 10.1186/1471-2210-4-25
Mosmann T Rapid colorimetric assay for cellular growth and survival: application to proliferation and cytotoxicity assays J Immunol Meth 1983 65 55 63 10.1016/0022-1759(83)90303-4
Deroanne CF Bonjean K Servotte S Devy L Colige A Clausse N Blacher S Verdin E Foidart JM Nusgens BV Castronovo V Histone deacetylases inhibitors as anti-angiogenic agents altering vascular endothelial growth factor signaling Oncogene 2002 21 427 436 11821955 10.1038/sj.onc.1205108
Kalfin R Righi A Del Rosso A Bagchi D Generini S Matucci Cerinic M Das DK Activin, a grape seed-derived proanthocyanidin extract, reduces plasma levels of oxidative stress and adhesion molecules (ICAM-1, VCAM-1, and E-selectin) in systemic sclerosis Free Rad Res 2002 36 819 825 10.1080/1071576021000005249
Bito T Roy S Sen CK Packer L Pine bark extract pycnogenol downregulates IFN-γ-induced adhesion of T cells to human keratinocytes by inhibiting inducible ICAM-1 expression Free Rad Biol Med 2000 28 219 227 11281289 10.1016/S0891-5849(99)00229-4
Bagchi D Sen CK Ray SD Das DK Bagchi M Preuss HG Vinson JA Molecular mechanisms of cardioprotection by a novel grape seed proanthocyanidin extract Mut Res 2003 523–524 87 97
Houck KA Ferrara N Winer J Cachianes G Li B Leung DW The vascular endothelial growth factor family-identification of a fourth molecular species and characterization of alternative splicing of RNA Mol Endocrinol 1991 5 1806 1814 1791831
Cohen T Nahari D Cerem-Weiss L Neufeld G Levi B Interleukin-6 induces the expression of vascular endothelial growth factor J Biol Chem 1996 271 736 741 8557680 10.1074/jbc.271.2.736
Li J Perrella MA Tsai JC Yet SF Hsieh CM Yoshizumi M Patterson C Endege WO Zhou F Lee ME Induction of vascular endothelial growth factor gene expression by interleukin-1 beta in rat aortic smooth muscle cells J Biol Chem 1995 270 308 312 7814392 10.1074/jbc.270.1.308
Ryuto M Ono M Izumi H Yoshida S Weich HA Kohno K Kuwano M Induction of vascular endothelial growth factor by tumor necrosis facto alpha in human glioma cells – possible role of SP-1 J Biol Chem 1996 271 28220 28228 8910439 10.1074/jbc.271.45.28220
Khanna S Roy S Bagchi D Bagchi M Sen CK Upregulation of oxidant-induced VEGF expression in cultured keratinocytes by a grape seed proanthocyanidin extract Free Rad Biol Med 2001 31 38 42 11425488 10.1016/S0891-5849(01)00544-5
Khanna S Venojarvi M Roy S Sharma N Trikha P Bagchi D Bagchi M Sen CK Dermal wound healing properties of redox-active grape seed proanthocyanidins Free Rad Biol Med 2002 33 1089 1096 12374620 10.1016/S0891-5849(02)00999-1
Maffei Facino R Carini M Aldini G Berti F Rossoni G Bombardelli E Morazzoni P Procyanidins from Vitis vinifera seeds protect rabbit heart from ischemia/reperfusion injury: antioxidant intervention and/or iron and copper sequestering ability Planta Med 1996 62 495 502 9000880
Aucamp J Gaspar A Hara Y Apostolides Z Inhibition of xanthine oxidase by catechins from tea (Camellia sinensis) Anticancer Res 1997 17 4381 4385 9494537
Bagchi D Garg A Krohn RL Bagchi M Bagchi DJ Balmoori J Stohs SJ Protective effects of grape seed proanthocyanidins and selected antioxidants against TPA-induced hepatic and brain lipid peroxidation and DNA fragmentation and peritoneal macrophage activation in mice Gen Pharmacol 1998 30 771 776 9559333
Zhao J Wang J Chen Y Agarwal R Anti-tumor-promoting activity of a polyphenolic fraction isolated from grape seeds in the mouse skin two-stage initiation-promotion protocol and identification of procyanidin B5-3'-gallate as the most effective antioxidant constituent Carcinogenesis 1999 20 1737 1745 10469619 10.1093/carcin/20.9.1737
Bagchi D Bagchi M Stohs SD Ray SD Sen CK Preuss HG Cellular protection with proanthocyanidins derived from grape seeds Ann NY Acad Sci 2002 957 260 270 12074978
|
16091140
|
PMC1198248
|
CC BY
|
2021-01-04 16:36:23
|
no
|
J Inflamm (Lond). 2005 Aug 9; 2:9
|
utf-8
|
J Inflamm (Lond)
| 2,005 |
10.1186/1476-9255-2-9
|
oa_comm
|
==== Front
J Neuroengineering RehabilJournal of NeuroEngineering and Rehabilitation1743-0003BioMed Central London 1743-0003-2-171598751410.1186/1743-0003-2-17ResearchWearable feedback systems for rehabilitation Sung Michael [email protected] Carl [email protected] Alex [email protected] The Media Laboratory, Massachusetts Institute of Technology, Cambridge, MA, USA2 Massachusetts General Hospital, Department of Psychiatry, Boston, MA, USA2005 29 6 2005 2 17 17 10 2 2005 29 6 2005 Copyright © 2005 Sung et al; licensee BioMed Central Ltd.2005Sung et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
In this paper we describe LiveNet, a flexible wearable platform intended for long-term ambulatory health monitoring with real-time data streaming and context classification. Based on the MIT Wearable Computing Group's distributed mobile system architecture, LiveNet is a stable, accessible system that combines inexpensive, commodity hardware; a flexible sensor/peripheral interconnection bus; and a powerful, light-weight distributed sensing, classification, and inter-process communications software architecture to facilitate the development of distributed real-time multi-modal and context-aware applications. LiveNet is able to continuously monitor a wide range of physiological signals together with the user's activity and context, to develop a personalized, data-rich health profile of a user over time. We demonstrate the power and functionality of this platform by describing a number of health monitoring applications using the LiveNet system in a variety of clinical studies that are underway. Initial evaluations of these pilot experiments demonstrate the potential of using the LiveNet system for real-world applications in rehabilitation medicine.
==== Body
Background and Introduction
Over the next decade, dramatic changes in healthcare systems are needed worldwide. In the United State's alone, 76 million baby boomers are reaching retirement age within the next decade [1]. Current healthcare systems are not structured to be able to adequately service the rising needs of the aging population, and a major crisis is imminent. The current system is dominated by infrequent and expensive patient visits to physician offices and emergency rooms for prevention and treatment of illness. The failure to do more frequent and regular health monitoring is particularly problematic for the elderly with multiple co-morbidities and often tenuous and rapidly changing health states. Even more troubling is the fact that current medical specialists cannot explain how most problems develop because they usually only see patients when something has already gone wrong.
Given this impending healthcare crisis, it is imperative to extend healthcare services from hospitals into home environments. Although there has been little success in extending health care into the home, there clearly is a huge demand. In 1997, Americans spent $27 billion on health care outside of the health care establishment, and that amount has been increasing [2]. Moreover, our dramatically aging population makes it absolutely necessary to develop systems that keep people out of hospitals. By 2030, nearly 1 out of 2 households will include someone who needs help performing basic activities of daily living and labor-intensive interventions will become impractical because of personnel shortage and cost [2].
The best solution to these problems lies in more proactive healthcare technologies that put more control into the hands of patients. The vision is a healthcare system that will help an individual to maintain their normal health profile by providing better monitoring and feedback, so that the earliest signs of health problems can be detected and corrected. This can be accomplished affordably by continuously monitoring a wide range of vital signals, providing early warning systems for people with high-risk medical problems, and "elder care" monitoring systems that will help keep seniors out of nursing homes and in their independent living arrangements.
Most available commercial mobile healthcare platforms have focused on data acquisition applications, with little attention paid to enabling real-time, context-aware applications. Companies such as VivoMetrics [3], Bodymedia [4], and Mini-Mitter [5], have extended the basic concept of the ambulatory Holter monitor (enabling a physician to record a patient's ECG continuously for 24–48 hours), which for three decades has been the only home health monitor with widespread use [6]. Additions to this individual monitoring paradigm have been extended along two fronts: medical telemetry and real-time critical health monitoring. Regarding the former, various inpatient medical telemetry systems have been developed in recent years, focusing on providing an infrastructure for transporting and storing data from the patient to caregivers for later analysis [7]. In terms of the latter, a few systems have extended the health monitoring concept by augmenting a physiological monitor (usually based on a single physiological sensor) with specialized algorithms for real-time monitoring within specific application domains, such as heart arrhythmia, epileptic seizures, and sleep apnea, which can potentially trigger alerts when certain critical conditions or events occur [8,9]. However, the development of proactive healthcare technologies beyond these basic telemedicine and individual event monitoring applications has been rather slow. The main limitation has been the large costs and inflexibility of limited monitoring modalities associated with these technologies and the impracticality for long-term use in general settings.
This paper presents LiveNet, a flexible distributed mobile platform that can be deployed for a variety of proactive healthcare applications that can sense one's immediate context and provide feedback. Based on cost-effective commodity PDA hardware with customized sensors and data acquisition hub plus a lightweight software infrastructure, LiveNet is capable of local sensing, real-time processing, and distributed data streaming. This integrated monitoring system can also leverage off-body resources for wireless infrastructure, long-term data logging and storage, visualization/display, complex sensing, and computation-intensive processing. The LiveNet system allows people to receive real-time feedback from their continuously monitored and analyzed health state. In addition, LiveNet can communicate health information to caregivers and other members of an individual's social network for support and interaction. Thus, by combining general-purpose commodity hardware with specialized health/context sensing within a networked environment, it is possible to build a multi-functional mobile healthcare device that is at the same time a personal real-time health monitor, multimodal feedback interface, context-aware agent, and social network support enabler and communicator.
With the development of increasingly powerful diagnostic sensing technology, doctors can obtain more context specific information directly, instead of relying on a patient's recollection of past events and symptoms, which tend to be vague, incomplete, and error prone. While many of these specialized sensing technologies have improved with time, most medical equipment is still a long way off from the vision of cheap, small, mobile, and non-invasive monitors. Modern imaging technology costs thousands of dollars per scan, requires room-sized equipment chambers, and necessitates uncomfortable and time-consuming procedures. Personal health systems, on the other hand, must be lightweight, easy-to-use, unobtrusive, flexible, and non-invasive to make headway as viable devices that people will use.
As such, there is tremendous potential for basic non-invasive monitoring as a complement to more invasive diagnostic sensing devices. The LiveNet system focuses on using combinations of non-invasive sensing and contextual features (for example, heart rate, motion, voice features, skin conductance, temperature/heat flux, location) that can be correlated with more involved clinical physiology sensing such as pulse oximetry, blood pressure, and multi-lead ECG. Sensors in the LiveNet system can continuously monitor autonomic physiology, motor activity, sleep patterns, and other indicators of health. The data from these sensors can then be used to build a personalized profile of performance and long-term health over time tailored to the needs of the patient and their healthcare providers. This unique combination of features also allows for quantification of personal contextual data such as amount and quality of social interactions and activities of daily living. This type of information is potentially very useful for increasing the predictive power of diagnostic systems.
The most important aspect of the LiveNet system is that it enables practical, long-term, context specific continuous monitoring. Continuous monitoring ensures the capture of relevant events and the associated physiology wherever the patient is, expanding the view of healthcare beyond the traditional outpatient and inpatient settings. Long-term monitoring has the potential to help create new models of health behavior. For example, long-term monitoring may provide important insights into the efficacy and effectiveness of medication regimes on the physiology and behavior of a patient over time at resolutions currently unobtainable. In addition, progress in terms of understanding human physiology and behavior will result from the fact that long-term trends can be explored in detail. Such advances include tracking the development and evolution of diseases, development of predictors of response to treatment and relapse prevention, monitoring changes in physiology as people grow older, comparing physiology across different populations (gender, ethnicity, etc), and even knowing characteristic physiology patterns of people who are healthy (this last example is particularly important when it is necessary as a diagnostic methodology designed to quantitatively define abnormal behavior). The goal is to be able to detect repeating patterns in complex human behavior by analyzing the patterns in data collected from the LiveNet system. From continuous monitoring, a very fine granularity of quantitative data can be obtained, in contrast to the surveys and history-taking that has been the mainstay of long-term studies and health interventions to date.
The LiveNet System
There are three major components to the LiveNet system: a personal data assistant (PDA) based mobile wearable platform, the software network and resource discovery application program interface (API), and a real-time machine learning inference infrastructure. The LiveNet system demonstrates the ability to use standardized PDA hardware tied together with a flexible software architecture and modularized sensing infrastructure to create a system platform where sophisticated distributed healthcare applications can be developed. While the current system implementations are based on PDAs, the software infrastructure is designed to be portable to a variety of mobile devices, including cell phones, tablet computers, and other convergence devices. As such, the system leverages commercial off-the-shelf components with standardized base-layer communication protocols (e.g., TCP/IP); this allows for the rapid adoption and deployment of these systems into real-world settings.
The LiveNet system is based on the MIThril wearable architecture developed at the Massachusetts Institute of Technology (MIT) Media Laboratory [10]. This proven architecture combines inexpensive commodity hardware, a flexible sensor/peripheral interconnection bus, and a powerful light-weight distributed sensing, classification, and inter-process communications software layer to facilitate the development of distributed real-time multi-modal and context-aware applications.
The LiveNet hardware and software infrastructure provides a flexible and easy way to gather heterogeneous streams of information, perform real-time processing and data mining on this information, and return classification results and statistics. This information can result in more effective, context-aware and interactive applications within healthcare settings.
A number of key attributes of the LiveNet system that make it an enabling distributed healthcare system include:
• Hierarchical, distributed modular architecture
• Based on standard commodity/embedded hardware that can be improved with time
• Wireless capability with resource posting/discovery and data streaming to distributed endpoints
• Leverages existing sensor designs and commercial sensors for context-aware applications that can facilitate interaction in a meaningful manner and provide relevant and timely feedback/information
• Unobtrusive, minimally invasive, and minimally distracting
• Abstracted network communications with secure sockets layer (SSL) encryption with real-time data streaming and resource allocation/discovery
• Continuous long-term monitoring capable of storing a wide range of physiology as well as contextual information
• Real-time classification/analysis and feedback of data that can promote and enforce compliance with healthy behavior
• Trending/analysis to characterize long-term behavioral trends of repeating patterns of behavior and subtle physiological cues, as well as to flag deviations from normal behavior
• Enables new forms of social interaction and communication for community-based support by peers and establishing stronger social ties within family groups
LiveNet Mobile Technology
The LiveNet system is currently based on the Sharp Zaurus (Sharp Electronics Corporation, U.S.A.), a Linux-based PDA mobile device that leverages commercial development and an active code developer community. Although LiveNet can utilize a variety of Linux-based devices, the Zaurus PDA provides a very convenient platform. This device allows applications requiring real-time data analysis, peer-to-peer wireless networking, full-duplex audio, local data storage, graphical interaction, and keyboard/touch screen input.
In order to effectively observe contextual data, a flexible wearable platform must have a means to gather, process, and interpret this real-time contextual data [34]. To facilitate this, the LiveNet system includes a modular sensor hub called the Swiss-Army-Knife 2 (SAK2) board that can be used to instrument the mobile device for contextual data gathering.
The SAK2 is a very flexible data acquisition board that serves as the central sensor hub for the LiveNet system architecture. The SAK2 incorporates a powerful 40 MHz PIC microcontroller, high efficiency regulated power (both 5 V and 3.3 V to power the board and sensor network) from a flexible range of battery sources, a 2.4 GHz wireless tranceiver capable of megabit data rates, compact flash based memory storage, and various interface ports (I2C, RS-232 serial, daughter board connector).
The SAK2 board was designed primarily to interface a variety of sensing technologies with mobile device-based wearable platforms to enable real-time context-aware, streaming data applications. The SAK2 is an extremely flexible data acquisition hub, allowing for a wide variety of custom as well as third-party sensors to interface to it. In addition to being a sensor hub, the SAK2 can also operate in stand-alone mode (i.e., without a Zaurus or mobile PDA host) for a variety of long-term data acquisition (using the CF card connector) and real-time interactive applications.
Physiologic and Contextual Sensing Technology
In order to support long-term health monitoring and activities of daily living applications, a specialized extensible, fully integrated physiological sensing board called the BioSense was developed as a special add-on board to the SAK2. The board incorporates a three dimensional (3D) accelerometer, ECG, EMG, galvanic skin conductance, a serial-to-I2C converter (which can allow the simultaneous attachment of multiple 3rd party serial-based sensing devices to the sensor network), and independent amplifiers for temperature/respiration/other sensors that can be daisy-chained to provide a flexible range of amplification for arbitrary analog input signals. Toward developing more non-invasive sensing technologies, we have started a collaboration with the Fraunhofer Institute to shrink the BioSense hardware to create a microminiaturized embedded system that can be incorporated in wearable fabrics. A prototype of a working lead-less lightweight ECG shirt based on conductive textiles has already been created.
Along with the core physiological sensing capabilities of the LiveNet system with BioSense daughter board, a whole host of other custom and third party sensors can be seamlessly integrated with the system, including:
• Wearable Multiple Sensor Acquisition (WMSAD) Board, providing a 3D accelerometer, infrared (IR) tag, IR tag readers (vertical, for in-door location in place of GPS, and horizontal for peer or object identification), and microphone for telephony-grade 8-kHz audio. This is interfaced to the SAK2 via the I2C port. [11].
• Squirt IR Tags: IR beacons that can broadcast unique identifiers (up to 4 independent signals from separately mounted and direction-adjustable IR-LEDS) [12]. These can be used to tag individuals, objects, locations (such used in arrays on the ceiling to identify location to within meter resolution within indoor settings where GPS is not effective), or even as environmental sensors to identify the actuation of certain events such as opening/closing of drawers, cabinets, or doors.
• IR Tag Reader: to be used in conjunction with the Squirt tags to be able to identify tagged objects, people, or even locations [12].
• Accelerometer Board: 3D accelerometer board very useful for a variety of activity classification. It has been demonstrated that a single accelerometer board can be used to accurately classify activity state (standing, walking, running, lying down, biking, walking up stairs, etc). Interfaced to the SAK2 using the sensor port.
• BodyMedia SenseWear: An integrated health sensor package which provides heart rate (via a Polar heart strap), galvanic skin response, 2D accelerometer, temperature (ambient and skin), and heat flux in a small form-factor package worn on the back of the arm [4]. The SAK2 can interface to the SenseWear wirelessly via a 900-MHz tranceiver attached to the serial-toI2C bridge (the tranceiver interface and heart rate monitor was discontinued in the SenseWear Pro 2)
• MITes Environmental Sensor: a wireless 3D accelerometer using the nRF 2.4 GHz protocol has been developed by the house_n group at the Media Lab for wireless environmental sensors for monitoring human activities in natural settings [13]
• Socio-Badges: Multifunctional boards with on-board DSP processor capable of processing audio features, RF tranceiver, IR transceiver, a brightness-controllable LED output display, vibratory feedback, navigator switch, flash memory, audio input/microphone and optional LCD display [14]. This badge is meant for social-networking experiments and other interactive distributed applications.
The sensor hub also allows us to interface with a wide range of commercially available sensors, including pulse oximetry, respiration, blood pressure, EEG, blood sugar, humidity, core temperature, heat flux, and CO2 sensors. Any number of these sensors can be combined through junctions to create a diversified on-body sensor network. The LiveNet system can also be outfitted with BlueTooth, Secure Data (SD), or Compact Flash (CF) based sensors and peripherals, and other I/O and communication devices including GSM/ GPRS/ CDMA/ 1 × RTT modems, GPS units, image and video cameras, memory storage, and even full-VGA head-mounted displays.
With the combined physiological sensing board and third-party sensors, a fully outfitted LiveNet system can simultaneously and continuously monitor and record 3D accelerometer, audio, ECG, EMG, galvanic skin response, temperature, respiration, blood oxygen, blood pressure, heat flux, heart rate, IR beacon, and up to 128 independently channeled environmental activity sensors. The sensor data and real-time classification results from a LiveNet system can also be streamed to off-body servers for subsequent processing, trigger alarms or notify family members and caregivers, or displayed/processed by other LiveNet systems or computers connected to the data streams for complex real-time interactions.
Software
The software architecture allows designers to quickly design distributed, group-based applications that use contextual information about the members of a group. Layered on top of standard libraries, this middleware comprises three important parts: the Enchantment Whiteboard, the Enchantment Signal system, and the MIThril Real-Time Context Engine [10]. Respectively, these three layers provide the ability to easily coordinate between distributed applications, transmit high bandwidth signals between applications, and create classification modules that make a group's changing contextual information available to applications.
The Enchantment Whiteboard system is a distributed, client/server, inter-process communication system that provides a lightweight way for applications to communicate. This system processes, publishes, and receives updates, decoupling information from specific processes. This is particularly useful in mobile, group based applications where group members may not be known a priori and may come and go over time.
For higher bandwidth signals, especially those related to the sharing and processing of sensor data for context aware applications, we developed the Enchantment Signal system. The Signal system is intended to facilitate the efficient distribution and processing of digital signals in a network-transparent manner. The Signal system is based on point-to-point communications between clients, with signal "handles" being posted on Whiteboards to facilitate discovery and connection. In the spirit of Whiteboard interactions, the Signal API abstracts away any need for signal produces to know who, how many, or even if, there are any connected signal consumers.
The MIThril Real-Time Context Engine is an open-source, lightweight, and modular architecture for the development and implementation of real-time context classifiers for wearable applications. Using the context engine, we can implement lightweight machine learning algorithms (capable of running on an embedded system like the Zaurus PDA) to process streaming sensor data, allowing the systems to classify and identify various user-state context in real-time.
Sample Applications
In the following section, a number of real-world case examples of clinical applications built upon various parts of the LiveNet system are detailed. These examples demonstrate the modular, configurable nature of the LiveNet infrastructure and the flexibility of the architecture to accommodate a variety of high bandwidth, real-time applications.
Health and Clinical Classification
The LiveNet system has proven to be a convenient, adaptable platform for developing real-time monitoring and classification systems using a variety of sensor data, including accelerometer-based activity-state classification that can differentiate between a variety of activities (for example, running, walking, standing, biking, climbing stairs) [15], accelerometer-based head-nodding/shaking agreement classifiers, GSR-based stress and emotional arousal detectors, and audio-based speech feature classifiers that can help characterize conversation dynamics (for example, talking time, prosody, stress) [16].
Work on these real-time classifiers has also been extended to include a variety of health conditions. Examples of current collaborations between the MIT Wearable Computing Group [17] and healthcare providers have lead to a variety of pilot studies including a hypothermia study with the United States Natick Army Laboratories in Natick; a study on the effects of medication on the dyskinesia state of Parkinson's patients with neurologists at Harvard Medical School; a pilot epilepsy classifier study with the University of Rochester Center for Future Health; and a study of the course of depression treatment with psychiatrists at Harvard Medical School.
Critical Soldier Monitoring
Army Rangers and other soldiers must perform physically and mentally demanding tasks under challenging environmental conditions ranging from extreme heat to extreme cold. Thermoregulation, or the maintenance of core body temperature within a functional range, is critical to sustained performance. A research collaboration with the Army Research Institute for Environmental Medicine (ARIEM) at the Army Natick Labs was initiated to study the effects of harsh environments on soldier physiology through the use of non-invasive sensing. Specifically, non-invasive accelerometer sensing was used to determine hypothermia and cold exposure state, as part of a broader initiative to develop a physiologic monitoring device for soldiers under the US. Army's Objective Force Warrior Program.
In the study, a real-time wearable monitor was developed using the LiveNet system that is capable of accurately classifying shivering motion through accelerometer sensing and analysis using statistical machine learning techniques [18]. Real-time working classifier systems were developed from Gaussian Mixture Models using frequency features derived from calculating a finite Fourier transform (FFT) on the raw accelerometer data. Preliminary data demonstrate that shivering can be accurately distinguished with up to 95% accuracy from general body movements in various activities using continuous accelerometer sensing. Results also indicate that specific modes of shivering (subjects in the study all exhibited a light shiver at a characteristic frequency at the start of the protocol that progressed into a more noisy and energetic shivering response spread across more frequency bands, and ending in a dampened shivering toward the end of the protocol) may correlate with core body temperature regimes, as a person is exposed to cold over time. In fact, preliminary results from six subjects show that we can triage a soldier into three core body temperature regimes (Baseline/Cold/Very Cold) with accuracies in the 92–98% range using HMM (Hidden Markov Models) modeling techniques. HMM modeling has the advantage of being able to accurately model the time-dependent changes in shivering over time as an individual is exposed to cold. This exploratory research shows promise of eventually being able to develop robust real-time health monitoring systems capable of classifying cold exposure of soldiers in harsh cold environments with non-invasive sensing and minimal embedded computational resources.
Parkinson's Disease Monitoring
LiveNet promises to be especially effective for monitoring medical treatments. Currently, doctors prescribe medications based on population averages rather than individual characteristics, and they check the appropriateness of the medication levels only occasionally. With such a data-poor system, it is not surprising that medication doses are frequently over- or underestimated and that unforeseen drug interactions can occur. Stratifying the population into phenotypes using genetic typing will improve the problem, but only to a degree and only in limited ways currently.
Continuous monitoring of physiologic and behavioral parameters may be extremely effective in tailoring medications to the individual Parkinson's patient. In Parkinson's patients, there are a variety of symptoms and motor complications that can occur, ranging from tremors (rhythmic involuntary motions), akinesia (absence or difficulty in producing motion), hypokinesia (decreased motor activity), bradykinesia (slow down of normal movement), and dyskinesia (abnormal or disruptive movements). For these patients to function at their best, medications must be optimally adjusted to the diurnal variation of these symptoms. In order for this to occur, the managing clinician must have an accurate picture of how a patient's symptoms fluctuates throughout a typical day's activities and cycles. In these situations, a patient's subjective self-reports are not typically very accurate, so objective clinical assessments are necessary.
An automated Parkinson symptom detection system is needed to improve clinical assessment of Parkinson's patients. To achieve this, Dr. Klapper, from the Harvard Medical School, combined the LiveNet system's wearable accelerometers with neural network algorithms to classify the movement states of Parkinson's patients and provide a timeline of how the severity of the symptoms and motor complications fluctuate throughout the day [19,20]. Two pilot studies were performed, consisting of seven patients, with the goal of assessing the ability to classify hypokinesia, dyskinesia, and bradykinesia based on accelerometer data, clinical observation (using standard clinical rating scales), and videotaping. Using the clinical ratings of a patient as the gold standard, the result was highly accurate identification of bradykinesia and hypokinesia. In addition, the studies classified the two most important clinical problems – predicting when the patient "feels off" or is about to experience troublesome dyskinesia – with nearly 100% accuracy. Future collaborations will focus on integrating the physiologic responses in an effort to identify predictors of relapse in addition to the motion data in Parkinson's patients.
Epilepsy Seizure Detection
A pilot study has also been initiated with the University of Rochester's Strong Hospital [21] to characterize and identify epileptic seizures through accelerometry and to begin to develop an ambulatory monitor with a real-time seizure classifier using the LiveNet system. Typically, epilepsy studies focus on EEG and EMG-based physiology monitoring. However, as demonstrated by the Parkinsons' and activity classification studies, accelerometry is a very powerful context sensor that can be applied to the domain of epilepsy. The study protocol is currently being designed, and we hope to have subject run in the Fall of 2005.
Of particular note to patients who have epilepsy is the fact that it can manifest itself in an extremely wide range of idiosyncratic motions, in contrast to Parkinsons' patients, whose movements typically follow distinct, characteristic motions. However, motions from the epileptic seizures of a particular individual are normally fairly consistent. As such, a motion classification system specifically tailored to a particular individual could be highly effective at being able to identify an epileptic seizure at onset and at sub-threshold levels of awareness. In addition, many times, the epileptic individual has no recollection of a seizure, so a system that could determine if a seizure has occurred could be very useful for doctors to be able to properly diagnose the type and pattern of epilepsy in patients or to develop applications to alert caregivers to changes that could lead to medication adjustments earlier in the course of the illness. Again, future directions will involve continuous physiologic and voice feature analysis in combination with the motion sensors to increase the accuracy and understanding of patients with epilepsy.
General Activity Classification
Being able to predict an individual's immediate activity state is one of the most useful sources of contextual information. For example, knowing whether a person is driving, sleeping, or exercising could be useful for a health wearable to calculate general energy expenditure or to initiate an action. Many studies on activity classification have been conducted because of the importance to context-aware systems. Most previous studies on accelerometer-based activity classification that involves multiple activities states focuses on using multiple accelerometers and requires the specific placement of sensors on different parts of the body. In one study, it was shown that it is possible to obtain activity classification with an average of 84% for 20 daily activities (such as vacuuming, eating, folding laundry, etc), with the additional finding that classification accuracy dropped only slightly by decreasing the number of sensors to two including the wrist and waist) [22].
In contrast, we have conducted a pilot study on the use of the minimum set of sensors required to accomplish accurate activity classification. The ultimate goal is to use only a single sensor in random orientation placed close to a person's center of mass (i.e., near waist level), as this replicates the minimum setup requirements of a sensor-enabled mobile phone in the pocket of an individual. The goal is to demonstrate that accurate activity classification can be performed without the need for an extreme level of instrumentation (for example, some systems use up to 30 sensors [23]) or particular delicacy in the setup in order to achieve good classification results. This way, we are able to potentially reduce the cost of a recognition system as well as reducing the overall burden when using the technology.
Using a LiveNet system we have been able to discriminate between a set of major activities (for example, lying down, walking, running, sitting in the office, watching TV, and walking up/down stairs) with classification results in the 80–95% accuracy range using only a single accelerometer located on the torso of an individual [15]. This research is important as it indicates that it is feasible to do activity classification on embedded hardware without any specialized setup, wires, or other unwieldy parts. By integrating the accelerometer into an existing device that people are comfortable carrying around (for example, a cell phone), we can significantly lower the bar for developing a practical activity classification system to the mass market that is completely transparent to the user. When combined with physiological measurements such as heart rate and breathing rate, these measurements can then be collected to build a personalized profile of your body's performance and your nervous system's activation throughout your entire day, and assembled over a period of months or years to show long-term changes in overall cardiac fitness. In the future, computer software 'agents' (automatic computer programs) could even give you gentle reminders to keep up your routine if your activity level started to decline and make suggestions to optimize your performance.
Depression Therapy Trending
Mental diseases rank among the top health problems worldwide in terms of cost to society. Major depression, for instance, is the leading cause of disability worldwide and in the U.S. Depressive disorders affects approximately 19 million American adults and has been identified by both the World Health Organization and the World Bank as the second leading cause of disability in the United States and worldwide [28,29].
Toward understanding the long-term biology associated with severe depression, we have recently initiated a pilot study to assess the physiological and behavioral responses to treatment in major depression in subjects in an inpatient psychiatric unit prior to, during, and following electroconvulsive therapy (ECT). This study, the first of its kind, intends to correlate basic physiology and behavioral changes with depression and mood state through a 24-hour, long-term, continuous monitoring of clinically depressed patients undergoing ECT. We are using non-invasive mobile physiologic sensing technology in combination with sensing devices on the unit to develop physiological and behavioral measures to classify emotional states and track the effects of treatment over time. This project is a joint collaboration with the Massachusetts General Hospital (MGH) Department of Psychiatry.
The goal of this study is to test the LiveNet system based on the known models of depression and prior clinical research in a setting with combined physiologic and behavioral measures with continuous ambulatory monitoring. It is anticipated that changes in these measures (namely, GSR response, hear rate/heart rate variability, motor activity, vocal features, and movement patterns) will correlate with improvements in standard clinical rating scales and subjective assessment following treatment for depression throughout the course of hospitalization. In the future, these correlates may be used as predictors of those patients most likely to respond to ECT, for early indicators of clinical response, or for relapse prevention.
The collaboration with MGH will serve to establish the LiveNet system's capabilities for engaging in significant long-term ambulatory clinical studies. The implications and clinical significance of the proposed research are broad. The development and refinement of a methodology that objectively and accurately monitors treatment response in major depression has implications for the diagnosis, treatment, and relapse prevention. Once a reliable index of physiologic and behavioral metrics for depression has been established, other environments outside of the inpatient setting become potential targets for assessment. The methodology developed also has the potential to help in the assessment, early diagnosis, and treatment prediction of other severe psychopathologies that have likely physiologic correlates and involve difficulty with social interactions. These include communication disorders and pervasive developmental disorders in children as well as the pre-clinical assessment of severe psychotic, mood, anxiety, and personality disorders. As our understanding of the central nervous system control of autonomic arousal improves and the neurobiology of depression continues to be discovered, future questions about the subtypes of depression and endophenotyping for genetic studies of depression can be studied in the ambulatory setting. This will lead to more sophisticated neurobiologic models of the mechanism of healing and ultimately to increased efficiency and efficacy of treatment.
Quantifying Social Engagement
Social interaction is a complex and ubiquitous human behavior involving attitudes, emotions, nonverbal and verbal cues, and cognitive function. Importantly, impairment in social function is a hallmark for nearly every diagnostic category of mental illness including mood and anxiety disorders as well as dementia, schizophrenia, and substance abuse [24]. In addition, social isolation can be a significant stress for patients undergoing rehabilitation from surgical and medical procedures and illnesses. Thus, an important challenge for our behavior modeling technology is to build computational models that can be used to predict the dynamics of individuals and their social interactions.
Using LiveNet, we can collect data about daily interactions with family, friends, and strangers and quantify information such as how frequent are the interactions, the dynamics of the interactions, and the characteristics of such interactions using simple infrared (IR) sensors and IR tags to identify individuals. Using simple voice features (such as talking/non-talking, voice patterns, and interactive speech dynamics measures) derived from microphones, we can obtain a variety of useful social interaction statistics. We can even model an individual's social network and how that network changes over time by analyzing statistical patterns of these networks as they evolve [25]. Data on social function can be used as both a marker of improvement or rehabilitation progress or as an indicator of relapse and for use in relapse prevention.
Long-Term Behavior Modeling and Trending
The LiveNet platform also lends itself naturally to be able to do a wide variety of long-term healthcare monitoring applications for physiological and behavioral trends that vary slowly with time by using the currently available physiological sensors. This has important implications for rehabilitation medicine. The ambulatory physiological and contextual sensing and the health classifiers discussed in Section 3.1 can be combined together in a hierarchical manner to develop time-dependent models of human behavior at longer timescales. Current systems are purely reactive (e.g. sounding an alarm after a person has a heart attack or falls down), and are dependent on classifying and determining in real-time when certain events have occurred.
While this type of application is very useful and potentially-life saving, these systems typically do not have any sense of the history of an individual and can only react to instantaneous events. By combining long-term trending with multimodal analysis, it is possible to develop more proactive systems and personalized data that can be used to catch problems before they manifest themselves (e.g. instead of reacting to a heart attack, one can predict beforehand that a heart attack is imminent). However, a proactive system requires more resources, as it must have context-aware and inference capabilities to be able to determine what the right information is to be directed at the right people, to the right places, at the right times, and for the right reasons. While challenging, small advances have been made in this regard.
In order to fully accomplish the goal of preventive monitoring, large databases in living situations are needed. The LiveNet system provides a convenient infrastructure to implement and rapidly prototype new proactive healthcare applications in this domain. It is very important from the proactive healthcare point of view that these individual classification systems also be able to determine trends in physiological/contextual state over time to provide not only immediate diagnostic power but also prognostic insight. We are collaborating on the MIT/TIAX PlaceLab, a cross-institutional research smart living environment [26], to provide a very robust infrastructure to be able to collect and study long-term health information in conjunction with data collected by LiveNet systems. We also have a collaboration with British Telecom to use LiveNet technology in similar long-term naturalistic home monitoring applications for eldercare.
The information collected from the multimodal sensors can then be used to construct activities of daily living, important information in being able to profile a person's healthy living style. Furthermore, these activities of daily living can initiate action on the part of the wearable PDA. Examples include experience sampling, a technique to gather information on daily activity by point of querying (which can be set to trigger based on movement or other sensed context by the PDA). The system can also proactively suggest alternative healthy actions at the moment of decision, where it has been demonstrated as being more effective at eliciting healthy behavior [27].
Real-Time Multimodal Feedback Systems in Rehabilitation
An obvious domain for LiveNet is in physiology monitoring with real-time feedback and classification. The dominant healthcare paradigm that exists is to stream physiology data from an individual to a centralized server, where the higher-power processing and data visualization could be performed to post-process the data. The wearable system served mainly as a data acquisition vehicle, with little feedback or interaction capabilities. Now, it is possible for significant localized processing as well as displaying the result, which will open up the door to real-time interactive health applications. In fact, commercial systems are just beginning to incorporate these types of increased functionality.
It is possible to use a system such as LiveNet to go a step farther and demonstrate that mobile systems are capable of significant local processing for real-time feature extraction and context classification as well as provide the distributed wireless infrastructure for streaming information between systems, all on commodity hardware that is commonplace and available today. This will enable the real-time classification of medical conditions without the need for other infrastructure, available wherever the individual goes. The distributed nature of LiveNet can also allow systems to stream raw physiology or its combination with derived metadata/context very easily to any specified source(s), whether it is other mobile systems, data servers, or output displays such as projections.
By providing local processing capabilities, the time that is required to receive feedback for relevant health events is dramatically reduced. Historically, the time delay required to receive feedback can potentially take weeks, and be both problematic because of the iterative nature of determining the optimized treatment path. For people who are on medication or embarking on a prolonged rehabilitation schedule, for example, this delay in the feedback loop is particularly onerous. A doctor will recommend a dosage and medication regimen to try out, and the person goes home and tries the medication schedule for a while. If the person does not respond favorably to this drug schedule, they have to reschedule an appointment with the doctor, go in, and potentially take more tests, before getting a recommendation on a new schedule (such is the case for people with thyroid conditions, for example). This same scenario is also true with lengthy rehabilitation programs such as cardiac rehabilitation. This results in a very time consuming process as well as a significant drain on healthcare resources. In addition to the fact that the feedback loop can be very long in duration, the doctor is literally in the dark about the efficacy of the treatment, and so an iterative trial-and-error process is required.
Using the LiveNet system, it is possible to effectively reduce the time delay to process and receive health feedback. This is particularly true when the doctor can be either removed from the equation or visit length and frequency can be reduced, such as with real-time diagnostic systems that can provide effectively instantaneous classification on health state and context. Given that medication compliance is a major healthcare issue, especially among the elderly, with estimated costs of upwards of $100 billion annually [30], systems that can help remind and support compliance with appropriate feedback will help to promote healthy, preventive behavior.
Also, potential advances in more personalized medication and rehabilitation scheduling can be improved based on measured, quantitative physiological symptoms and behavioral responses, not based only on time scheduled approximations as the current practice. The effects of medication and rehabilitation treatment can be logged and recorded quantitatively and compared to changes in physiology, eventually with the goal of developing a real-time monitoring and drug delivery system. Future research will extend this work by developing real-time, closed-loop systems that can track the effects of individualized treatment over time.
Time-stamping of relevant events (either simple events or elaborated notes) in order to correlate these events with accurate continuous physiology and behavior data in an ambulatory setting also offers great potential. From this health information, real-time correlations to specific medical conditions as well as predictions of adverse outcomes can be made. This also has a potential impact in the research of physiology in the domain of clinical medication and intervention research, providing a streamlined path for Electronic Data Capture (EDC), where accurate reporting is an issue and human transcription errors and recall bias from surveys can be reduced. Potential benefits provided by continuous monitoring include automated and real-time data capture from patients for accurate reporting, feedback and notification for enforcing medication compliance in patients, assessment of the degree of medication compliance, removal of human error inherent in manual transcription/data entry, high-resolution time-stamping for accurate temporal characterization of events, and the ability to accurately correlate quantitative physiologic data to events for diagnosis and characterization
Conclusion
The ability to provide new wearable technology for medical and surgical rehabilitation services is emerging as an important option for clinicians and patients. Wearable technology provides a convenient platform to be able to quantify the long-term context and physiological response of individuals. This, in turn, will support the development of individualized treatment systems with real-time feedback to help promote proper behavior. The ultimate goal of the research is to eventually be able to use LiveNet for developing practical monitoring systems and therapeutic interventions in ambulatory, long-term use environments.
Figure 1 LiveNet wearable performing real-time FFT analysis and activity classification on accelerometer data, visualizing the results, as well as wirelessly streaming real-time ECG/GSR/temperature and classification results to a remote computer with a projection display as well as peer LiveNet systems.
Figure 2 LiveNet system, composed of the Zaurus PDA (top left), with SAK2 data acquisition/sensor hub and BioSense physiological sensing board (middle), battery source (top right), sensor bus hub (lower right), 3D accelerometer board (middle left), and WMSAD multisensor board (lower left).
Figure 3 LiveNet wearable configured for non-invasive real-time soldier physiology monitoring.
Figure 4 LiveNet wearable streaming real-time ECG/motion/stress information to a remote display over a wireless network link.
==== Refs
Tavormina C Embracing the eden alternative in long-term care environments Geriatr Nurs 1999 20 158 161 10661107
Abowd G Improving outlooks for aging baby boomers Testimony before the Senate on Special Aging, 107th Congress 2003
LifeShirt, Vivometrics Corporation
SenseWear, BodyMedia Corporation
Mini Logger, Minimitter Corporation
Holter NJ Generelli JA Remote recording of physiologic data by radio Rocky Mountain Med Journal 1949 747 751
Bashshur R Sanders J Shannon G Telemedicine: Theory and Practice 1997 Springfield: Charles C. Thomas
Food and Drug Administration Medical devices: cardiovascular devices: reclassification of the arrhythmia detector and alarm. final rule Fed Regist 2003
Kirk V Bohn S Flemons W Remmers J "Comparison of Home Oximetry Monitoring with Laboratory Polysomnography in Children" Department of Pediatrics, University of Calgary, Canada 2003
DeVaul R Sung M Gips J Pentland A MIThril 2003: applications and architecture IEEE International Symposium of Wearable Computing 2003
Elledge C Wearable multiple sensor acquisition board MEng Thesis MIT, Dept Media Art and Sciences 2003
MIThril Sensors 2003
Tapia E Marmasse N Intille S Larson K MITes: Wireless portable sensors for studying behavior House_n Technical Report 2004
M. Laibowitz, Uber-Badge
Sung M Non-invasive activity classification using single accelerometer Technical Report, MIT Media Lab 2004
Basu S Conversational scene analysis PhD Thesis 2002 MIT, Dept. of Electrical Engineering and Computer Science
The M.I.T. Wearable Computing Group webpage
Sung M Shivering motion/hypothermia classification for wearable soldier health monitoring systems 2003 IEEE International Symposium of Wearable Computers
Weaver J A wearable health monitor to aid parkinson disease treatment MS Thesis 2003 MIT, Dept. Media Arts and Sciences
Klapper D Use of a wearable ambulatory monitor in the classification of movement states in parkinson's disease MS Thesis 2003 Harvard-MIT, Health Sciences and Technology Program
Rochester Center for Future Health
Bao L Physical activity recognition from acceleration data under semi-naturalistic conditions MEng Thesis 2003 MIT, Dept. Electrical Engineering and Computer Science
Van Laerhoven K Schmidt A Gellersen H Multi-sensor context aware clothing IEEE International Symposium on Wearable Computers 2002
DSM-IV, T.F.o.: Diagnostic and Statistical Manual of Mental Disorders 1994 Washington, D.C.: American Psychiatric Society 70
Eagle N Pentland A Social network computing Conference on Ubiquitous Computing (UbiComp) 2003
MIT/TIAX PlaceLab
Intille S Kukla C Ma X Eliciting user preferences using image-based experience sampling and reflection Conference on Human Factors and Computing Systems 2002
Kessler RC Berglund P Demler O Jin R Koretz D Merikangas KR Rush AJ Walters EE Wang PS The epidemiology of major depressive disorder: results from the national comorbidity survey replication (NCS-R) J Am Med Assoc 2003 289 3095 3105 10.1001/jama.289.23.3095
Mental Health A Report of the Surgeon General 1999 U.S. Department of Health and Human Services
Wertheimer A Santella T Medication compliance research: still so far to go Journal of Applied Research in Clinical and Experimental Therapeutics 2003 3
Kumar V Pentland A DiaBetNet: learning and predicting for better health and compliance Diabetics Technology Society meeting Foster City, CA Oct 31-Nov 2
Marmasse N Providing lightweight telepresence in mobile communication to enhancecollaborative living PhD Thesis 2004 MIT, Media Arts and Sciences
Fogarty J Connecting versus calming: interruptibility, presence, and availability Position paper, Workshop on Forecasting Presence and Availability, CHI 2004
Gerasimov V Every sign of life Doctoral dissertation 2003 MIT, Dept. Media Arts and Sciences
|
15987514
|
PMC1198249
|
CC BY
|
2021-01-04 16:37:40
|
no
|
J Neuroengineering Rehabil. 2005 Jun 29; 2:17
|
utf-8
|
J Neuroeng Rehabil
| 2,005 |
10.1186/1743-0003-2-17
|
oa_comm
|
==== Front
Lipids Health DisLipids in Health and Disease1476-511XBioMed Central London 1476-511X-4-161609822510.1186/1476-511X-4-16ResearchNPC1L1 haplotype is associated with inter-individual variation in plasma low-density lipoprotein response to ezetimibe Hegele Robert A [email protected] Justin [email protected] Matthew R [email protected] Jian [email protected] Vascular Biology Group, Robarts Research Institute, London, Ontario, Canada2 Department of Medicine, Schulich School of Medicine and Dentistry, University of Western Ontario, London, Ontario, Canada2005 12 8 2005 4 16 16 25 7 2005 12 8 2005 Copyright © 2005 Hegele et al; licensee BioMed Central Ltd.2005Hegele et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
NPC1L1 encodes a putative intestinal sterol transporter which is the likely target for ezetimibe, a new type of lipid-lowering medication. We previously reported rare non-synonymous mutations in NPC1L1 in an individual who had no plasma lipoprotein response to ezetimibe. We next hypothesized that common variants in NPC1L1 would underlie less extreme inter-individual variations in the plasma LDL cholesterol response to ezetimibe.
Results
In 101 dyslipidemic subjects, we found that NPC1L1 haplotype was significantly associated with inter-individual variation in the response of plasma LDL cholesterol to treatment with ezetimibe for 12 weeks. Specifically, about one subject in eight lacked the common NPC1L1 haplotype 1735C-25342A-27677T and these subjects had a significantly greater reduction in plasma LDL cholesterol with ezetimibe than subjects with at least one copy of this haplotype (-35.9+4.0 versus -23.6+1.6 percent reduction, P = 0.0054). This was paralleled by a similar non-significant trend of between-haplotype difference in reduction of total cholesterol.
Conclusion
These preliminary pharmacogenetic results suggest that NPC1L1 variation is associated with inter-individual variation in response to ezetimibe treatment.
==== Body
Background
Ezetimibe, the first member of a new class of medications, primarily reduces plasma concentration of low-density lipoprotein (LDL) cholesterol by blocking sterol absorption in enterocytes [1]. Ezetimibe probably interferes with the normal function of the NPC1L1 gene product, which appears to govern sterol absorption in the small intestine [2-5]. The mean plasma LDL cholesterol reduction seen with ezetimibe is 20 to 25%, and this has been remarkably consistent across patient subgroups defined by age, gender, ethnic background and concomitant use of other lipid regulating agents, such as statin drugs [6-9]. But despite the concordance in mean reductions, there is a wide range of inter-individual variation in the LDL cholesterol response to ezetimibe. A possible genetic basis for this inter-individual variation was suggested by our previous observation of rare non-synonymous NPC1L1 mutations in a non-responder to ezetimibe [10]. During the course of those studies, we identified several single nucleotide polymorphisms (SNPs) in NPC1L1 [10]. These SNPs have enabled assessment of common genetic variation at NPC1L1, which we hypothesized would underlie less extreme inter-individual variations in the plasma LDL cholesterol response to ezetimibe.
Results
Clinical and demographic data
Descriptive baseline clinical features are shown in Table 1. All subjects had baseline and follow-up fasting lipoprotein profiles measured. Mean follow-up was 84 days (12 weeks). None of the 101 subjects withdrew from treatment and compliance was excellent as judged by tablet counting. No adverse events were reported. Mean responses to ezetimibe treatment are shown in Table 2. The range of LDL cholesterol responses to ezetimibe treatment are shown in Figure 1. There was no difference in mean LDL cholesterol change in female versus male subjects or in subjects on ezetimibe monotherapy versus subjects on ezetimibe in combination with statin treatment.
Table 1 Baseline (mean ± standard deviation) and on-treatment clinical and biochemical attributes
overall males females
number 101 61 40
age (years) 55.6 ± 11.9 54.0 ± 10.7 58.0 ± 13.3
days on treatment 83.8 ± 60.6 90.1 ± 67.6 74.5 ± 47.9
percent on statin 69.3 78.7 55.0
plasma lipids and lipoproteins (mmol/L)
baseline
cholesterol
- total 6.57 ± 1.43 6.33 ± 1.54 6.93 ± 1.18
- LDL 4.51 ± 1.40 4.35 ± 1.53 4.72 ± 1.18
- HDL 1.18 ± 0.30 1.15 ± 0.29 1.22 ± 0.31
triglycerides 2.50 ± 2.36 2.63 ± 2.86 2.32 ± 1.33
on-treatment
cholesterol
- total 5.41 ± 1.32 5.10 ± 1.21 5.88 ± 1.35
- LDL 3.35 ± 1.15 3.21 ± 1.11 3.56 ± 1.19
- HDL 1.23 ± 0.45 1.16 ± 0.33 1.33 ± 0.60
triglycerides 2.25 ± 1.91 2.24 ± 2.12 2.25 ± 1.57
abbreviations: LDL, low-density lipoprotein; HDL, high-density lipoprotein;
Table 2 NPC1L1 genotype frequencies
SNP genotype number frequency
1735C>G C/C 56 0.55
C/G 40 0.40
G/G 5 0.05
25342A>C A/A 54 0.53
A/C 36 0.36
C/C 11 0.11
27677T>C T/T 66 0.65
T/C 30 0.30
C/C 5 0.05
abbreviation: SNP, single nucleotide polymorphism
Figure 1 Individual LDL-cholesterol response to ezetimibe 10 mg. Each bar represents the percent change in LDL-cholesterol from baseline for one study subject; these data are arranged in rank order to show the range of responses.
Genetic descriptors of study sample
NPC1L1 genotype frequencies are shown in Table 2. Allele frequencies were: 1) 0.75 and 0.25 for 1735C and 1735G, respectively; 2) 0.72 and 0.28 for 25342A and 25342C, respectively; and 3) 0.80 and 0.20 for 27677T and 27677C, respectively. Pairwise linkage disequilibrium correlation coefficients for SNP pairs 1735C>G:25342A>C, 1735C>G:27677T>C and 25342A>C:27677T>C were 0.24 (P = 0.017), 0.30 (P < 0.0001) and 0.44 (P < 0.0001), respectively. Thus, there was moderate but not strong linkage disequilibrium between these pairs of SNPs. Maximal likelihood haplotype definitions and frequencies are shown in Table 3. The most common haplotype (frequency ~0.6) was defined as 1735C-25342A-27677T and designated as "haplotype 2". For ANOVA, NPC1L1 haplotypes were collapsed to three groups, based on the presence or absence of haplotype 2. Thus, participants in this study had one of three possible diploid summary haplotypes: 2/2, 2/X or X/X, where X refers to any non-2 haplotype. There were 37, 51 and 13 subjects with haplotypes 2/2, 2/X and X/X, respectively.
Table 3 NPC1L1 haplotype definition and frequencies
designation sequence definition frequency
1 1735G-25342A-27677T 0.089
2 1735C-25342A-27677T 0.619
3 1735G-25342A-27677C 0.010
4 1735C-25342A-27677C 0.005
5 1735G-25342C-27677T 0.025
6 1735C-25342C-27677T 0.069
7 1735G-25342C-27677C 0.119
8 1735C-25342C-27677C 0.064
Genetic associations with plasma lipoproteins
ANOVA (Table 4) showed no significant differences between baseline plasma lipoproteins. However, the percent change in LDL cholesterol on ezetimibe from baseline was significantly different between haplotypes (P = 0.02) and the percent change in total cholesterol from baseline on ezetimibe tended to be different between haplotypes. The significant between-haplotype differences from ANOVA were further explored assuming a dominant model for presence of haplotype 2, in which subjects with haplotypes 2/2 or 2/X were scored as "A" and subjects with haplotype X/X were scored as "B". In this model, subjects with haplotype X/X had a significantly greater percent reduction in plasma LDL cholesterol from baseline on ezetimibe than did subjects with haplotypes 2/2 or 2/X (-35.9 ± 4.0 versus -23.6 ± 1.6 percent, P = 0.0054). The range of reduction in LDL cholesterol in subjects with haplotype X/X was -68.2 to -19.6 percent. Furthermore, subjects with haplotype X/X tended to have a greater percent reduction in plasma total cholesterol from baseline on ezetimibe than did subjects with haplotypes 2/2 or 2/X (-22.8 ± 3.4 versus -15.9 ± 1.3 percent, P = 0.058).
Table 4 Clinical and biochemical features according to NPC1L1 haplotype
haplotype
2/2 2/X X/X P-value
number 37 51 13
age (years) 56.6 ± 13.3 53.9 ± 11.3 58.4 ± 6.9 NS
percent female 48.6 37.3 38.5 NS
percent on statin 65.8 70.6 69.8 NS
baseline plasma lipids and lipoproteins (mmol/L)
cholesterol
- total 6.88 ± 1.33 6.31 ± 1.59 6.73 ± 1.01 NS
- LDL 4.64 ± 1.39 4.41 ± 1.51 4.49 ± 1.10 NS
- HDL 1.16 ± 0.34 1.19 ± 0.29 1.20 ± 0.21 NS
triglycerides 2.58 ± 1.75 1.99 ± 1.37 3.47 ± 3.78 NS
percent change on-treatment
cholesterol
- total -16.5 ± 12.2 -15.8 ± 13.0 -22.8 ± 8.4 NS (0.07)
- LDL -23.6 ± 13.2 -23.6 ± 14.7 -35.2 ± 13.5 0.02
- HDL +10.7 ± 43.0 +1.2 ± 14.3 +0.6 ± 17.0 NS
triglycerides -4.2 ± 35.6 +2.9 ± 44.9 -0 ± 30.3 NS
abbreviations: as in Table 1.
An additional post hoc analysis of individual SNPs found that the 11 homozygotes for the 25342C allele had a significantly greater percent reduction in plasma LDL cholesterol from baseline on ezetimibe than did other subjects (P = 0.02). Furthermore, the five homozygotes for the 27677C allele had a significantly greater percent reduction in plasma LDL cholesterol from baseline on ezetimibe than did other subjects (P = 0.013).
Discussion
In this very preliminary analysis of a small sample of subjects with hypercholesterolemia, we found that genetic variation in NPC1L1, as defined by a three-site SNP haplotype, was significantly associated with inter-individual variation in the response of plasma LDL cholesterol to 12 weeks of treatment with ezetimibe 10 mg daily. Specifically, about one subject in eight did not carry the common NPC1L1 haplotype 1735C-25342A-27677T (designated "haplotype 2"); these subjects were found to have a significantly greater reduction in plasma LDL cholesterol with ezetimibe than subjects with at least one copy of haplotype 2 (-35.9 ± 4.0 versus -23.6 ± 1.6 percent reduction, P = 0.0054). This was paralleled by a similar non-significant trend of between-haplotype differences in reduction of total cholesterol. There were no significant between-haplotype differences in ezetimibe-related changes in plasma triglycerides or HDL cholesterol.
As with many association studies, the present study has limitations [11], including: 1) a small sample size; 2) no replication sample; 3) no demonstrated functional consequences of the NPC1L1 SNPs or haplotype; 4) no intermediate phenotype, such as cholesterol absorption; and 5) the potential that the positive findings were related to linkage disequilibrium with unmeasured markers at or near the NPC1L1 locus. Also, we did not genotype all SNPs at this locus. However, previous genomic screening experiments [10] indicated that the remaining SNPs were rare, and thus less likely to add information to the three SNPs studied. Inclusion of these rare SNP genotypes in the extended haplotype would have further subdivided the data into very small-sized cells for statistical analysis.
Our study confirms the similarity of the mean LDL cholesterol response to ezetimibe, namely a 20 to 25% reduction, in various study samples and across a range of demographic features including sex and concomitant statin treatment (6–9). Figure 1 demonstrates that this consistent mean LDL cholesterol reduction occurs on the background of relatively wide inter-individual variation in response. Our findings further indicate that subjects who carry the most common NPC1L1 haplotype (namely haplotype 2), have an ezetimibe-related LDL cholesterol reduction that is within the expected 20 to 25% range. However, a small but substantial group of subjects without haplotype 2 experienced a significantly greater LDL cholesterol reduction, on the order of 35%.
Conclusion
The current finding of NPC1L1-associated inter-individual differences in LDL-cholesterol response to ezetimibe together with our earlier demonstration that rare missense mutations in NPC1L1 are associated with non-response to ezetimibe [10] support a relationship between this gene product and the mechanism of action of ezetimibe. Clearly, additional mechanistic and genetic studies are required. But these pharmacogenetic results, if confirmed, are consistent with the idea that the NPC1L1 is the ezetimibe target.
Methods
Study subjects
Between December 2003 and May 2004, 101 patients with primary hypercholesterolemia (defined as elevated LDL cholesterol) were treated with ezetimibe 10 mg daily according to national dyslipidemia guidelines [12]. Basic demographic attributes of study subjects are shown in Table 1. About one-third of patients were not taking any lipid-lowering medication and the remainder were stable on statin treatment of ≥12 weeks' duration prior to initiation of ezetimibe. Concomitant statin treatment included atorvastatin, rosuvastatin and simvastatin in 30, 28 and 12 subjects, respectively. All subjects provided informed consent and the study was approved by the Ethics Review panel of the University of Western Ontario (review number 07920E).
Biochemical and genetic analyses
The lipoprotein profile after a 12 hour fasting period was determined before initiation of ezetimibe treatment and again after a mean follow-up of 12 weeks. Lipoprotein determinations were performed according to the Ontario Lipoprotein Proficiency Program standards and LDL cholesterol was calculated using the Friedewald-Levy-Fredrickson formula [13].
Genomic DNA was extracted and three common informative NPC1L1 SNPs from across the coding sequence were chosen for genotyping. Allele-specific genotyping methods were used [10]. For the genotype of exon 2 SNP 1735C>G (trivial name L272L, dbSNP number 2072183), we amplified a 381 bp fragment containing exon 2 using primers 5' GCT CAA CTT CCA GGG AGA CA and 5' AGC TTG TCA GAG AGG CTG G. This was followed by treatment with shrimp alkaline phosphatase (SAP) and ExoI to remove primers and unincorporated dNTPs, followed by ddNTP extension (SnaPShot, PE Applied Biosystems, Mississauga, ON) with primer 5' ATA GGC ATG AGC CAC TGC AC and analysis on a 3730 DNA Sequencer (PE Applied Biosystems, Mississauga, ON). For the genotype of intron 18 SNP 25342A>C, we amplified a 766 bp fragment containing intron 18 using primers 5' GCC CAG GTA GAA GGT GGA GTC and 5' CGT TGT TTG AGA CAT ACA TAG CTG. This was followed by treatment with SAP and ExoI to remove primers and unincorporated dNTPs, gel purification and ddNTP extension with primer 5' CTG CCT GAC ACC TGG CTC TGA and fragment analysis. For the genotype of exon 20 SNP 27677T>C (trivial name V1296V), we amplified the 558 bp fragment containing exon 20 using primers 5' GAA GCT TGG GCT GTG AAC A and 5' CCA CTA TGG GAG CAG AGG AG. This was followed by treatment with SAP and ExoI to remove primers and unincorporated dNTPs, gel purification and ddNTP extension (SnaPShot, PE Applied Biosystems, Mississauga, ON) with primer 5' TCT CTC CGC AGG GCC TGA CGT, and fragment analysis.
Statistical analysis
Analyses were performed using SAS version 8.2 (Cary, NC), with a nominal level of significance defined as P < 0.05. Significance of the deviation of SNP genotype frequencies from Hardy-Weinberg equilibrium was assessed using chi-square analysis. Pairwise linkage disequilibrium between NPC1L1 alleles was determined using correlation coefficients as described [14]. Three-site maximal likelihood haplotypes were constructed using PHASE version 2.0 [15]. Analysis of variance (ANOVA) was used to identify significant sources of variation for quantitative plasma phenotypes, using F-values computed from type III sums of squares, which is most appropriate for unbalanced study designs. The dependent variables in ANOVA were percent change from baseline of plasma total, LDL and high-density lipoprotein (HDL) cholesterol and triglycerides. The independent variable in ANOVA was NPC1L1 haplotype, with age and sex as covariates within each model.
Authors' contributions
Robert A. Hegele: experimental design, manuscript preparation
Justin Guy: data generation and interpretation, manuscript approval
Matthew R. Ban: data analysis, manuscript approval
Jian Wang: data analysis, editing, manuscript approval
Acknowledgements
Supported by the Jacob J. Wolfe Distinguished Medical Research Chair, the Edith Schulich Vinet Canada Research Chair (Tier I) in Human Genetics, a Career Investigator award from the Heart and Stroke Foundation of Ontario, and operating grants from the Canadian Institutes for Health Research, the Heart and Stroke Foundation of Ontario and the Ontario Research and Development Challenge Fund (Project #0507).
==== Refs
Davidson MH Ezetimibe: a novel option for lowering cholesterol Expert Rev Cardiovasc Ther 2003 1 11 21 15030293 10.1586/14779072.1.1.11
Altmann SW Davis HR JrZhu LJ Yao X Hoos LM Tetzloff G Iyer SP Maguire M Golovko A Zeng M Wang L Murgolo N Graziano MP Niemann-Pick C1 Like 1 protein is critical for intestinal cholesterol absorption Science 2004 303 1201 1204 14976318 10.1126/science.1093131
Davis HR Zhu LJ Hoos LM Tetzloff G Maguire M Liu J Yao X Iyer SP Lam MH Lund EG Detmers PA Graziano MP Altmann SW Niemann-pick C1 Like 1 (NPC1L1) is the intestinal phytosterol and cholesterol transporter and a key modulator of whole body cholesterol homeostasis J Biol Chem 2004 279 33586 33592 15173162 10.1074/jbc.M405817200
Davies JP Scott C Oishi K Liapis A Ioannou YA Inactivation of NPC1L1 causes multiple lipid transport defects and protects against diet-induced hypercholesterolemia J Biol Chem 2005 280 12710 12720 15671032 10.1074/jbc.M409110200
Iyer SP Yao X Crona JH Hoos LM Tetzloff G Davis HR JrGraziano MP Altmann SW Characterization of the putative native and recombinant rat sterol transporter Niemann-Pick C1 Like 1 (NPC1L1) protein Biochim Biophys Acta 2005 1722 282 292 15777641
Knopp RH Dujovne CA Le Beaut A Lipka LJ Suresh R Veltri EP Ezetimbe Study Group Evaluation of the efficacy, safety, and tolerability of ezetimibe in primary hypercholesterolaemia: a pooled analysis from two controlled phase III clinical studies Int J Clin Pract 2003 57 363 368 12846338
Davidson MH Ballantyne CM Kerzner B Melani L Sager PT Lipka L Strony J Suresh R Veltri E for the Ezetimibe Study Group Efficacy and safety of ezetimibe coadministered with statins: randomised, placebo-controlled, blinded experience in 2382 patients with primary hypercholesterolemia Int J Clin Pract 2004 58 746 55 15372846 10.1111/j.1368-5031.2004.00289.x
Stein E Stender S Mata P Sager P Ponsonnet D Melani L Lipka L Suresh R Maccubbin D Veltri E for Ezetimibe Study Group Achieving lipoprotein goals in patients at high risk with severe hypercholesterolemia: efficacy and safety of ezetimibe co-administered with atorvastatin Am Heart J 2004 148 447 55 15389231 10.1016/j.ahj.2004.03.052
Pearson TA Denke MA McBride PE Battisti WP Brady WE Palmisano J A community-based, randomized trial of ezetimibe added to statin therapy to attain NCEP ATP III goals for LDL cholesterol in hypercholesterolemic patients: the ezetimibe add-on to statin for effectiveness (EASE) trial Mayo Clin Proc 2005 80 587 595 15887425
Wang J Williams CM Hegele RA Compound heterozygosity for two non-synonymous polymorphisms in NPC1L1 in a non-responder to ezetimibe Clin Genet 2005 67 175 177 15679830 10.1111/j.1399-0004.2004.00388.x
Hegele RA SNP judgments and freedom of association Arterioscler Thromb Vasc Biol 2002 22 1058 1061 12117714 10.1161/01.ATV.0000026801.56080.14
Genest J Frohlich J Fodor G McPherson R Recommendations for the management of dyslipidemia and the prevention of cardiovascular disease: summary of the 2003 update Can Med Assoc J 2003 169 921 924 14581310
Roberts WC The Friedewald-Levy-Fredrickson formula for calculating low-density lipoprotein cholesterol, the basis for lipid-lowering therapy Am J Cardiol 1988 62 345 356 3165237 10.1016/0002-9149(88)90248-2
Hegele RA Plaetke R Lalouel JM Linkage disequilibrium between DNA markers at the low-density lipoprotein receptor gene Genet Epidemiol 1990 7 69 81 1970320 10.1002/gepi.1370070114
Stephens M Donnelly P A comparison of Bayesian methods for haplotype reconstruction from population genotype data Am J Hum Genet 2003 73 1162 1169 14574645 10.1086/379378
|
16098225
|
PMC1198250
|
CC BY
|
2021-01-04 16:39:18
|
no
|
Lipids Health Dis. 2005 Aug 12; 4:16
|
utf-8
|
Lipids Health Dis
| 2,005 |
10.1186/1476-511X-4-16
|
oa_comm
|
==== Front
Mol CancerMolecular Cancer1476-4598BioMed Central London 1476-4598-4-221600198210.1186/1476-4598-4-22ResearchHistone acetylation and histone deacetylase activity of magnesium valproate in tumor and peripheral blood of patients with cervical cancer. A phase I study Chavez-Blanco Alma [email protected] Blanca [email protected] Enrique [email protected] Lucia [email protected] Lucely [email protected] Myrna [email protected] David [email protected] Aurora [email protected] Patricia [email protected] Pilar [email protected] Carlos [email protected] Gustavo [email protected] Catalina [email protected] Enrique [email protected] Alfonso [email protected] Unidad de Investigacion Biomedica en Cáncer, Instituto Nacional de Cancerología/Instituto de Investigaciones Biomédicas (INCan/IIB), Universidad Nacional Autonoma de Mexico (UNAM), Mexico City. Mexico2 Division of Clinical Research, Instituto Nacional de Cancerología (INCan), Mexico City, Mexico3 Laboratorio de Química Medicinal FES-Cuautitlán, UNAM, Mexico2005 7 7 2005 4 22 22 9 5 2005 7 7 2005 Copyright © 2005 Chavez-Blanco et al; licensee BioMed Central Ltd.2005Chavez-Blanco et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The development of cancer has been associated with epigenetic alterations such as aberrant histone deacetylase (HDAC) activity. It was recently reported that valproic acid is an effective inhibitor of histone deacetylases and as such induces tumor cell differentiation, apoptosis, or growth arrest.
Methods
Twelve newly diagnosed patients with cervical cancer were treated with magnesium valproate after a baseline tumor biopsy and blood sampling at the following dose levels (four patients each): 20 mg/kg; 30 mg/kg, or 40 mg/kg for 5 days via oral route. At day 6, tumor and blood sampling were repeated and the study protocol ended. Tumor acetylation of H3 and H4 histones and HDAC activity were evaluated by Western blot and colorimetric HDAC assay respectively. Blood levels of valproic acid were determined at day 6 once the steady-state was reached. Toxicity of treatment was evaluated at the end of study period.
Results
All patients completed the study medication. Mean daily dose for all patients was 1,890 mg. Corresponding means for the doses 20-, 30-, and 40-mg/kg were 1245, 2000, and 2425 mg, respectively. Depressed level of consciousness grade 2 was registered in nine patients. Ten patients were evaluated for H3 and H4 acetylation and HDAC activity. After treatment, we observed hyperacetylation of H3 and H4 in the tumors of nine and seven patients, respectively, whereas six patients demonstrated hyperacetylation of both histones. Serum levels of valproic acid ranged from 73.6–170.49 μg/mL. Tumor deacetylase activity decreased in eight patients (80%), whereas two had either no change or a mild increase. There was a statistically significant difference between pre and post-treatment values of HDAC activity (mean, 0.36 vs. 0.21, two-tailed t test p < 0.0264). There was no correlation between H3 and H4 tumor hyperacetylation with serum levels of valproic acid.
Conclusion
Magnesium valproate at a dose between 20 and 40 mg/kg inhibits deacetylase activity and hyperacetylates histones in tumor tissues.
==== Body
Background
The development of cancer has been associated with epigenetic alterations such as deregulation of DNA methylation and aberrant histone deacetylase (HDAC) activity [1]. These epigenetic phenomena co-participate in regulation of gene transcription: for instance, histone deacetylases which deacetylate histone core tails are recruited by DNA methyltransferases and methyl-binding proteins, which in turn lead to tighter chromatin packaging, reducing access of transcriptional factors to DNA [2,3].
HDACs are seen as a potential target for cancer treatment. HDAC inhibition has been reported to induce tumor cell differentiation, apoptosis, or growth arrest, depending on the experimental system [4-7]. Previous studies have shown that histone acetylation can increase the efficiency of several anticancer drugs targeting the DNA [8-11]; studies also shown that HDAC activity inhibitors enhance in vitro sensitivity of tumor cells to radiation [12-14]. However, a significant impediment in targeting HDAC has been lack of a clinically applicable HDAC inhibitor.
Valproic acid, an 8-carbon, branched-chained fatty acid, is a well-known and effective antiepileptic drug [15]. Its pharmacologic effects involve a variety of mechanisms including increased gamma-amino butyric acid (GABA)-ergic transmission, reduced release and/or effects of excitatory amino acids, blockade of voltage-gated sodium channels, and modulation of dopaminergic and serotoninergic transmission [16]. Because of its effectiveness, good tolerability, and oral bioavailability, valproic acid is widely used as a chronic anti-convulsant therapy.
It was recently reported that valproic acid is an effective inhibitor of histone deacetylases at concentrations well within the therapeutic range used for epilepsy. As such, valproic acid relieves repression of transcription factors that recruit histone deacetylases and activates transcription from diverse promoters. Valproic acid causes hyperacetylation of the N-terminal tails of histones H3 and H4 in vitro and in vivo. It inhibits HDAC activity, most probably by binding to the catalytic center and thereby blocking substrate access [17,18].
In contrast to other HDAC inhibitors, valproic acid has a good tolerability and safety profile as demonstrated by 35 years of use as a chronic therapy for epileptic disorders. It has a serum half-life of 9–18 h and is administered orally [15,16]; however, the dose needed for observing its HDAC inhibitory activity in tumors of cancer patients is yet unknown. On these basis, we performed a phase I study to find the biologically "adequate" dose of magnesium valproate that would induce achieve histone acetylation and inhibition of HDAC in tumors of patients with cervical cancer.
Results
Study group
A total of 12 patients were studied. All of were chemotherapy- or radiation-naive and had a macroscopic tumor accessible for punch biopsy. Patient mean age was 63.3 years (44–72 years), while the majority of cases were squamous histology and were staged as FIGO stages IIB and IIIB. Status performance was 0–1 in most patients (Table 1).
Table 1 Clinical characteristics of patients
Number 12
Mean age (years) 63.3 (72–44)
Histology
Squamous 8 (66%)
Adenocarcinoma 4 (33%)
FIGO stage*
IIB 3 (25%)
IIIB 8 (66%)
IVB 1 (08%)
Performance status**
0 1 (17%)
1 9 (75%)
2 2 (08%)
*FIGO = International Federation of Gynecology and Obstetrics; ** World Health Organization (WHO) criteria.
Treatment compliance and side effects
All patients completed the study medication. Weight of patients varied from 45–82 kg and mean daily dose for all patients was 1,890 mg. According to dose level, mean daily dose for the 20-mg/kg dose level was 1245 mg (range, 1000–1400 mg), while it was 2000 mg (1800–2100) for the 30 mg/kg- and 2425 mg (1800–3300) for the 40-mg/kg dose level. Treatment in general was well-tolerated. Table 2 shows toxicity recorded according to the Common Toxicity Criteria (CTC). As expected, the most common side effect was depressed level of consciousness that in no case was grade 3 or 4. Thus, three of four patients at the 20-mg and 30-mg dose levels presented this event graded as 2, whereas one and three patients were graded 1 and 2, respectively, at the 40-mg dose; by definition, these side effects did not interfere with the daily living activities of patients. The next effect in frequency was fatigue; also grade 2 in three patients at the 20-mg dose level and in one patient at the 30-mg dose. Other side effects such as nausea, diarrhea, anorexia, and dizziness/lightheadedness were uncommon and mild. There were no changes in the values of non-hematological or hepatic parameters except by lymphopenia grade 1 in a patient receiving the lowest dose level. All toxicities disappeared within the ensuing week.
Table 2 Toxicity to valproate expressed by number of patients suffering the event
Toxicity 4 patients at each dose level (mg/kg)
20 30 40
Depressed level of consciousness* 3 (g2) 3(g2) 1(g1), 3(g2)
Nausea - 1 (g1), 1(g2) 1 (g2)
Diarrhea - - -
Fatigue 3 (g2) 1(g2) -
Anorexia - 1(g2) -
Dizziness/lightheadedness - 1(g2) -
*Grade 2 toxicity. All others were grade 1.
Depressed level of consciousness (g1: somnolence or sedation not interfering with function; g2: somnolence or sedation interfering with function but not interfering with activities of daily living; g3: obtundation or stupor, difficulty to arouse, interfering with daily living; g4) coma.
Histone acetylation of tumors
Pre- and post-treatment tumor samples of all 12 patients were collected; however, the effect of valproate treatment on histone acetylation by Western blot of H3 and H4 could not be assessed in two patients because amount and quality of tumor samples of either pre- or post-treatment biopsies were not adequate. These two patients (numbers 2 and 4) belonged to the 20-mg/kg dose level. Figures 1, 2, and 3 are Western blots of the patients analyzed. The first observation is the ample heterogeneity in degree of baseline acetylation in both H3 and H4 histones. Pre-treatment band was hardly seen for acetylated H3 in patients 1 and 11, whereas it was very strong in patient 6; likewise, acetylated H4 was very weak in patients 3, 5, and 11. In assessment of valproate treatment effect, there was variable increase in band intensity, indicative of hyperacetylation of H3 in all patients except in patient 6. With regard to acetylation of H4, seven patients (1, 3, 5, 6, 7, 11, and 12) had clear hyperacetylation, whereas in the remaining individuals (patients 8, 9, and 10) the effect was minor or non-existent. As can be seen, both H3 and H4 hyperacetylation was observed in patients 1, 3, 5, 7, 11 and 12.
Figure 1 Western blots for anti-acetylated H3 and H4 histones pre- and post-treatment in patients receiving a dose of 20 mg/kg. Patients 2 and 4 are missed due to insufficient sample. Positive and negative controls are HeLa cells treated or not with trichostatinA. Loading control are gels stained with coomassie blue. Graphs in the inferior panel are HDAC activity expressed as optical densities (ODs), in the same scale as positive and negative controls that are also HeLa cells extracts with and without trichostatinA treatment. At the bottom are values of valproic acid in serum.
Figure 2 Western blots for anti-acetylated H3 and H4 histones pre- and post-treatment in patients receiving a dose of 30 mg/kg. Positive and negative controls are HeLa cells treated or not with trichostatin A. Loading control are gels stained with coomassie blue. Graphs in the inferior panel are HDAC activity expressed as ODs, in the same scale as positive and negative controls that are also HeLa cells extracts with and without trichostatinA treatment. At the bottom are values of valproic acid in serum.
Figure 3 Western blots for anti-acetylated H3 and H4 histones pre- and post-treatment in patients receiving a dose of 40 mg/kg. Positive and negative controls are HeLa cells treated or not with trichostatin A. Loading control are gels stained with coomassie blue. Graphs in the inferior panel are HDAC activity expressed as ODs, in the same scale as the positive and negative controls that are also HeLa cells extracts with and without trichostatin A treatment. At the bottom are values of valproic acid in serum.
Serum Levels of valproic acid
Blood serum levels of valproic acid after the 5-day treatment period are shown in Figures 1, 2, and 3, respectively. Samples were taken and analyzed in all cases. Levels ranged from 73.6–170.49 μg/mL. There was lack of correlation between serum levels with dose level. Thus, values for patients were as follows: at the 20-mg/kg dose level 80.79, 98.92, 109.12, and 87.43, for a mean of 94.06 μg/mL; for the 30-mg/kg level 146.56, 81.20, 170.49, and 95.60, for a mean of 123.46 μg/mL, and finally for the highest dose level of 40 mg/kg, corresponding values were 88.47, 94.18, 107.47, and 73.61, with a mean of 90.93 μg/mL.
Histone deacetylase assay in tumors
To investigate whether a decrease in histone acetylase activity could be achieved by valproate treatment in the tumors, enzyme activity was evaluated in tumor biopsies extracts by colorimetric commercial HDAC activity assay in 10 patients; Results are also shown in Figures 1, 2, and 3. Using the same scale for positive and negative control, we found that 8 of 12 (75%) patients had a decrease in deacetylase activity equal to or greater that observed in HeLa cell extracts treated or not with trichostatinA, whereas patients 3 and 8 (25%) had either no change or a mild increase in deacetylase activity. Pooling the absolute values of optical densities (ODs) for pre- and post-treatment samples, there was a statistically significant difference between values (mean 0.36 vs. 0.21, respectively, for a 0.142 mean difference 95% confidence interval (CI 95%) 0.020–0.264), two-tailed t test (p < 0.0264). Our small sample size did not allow for establishing a correlation with H3 and H4 histone acetylation. Remarkably, cases 3 and 8-who showed no HDAC inhibition had tumor hyperacetylation. The patient 3 in both histones whereas the patient 8 only in H3.
H3 and H4 Acetylation in PBMN cells
Due to limitations in amount of blood samples, histone acetylation by Western blot could be assayed only in four patients (1, 2, 5, and 8). Figure 4 shows that all four showed hyperacetylation after treatment in comparison with pre-treatment samples for both H3 and H4 histones.
Figure 4 Western blots for anti-acetylated H3 and H4 histones pre- and post-treatment in PBMN cells of patients 1, 2, 5, and 8. Patient 1 and 2 received a dose of 20 mg/kg, whereas patients 5 and 8 each received a dose of 30 mg/kg. Positive and negative controls are HeLa cells treated or not with thrichostatin A. Loading control are gels stained with Coomassie blue.
Discussion
The balance between histone deacetylases and histone acetyl transferase activities is a major player in regulation of gene transcription [19]. HDAC inhibition induces accumulation of hyperacetylated nucleosome core histones in the majority of chromatin regions but affects expression of only a small subset of genes, leading to transcriptional activation of some but repression of an equal or larger number of other genes [20,21]. This led to testing HDAC inhibitors as anticancer agents in a variety of tumor models and clinical studies. To date, a number of structurally distinct classes of compounds have been identified as HDAC inhibitors including hydroxamates, cyclic tetrapeptides, benzamides, and short-chain fatty acids [22].
The discovery that valproic acid-which belongs to the short-chain fatty acids category and resulted to be an effective HDAC inhibitor [17,18]-encouraged investigation of this agent as a potential cancer therapy agent. Valproic acid can be administered as such or as sodium or magnesium salts, but all three forms have the same pharmacokinetic behaviour and anticonvulsant effect [23]. Here we demonstrate that magnesium valproate, when used at only slightly higher doses than those used as anticonvulsant, not only produces H3 and H4 histone hyperacetylation in PBMN cells but also leads to hyperacetylation of H3 and H4 and inhibition of HDAC activity in tumors.
Preclinical data has shown that valproic acid inhibits the majority of class I and II HDAC at CI50, varying from 0.7–1.3 mM [24]; however, the concentration needed in vitro to achieve its biological effects varies from system to system, but in some cases the concentration required is lower than 0.5 mM in tumors such as glioma, hematopoietic, or breast cancer cell lines [24-26]. On the other hand, it is also known that the inhibitory effect of valproic acid on HDAC begins and disappears within hours after exposure, with a rapid return of baseline acetylation on histones [24]. This knowledge suggests that valproic acid, for effective use in the clinical setting, should be administered continuously and perhaps for long periods. Nonetheless, despite the fact that the therapeutic window of valproic acid is very wide (50–100 μg/mL), there is a clear dose-effect relationship with depressed level of consciousness [27] therefore we felt important to investigate if doses below 60 mg/Kg which produce grade III toxicity lead to histone hyperacetylation in tumors.
These facts led to us to perform this phase I study to find the "adequate" biological dose of magnesium valproate. The advent of the so called "targeted therapies" has led to reconsider the appropriateness of the traditional or classical designs for phase I trials which takes into account toxicity as the most important parameter for dose escalation. Instead some authors have proposed that the endpoint in phase I studies of targeted therapies should be a change in the level or activity of the target or enzyme or other surrogate marker [28]. Because it is clear that the antiproliferative, differentiating and/or pro-apoptotic effects of HDAC inhibitors result from the histone hyperacetylation it seems logical to evaluate the acetylation status of H3 and H4 histones as the endpoint of the trial. On the other hand, not necessarily, the highest hyperacetylation should occur at the higher dose of the tested agent, being possible that a plateau in the effect can be seen in a range of doses that do not produce limiting toxicity. In this sense, the concept of "adequate" instead of the "optimal" biological dose emerges because, by one hand, the number of patients required for finding the optimal dose may be to high for an agent that by itself (or as a single agent) is expected to produce a negligible clinical response rate or tumor shrinkage which equate to expose a larger number of patients to a potentially ineffective agent by its own [29,30]. Another issue for selecting adequate instead of optimal dose design is the fact that it allows investigators to quickly proceed to phase II trials (usually along with chemotherapy or radiation). Perhaps the only situation where an optimal dose design is preferred accounts when the study drug is planned to proceed to larger phase III trials [31].
On this basis, we opted to perform an "adequate" dose phase I. To our surprise, using doses as low as 20–40 mg/kg, the biological effect was observed, being remarkable that no patient presented any grade 3 event. In nine patients, depressed level of consciousness was grade 2. Molecular efficacy in terms of H3 and H4 hyperacetylation in tumors was observed in most patients; in fact if we take into account the effects on H3, at 20 mg/Kg 2 out of 2 evaluated (100%) had hyperacetylation, 3 out of 4 at 30 mg/Kg (75%), and 4 out of 4 (100%) patients at 40 mg/Kg. For H4 a very similar pattern emerged 2/2 (100%), 3/4 (75%) and 2/4 (50%) had hyperacetylation at 20 mg, 30 mg and 40 mg/Kg. When evaluating for both H3 and H4 acetylation the numbers are 2/2 (100%), 2/4 (50%), and 2/4 (50%). Overall, despite the limited number of patients, inherent to the study design it is clear that at the range of doses tested there is a plateau in the molecular response (100%, 50% and 50%) for the three tested doses. In cases like this when there is a tie in efficacy, the highest dose should be selected, and hence we have selected the dose of 40 mg/Kg for our phase II studies [29].
Currently, there is no information concerning the effect of HDAC inhibitors upon acetylation of H3 and H4 in solid tumors of patients. Two reports on depsipeptide have demonstrated hyperacetylation of PBMN cells in two patients with cutaneous T-cell lymphoma and two patients with refractory neoplasms as evaluated by immunoflouorescent labelling [32,33]; additionally, in a clinical study of SAHA for lymphoma and solid tumors it was reported that three of five patients showed increased accumulation of acetylated H3 in the post-treatment biopsy sample [34]. These data are in agreement with our findings indicating that the hyperacetylating effect is clinically achievable with these HDAC inhibitors.
The inhibitory activity of HDAC in cells shown by this class of compounds always parallels changes in H3 and H4 acetylation [24]. We demonstrate herein that valproic acid produces a decrease in HDAC activity in tumors which is accompanied by histone hyperacetylation suggesting that the H3 and H4 hyperacetylation occurring in tumors is a consequence of HDAC inhibition. On the other hand, it is remarkable that three out of the four cases assayed by H3 and H4 acetylation by western blot in PBMN cells (patients 1, 5 and 8) their corresponding tumors had hyperacetylation showing the concordance between the findings in tumors and PBMN cells. These data suggest that HDAC activity in the tumor or H3 and H4 acetylation status in the PBMN cells can be good surrogate markers for monitoring the effect of HDAC inhibitors in clinical trials.
The pharmacokinetics of valproic acid has extensively been studied. Because its half-life is between 8–16 hours, the steady-state concentrations can be determined as soon as at day 4 of treatment [35,36]. Under this rational we determined the levels of valproic acid at day six, within 8–10 hours of the last dose taken, hence the sera values obtained are representative of the levels at the steady-state. High interindividual variation in the pharmacokinetics of valproic acid in epileptic patients is well-known [37]. This phenomenon was observed in our study with serum concentrations ranging from 73.61 to 170.49 μg/mL, and in addition there was no correlation with dose level and mean concentration achieved; thus, for doses of 20, 30, and 40 mg/kg, mean concentrations were 94.06, 123.46, and 90.93 μg/mL, respectively. This finding, however, could be derived from the small number of patients we analyzed, because Tisdale et al. demonstrated significant linear correlation between VPA dose and serum concentration in a sample of 60 epileptic patients [37]. Alternatively, it can be suggested that like in cell lines, which show high variability in response to HDAC inhibitors [24-26], interpatient variability in tumors may exist.
There is one report in abstract form proceeding from a clinical study with valproic acid. Atmaca et al., performed a "traditional" dose escalating phase I study using valproic acid by intravenous infusion in patients with advanced cancer. In the study, 26 patients pre-treated and with progressive malignant disease received the drug by 1-h infusion split twice a day for 5 consecutive days, repeating the treatment at intervals of 2 weeks in cohorts of 60, 75, 90, and 120 mg/kg dosages. These investigators found maximum tolerated dose of 60 mg/kg because 9 of 26 patients presented grade 3/4 neurological toxicity. In addition, these authors reported hyperacetylation of PBMN cells in the majority of patients; nevertheless, they provided no information on serum levels achieved or the molecular response according to the dose level [38].
In a phase I-II study of the combination of decitabine and valproic acid for myeloblastic acute leukemia and myelodysplastic syndrome, a fixed dose of 15 mg/m2 of decitabine and valproic acid at 20, 35 and 50 mg/Kg were administered every 10 days. The toxicity observed was grade III neurotoxicity at 50 mg/Kg in 1 of 10 and grade II in 5 out 10 patients. Subsequently, VPA 50 mg/kg was chosen for the phase II portion of the study. Of 40 evaluable for response, an overall response rate of 22% (8 PR and 1 CR) was observed. Interestingly, the response rates according to the valproic acid dose were 33% for 20 mg, 11% for the 35 mg and 25% for the 50 mg/Kg dose level [39]. Finally, in a third phase I study for metastatic solid tumors valproic acid was administered as an IV loading dose followed by 5 doses of oral doses given every 12 hours followed by a dose of epirubicin at day 3. At the time of reporting, 16 patients have been treated at 4 dose levels: VPA 15, 30, 45, and 60 mg/kg; with epirubicin at 75 mg/m2. The maximum tolerated dose has not been reached and dose escalation is continuing. Major responses were observed in all tumor types including in anthracycline failures and in anthracycline-resistant cancers such as melanoma and cervical carcinoma. Plasma levels were not reported, however, there were above the concentrations needed for in-vitro synergy [40].
The extensive use of valproate as anticonvulsant has shown excellent tolerability in the therapeutic range between 50 and 100 μg/mL [15,16]. Despite the fact that levels >100 μg/mL are considered supra-therapeutic [41], it must be stressed that in the our present study four patients presented concentrations higher than that level (109.12, 146.56, 170.49, and 107.43 μg/mL), and that none of these patients presented any grade 3 toxicity, which indicates the feasibility of achieving these levels during treatment to maximize the chances of producing tumor hyperacetylation. In fact, significant clinical adverse effects of valproic acid ingestion, such as lethargy, coma, tachycardia, metabolic acidosis, and hypotension, are more likely to occur with concentrations >450 μg/mL [41].
All together, existing data suggest that molecular and clinical responses occur at valproic acid doses between 20 and 60 mg/Kg. The optimal biological dose of valproic acid remains to be determined, however, our results suggest the existence of a plateau in the histone acetylation at doses between 20 and 40 mg/Kg. Based on that we have chosen 40 mg/Kg as the dose to test in phase II trials.
Synergy of HDAC inhibitors with demethylating agents for reactivating expression of tumor suppressor genes as well as to induce antitumor effects is well-known [42-44]. In this regard, our group showed that hydralazine is an effective demethylating agent that reactivates the tumor suppressor genes expression silenced by methylation [45,46] and reported the results of a phase I study demonstrating the gene demethylating and reactivating activity of the drug at doses between 75 and 150 mg/day [47]. On these bases, we at present are testing these two drugs-demethylating hydralazine and the HDAC inhibitor valproic acid-plus chemotherapy or plus chemoradiation in phase II studies for solid tumors.
Conclusion
Our results provide evidence that magnesium valproate at doses between 20 mg and 40 mg/Kg, inhibits deacetylase activity and hyperacetylates histones in tumor tissues. Its clinical efficacy along with a demethylating agent plus chemotherapy or radiation is currently being tested in phase II studies.
Methods
Patient selection
Previously untreated patients with histological diagnosis of carcinoma of the cervix uteri entered into this phase I study. Patients were invited to participate in the study in the waiting time from diagnostic evaluation to commencing chemoradiation. The whole study period lasted only 6 days, hence valproic acid treatment was not repeated.
The following inclusion criteria were applied: 1) age between 18 and 75 years; 2) World Health Organization Performance Status 0–2; 3) hematological, renal, and hepatic functions as follows: hematological: hemoglobin ≥10 g/L; leukocytes >4,000/mm3, platelets >100,000/mm3, total bilirubin and transaminasas <1.5× the normal upper limit, and normal serum creatinine; 4) normal chest x-ray, and 5) signed informed consent for study medication and biopsies pre and post-treatment. Exclusion criteria included the following: 1) history of allergy to valproic acid; 2) any past or current central nervous system pathologic condition (epilepsy, etc.) that required pharmacologic treatment; 3) any current hepatic disease; 4) uncontrolled infection or other systemic diseases; 5) concomitant treatment with any experimental drug; 6) pregnant or nursing women; 7) mental illness, and 8) previous or concomitant malignant diseases other than non-melanoma skin cancer. The Institutional Regulatory Board approved the study protocol.
Clinical samples
Biopsies were taken from areas with visible macroscopic cervical tumor using a sterile biopsy punch the day before and the day six after the five days of magnesium valproate treatment. (The post-treatment biopsy and peripheral blood sampling were taken in the early morning, between 8 and 10 hours after the last dose of magnesium valproate). Part of the biopsy was sent to the Institution's Pathology Department for routine Hematoxilin & eosin diagnosis. The remaining biopsy specimen was immediately frozen at -20°C for biological analyses. In addition, a blood sample of 10 mL was drawn from the arm by venipuncture to obtain serum and the mononuclear cell fraction for protein extraction.
Magnesium valproate treatment
After tumor and blood sampling, patients were divided into the following groups and started oral magnesium valproate for a five-day period: I) 20 mg/kg, II) 30 mg/kg; III) 40 mg/Kg. Total dose was divided in three administrations every 8 h (8 AM, 4 PM and 12 PM) per oral route in enteric-coated tablets of 200 and/or 400 mg. Biological effects of treatment on tumor and peripheral blood samples as well as clinical and laboratory toxicity were assessed at the end of the study period (6 days). Patients then went to receive the definitive chemoradiation treatment as planned.
Toxicity was registered according to the Common Toxicity Criteria of the National Cancer Institute (CTC NCI) with special emphasis on the presence, throughout the treatment days, of known signs and symptoms associated with magnesium valproate treatment such as somnolence/sedation and fatigue. Therapeutic levels of valproic acid were also evaluated in blood serum at day 6, between 8 and 10 hours after the last dose of magnesium valproate.
Acid extraction of proteins
Acid extraction of histones was performed as described previously [48] with modifications. Briefly, tumor samples and mononuclear cell pellets were suspended in 10 volumes of PBS and centrifuged at 200 × g for 10 min. Cells were then suspended with five volumes of lysis buffer [10 mM HEPES (pH 7.9), 1.5 mM MgCl2, 10 mM KCl, 0.5 mM DTT, and 1.5 mM phenylmethylsulfonyl fluoride] and hydrochloride acid at a final concentration of 0.2 M and subsequently lysed on ice for 30 min. After centrifugation at 11,000 × g for 10 min at 4°C, the cell supernatant fraction that contained acid-soluble proteins was retained. Supernatant was dialyzed against 200 mL of 0.1 M acetic acid twice for 1–2 h each and then dialyzed against 200 mL of H2O for 1 h, 3 h, and overnight. Dialysis was performed using a Spectra/Pore 3 Dialysis Membranes 3,500 MWCO (Spectrum Laboratories, Inc., Rancho Dominguez, CA, USA).
Western blot analysis
Acid extracted proteins were analyzed by sodium duodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE)/immunoblotting with antibodies recognizing acetylated histones (rabbit polyclonal IgG, anti-acetyl-histone H4, and rabbit polyclonal-anti-acetyl-histone H3; Upstate Biotechnology, Lake Placid, NY, USA). Protein samples were separated along with molecular weight markers (Bio-Rad, Hercules, CA, USA) in 15% polyacrylamide gels. Gels were transferred onto 0.45-μm nitrocellulose membranes (Schleicher and Schuell, Keene, NH, USA). Gel loading equivalence was confirmed by Coomassie blue stain (Sigma, St Louis, MO, USA). Species-specific immunoglobulin G-horseradish peroxidase (IgG-HRP) secondary antibodies were purchased from Bio-Rad. Blots were developed with chemiluminescent substrate (Amersham International, Buckinghamshire England), and autoradiography was performed utilizing X-OMAT film (Kodak, Rochester, NY, USA).
Histone deacetylase assay
Assays were performed using the colorimetric HDAC activity assay from BioVision (BioVision Research Products, Mountain View, CA, USA) according to manufacturer instructions. Briefly, 50 μg of nuclear extracts from tumors were diluted in 85 μL of ddH2O; then, 10 μL of 10× HDAC assay buffer were added followed by addition of 5 μL of the colorimetric substrate; samples were incubated at 37° for 1 h. Subsequently, the reaction was stopped by adding 10 μL of lysine developer and left for additional 30 min at 37°C. Samples were then read in an ELISA plate reader at 400 nm. HDAC activity was expressed as relative OD values per μg of protein sample. The kit contains negative and positive controls that consist of nuclear extract of HeLa treated or not with trichostatinA, respectively.
Blood levels of valproic acid. Valproic acid was measured in serum using a fluorescence polarization immunoassay (FPIA) technology. This is a competitive binding immunoassay technology in which fluorescein-labeled and patient antigens compete for binding sites on antibody molecules [49]. Polarization of the fluorescent light emitted from labeled drug-antibody complex is inversely related with amount of drug in the sample. TDx/TDxFLx analyzer and the valproic acid kit REF9514 were obtained from Abbott Laboratories (Abbott Laboratories, Abbott Park, IL, USA). Blood samples were obtained in an sst-gel-clot activator tube 8 h after the last dose of magnesium valproate and assays were conducted following manufacturer instructions; 100-μL samples of serum were run by duplicate and values expressed as μg/mL.
Statistical analysis
All numerical data were expressed as average of values obtained ± standard deviation (SD). Statistical significance was determined by conducting a paired Student t test.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
PZ participated the collection and storing of clinical samples; A C-B and B S-P, performed the Western blots and histone deacetylase assays, A G-F and P G-L participated in the determination of serum valproic acid levels; E P-C and L T-C participated in analysis of results, LC, MC and DC cared for patients, EA, GC, CP-P and CT-B participated in discussion of results, A D-G conceived of the study and wrote the manuscript.
Acknowledgements
This work was supported by CONACyT grants SALUD-2002-C01-6579 and AVANCE C01-294, and by Psicofarma, S.A. de C.V., Mexico.
==== Refs
Verma M Maruvada P Srivastava S Epigenetics and cancer Crit Rev Clin Lab Sci 2004 41 585 607 15603512 10.1080/10408360490516922
Rountree MR Bachman KE Baylin SB DNMT1 binds HDAC2 and a new co-repressor, DMAP1, to form a complex at replication foci Nature Genet 2000 25 269 277 10888872 10.1038/77023
Nan X Ng HH Johnson CA Laherty CD Turner BM Eisenman RN Bird A Transcriptional repression by the methyl-CpG-binding protein MeCP2 involves a histone deacetylase complex Nature 1998 393 386 389 9620804 10.1038/30764
Shao Y Gao Z Marks PA Jiang X Apoptotic and autophagic cell death induced by histone deacetylase inhibitors Proc Natl Acad Sci U S A 2004 101 18030 18035 15596714 10.1073/pnas.0408345102
Michaelis M Suhan T Cinatl J Driever PH Cinatl J Jr Valproic acid and interferon-alpha synergistically inhibit neuroblastoma cell growth in vitro and in vivo Int J Oncol 2004 25 1795 1799 15547719
Cinatl J JrKotchetkov R Blaheta R Driever PH Vogel JU Cinatl J Induction of differentiation and suppression of malignant phenotype of human neuroblastoma BE(2)-C cells by valproic acid: enhancement by combination with interferon-alpha Int J Oncol 2002 20 97 106 11743648
Kawagoe R Kawagoe H Sano K Valproic acid induces apoptosis in human leukemia cells by stimulating both caspase-dependent and -independent apoptotic signaling pathways Leuk Res 2002 26 495 502 11916526 10.1016/S0145-2126(01)00151-5
Knupfer MM Pulzer F Schindler I Hernaiz Driever P Knupfer H Keller E Different effects of valproic acid on proliferation and migration of malignant glioma cells in vitro Anticancer Res 2001 21 347 351 11299760
Marks PA Richon VM Rifkind RA Histone deacetylase inhibitors: inducers of differentiation or apoptosis of transformed cells J Natl Cancer Inst 2000 92 1210 1215 10922406 10.1093/jnci/92.15.1210
Kim MS Blake M Baek JH Kohlhagen G Pommier Y Carrier F Inhibition of histone deacetylase increases cytotoxicity to anticancer drugs targeting DNA Cancer Res 2003 63 7291 7300 14612526
Castro-Galache MD Ferragut JA Barbera VM Martin-Orozco E Gonzalez-Ros JM Garcia-Morales P Saceda M Susceptibility of multidrug resistance tumor cells to apoptosis induction by histone deacetylase inhibitors Int J Cancer 2003 104 579 586 12594812 10.1002/ijc.10998
Camphausen K Scott T Sproull M Tofilon PJ Enhancement of xenograft tumor radiosensitivity by the histone deacetylase inhibitor MS-275 and correlation with histone hyperacetylation Clin Cancer Res 2004 10 6066 6071 15447991
Kim JH Shin JH Kim IH Susceptibility and radiosensitization of human glioblastoma cells to trichostatin A, a histone deacetylase inhibitor Int J Radiat Oncol Biol Phys 2004 59 1174 1180 15234053 10.1016/j.ijrobp.2004.03.001
Zhang Y Jung M Dritschilo A Jung M Enhancement of radiation sensitivity of human squamous carcinoma cells by histone deacetylase inhibitors Radiat Res 2004 161 667 674 15161353
Loscher W Basic pharmacology of valproate: a review after 35 years of clinical use for the treatment of epilepsy CNS Drugs 2002 16 669 669 12269861
Perucca E Pharmacological and therapeutic properties of valproate: a summary after 35 years of clinical experience CNS Drugs 2002 16 695 714 12269862
Phiel CJ Zhang F Huang EY Guenther MG Lazar MA Klein PS Histone deacetylase is a direct target of valproic acid, a potent anticonvulsant, mood stabilizer, and teratogen J Biol Chem 2001 276 36734 36741 11473107 10.1074/jbc.M101287200
Kramer OH Zhu P Ostendorff HP Golebiewski M Tiefenbach J Peters MA Brill B Groner B Bach I Heinzel T Gottlicher M The histone deacetylase inhibitor valproic acid selectively induces proteasomal degradation of HDAC2 EMBO J 2003 22 3411 3420 12840003 10.1093/emboj/cdg315
Vigushin DM Coombes RC Targeted histone deacetylase inhibition for cancer therapy Curr Cancer Drug Targets 2004 4 205 218 15032670 10.2174/1568009043481560
Shi H Wei SH Leu YW Rahmatpanah F Liu JC Yan PS Nephew KP Huang TH Triple analysis of the cancer epigenome: an integrated microarray system for assessing gene expression, DNA methylation, and histone acetylation Cancer Res 2003 63 2164 2171 12727835
Peart MJ Smyth GK van Laar RK Bowtell DD Richon VM Marks PA Holloway AJ Johnstone RW Identification and functional significance of genes regulated by structurally different histone deacetylase inhibitors Proc Natl Acad Sci U S A 2005 102 3697 3702 15738394 10.1073/pnas.0500369102
Mei S Ho AD Mahlknecht U Role of histone deacetylase inhibitors in the treatment of cancer (Review) Int J Oncol 2004 25 1509 1519 15547685
Chapman A Keane PE Meldrum BS Simiand J Vernieres JC Mechanism of anticonvulsant action of valproate Prog Neruobiol 1982 28 963 964
Gurvich N Tsygankova OM Meinkoth JL Klein PS Histone deacetylase is a target of valproic acid-mediated cellular differentiation Cancer Res 2004 64 1079 1086 14871841
Driever PH Knüpfer MM Cinatl J Wolff JFA Valproic acid for the treatment of pediatric malignant glioma Klin Pediatr 1999 211 323 328
Olsen CM Meussen-Elholm ETM Roste LS Tauboll E Antiepileptic drugs inhibit cell growth in the human breast cancer cell line MCF7 Mol Cell Endocrinol 2004 213 173 179 15062565 10.1016/j.mce.2003.10.032
Davis R Peters DH McTavish D Valproic acid: A reappraisal of its pharmacological properties and clinical efficacy in epilepsy Drugs 1994 47 332 372 7512905
Parulekar WR Eisenhauer EA Novel endpoints and design of early clinical trials Ann Oncol 2002 13 139 43 12401680
Hunsberger S Rubinstein LV Dancey J Korn EL Dose escalation trial designs based on a molecularly targeted endpoint Stat Med 2005 24 2171 2181 15909289 10.1002/sim.2102
Parulekar WR Eisenhauer EA Phase I trial design for solid tumor studies of targeted, non-cytotoxic agents: theory and practice J Natl Cancer Inst 2004 96 990 997 15240782
Korn EL Nontoxicity endpoints in phase I trial designs for targeted, non-cytotoxic agents J Natl Cancer Inst 2004 96 977 978 15240771
Piekarz RL Robey R Sandor V Bakke S Wilson WH Dahmoush L Kingma DM Turner ML Altemus R Bates SE Inhibitor of histone deacetylation, depsipeptide (FR901228), in the treatment of peripheral and cutaneous T-cell lymphoma: a case report Blood 2001 98 2865 2868 11675364 10.1182/blood.V98.9.2865
Sandor V Bakke S Robey R Kang MH Blagaskonny M Brooks R Piekarz R Tucker E Figg WD Chan KK Goldspiel B Sausville E Balcerzak SP Bates SE Phase I trial of the histone deacetylase inhibitor, depsipeptide (FR90 NSC 630176), in patients with refractory neoplasms Clin Cancer Res 1228 8 718 728 11895901
Kelly WK Richon VM O'Connor O Curley T MacGregor-Curtelli B Tong W Klang M Schwartz L Richardson S Rosa E Drobnjak M Cordon-Cordo C Chiao JH Rifkind R Marks PA Scher H Phase I clinical trial of histone deacetylase inhibitor: suberoylanilide hydroxamic acid administered intravenously Clin Cancer Res 2003 9 3578 3588 14506144
Miljkovic B Pokrajac M Varagic V Levic Z Single dose and steady state pharmacokinetics of valproic acid in adult epileptic patients Int J Clin Pharmacol Res 1991 11 137 141 1809700
Bruni J Wilder BJ Willmore LJ Perchalski RJ Villarreal HJ Steady-state kinetics of valproic acid in epileptic patients Clin Pharmacol Ther 1978 24 324 332 357068
Tisdale JE Tsuyuki RT Oles KS Penry JK Relationship between serum concentration and dose of valproic acid during monotherapy in adult outpatients Ther Drug Monit 1992 14 416 423 1448851
Atmaca A Maurer A Heinzel T Gotlicher M Neumann A Al-Batran SE Martin E Bartshc I Knuth A Jaegen E A dose-scalating phase I study with valproic acid (VPA) in patients (pts) with advanced cancer Procc ASCO 2004 23 (abstract 3169).
Garcia-Manero G Kantarjian H Sanchez-Gonzalez B Verstovsek S Ravandi F Ryttling M Cortes J Yang H Fiorentino J Rosner G Issa J Results of a Phase I/II Study of the Combination of 5-aza-2-deoxycytidine and 'valproic acid in patients with acute myeloid leukemia and myelodysplastic syndrome Procc ASCO 2005 24 (abstract 6544).
Munster PN Marchion DC Bicaku E Sullivan P Beam C Mahany J Lush R Sullivan DM Daud A Phase I trial of the histone deacetylase Inhibitor, valproic acid and the topoisomerase II inhibitor, epirubicin: A clinical and translational study Procc ASCO 2005 24 (abstract 3084).
Spiller HA Krenzelok EP Klein-Schwartz W Winter ML Weber JA Sollee DR Bangh SA Griffith JR Multicenter case series of valproic acid ingestion: serum concentrations and toxicity J Toxicol Clin Toxicol 2000 38 7557 7560 10.1081/CLT-100102388
Cameron EE Bachman KE Myohanen S Herman JG Baylin SB Synergy of demethylation and histone deacetylase inhibition in the re-expression of genes silenced in cancer Nat Genet 1999 21 103 107 9916800 10.1038/5047
Zhu WG Otterson GA The interaction of histone deacetylase inhibitors and DNA methyltransferase inhibitors in the treatment of human cancer cells Curr Med Chem Anti-Cancer Agents 2003 3 187 199 10.2174/1568011033482440
Mongan NP Gudas LJ Valproic acid, in combination with all-trans retinoic acid and 5-aza-2'-deoxycytidine, restores expression of silenced RARbeta2 in breast cancer cells Mol Cancer Ther 2005 4 477 486 15767557
Segura-Pacheco B Trejo-Becerril C Perez-Cardenas E Taja-Chayeb L Mariscal I Chavez A Acuna C Salazar AM Lizano M Duenas-Gonzalez A Reactivation of tumor suppressor genes by the cardiovascular drugs hydralazine and procainamide and their potential use in cancer therapy Clin Cancer Res 2003 9 1596 603 12738711
Angeles EE Vazquez-Valadez VH Vasquez-Valadez O Velazquez-Sanchez AM Ramirez A Martinez L Diaz-Barriga S Romero-Rojas A Cabrera G Lopez-Castañares R Duenas-Gonzalez A Computational studies of 1-hydrazinophthalazine (Hydralazine) as antineoplasic agent. Docking studies on methyltransferase Letters Drug Design Discovery 2005 4 282 286 10.2174/1570180054038413
Zambrano P Segura-Pacheco B Perez-Cardenas E Cetina L Revilla-Vazquez A Taja-Chayeb L Chavez-Blanco A Angeles E Cabrera G Sandoval K Trejo-Becerril C Chanona-Vilchis J Duenas-Gonzalez A A phase I study of hydralazine to demethylate and reactivate the expression of tumor suppressor genes BMC Cancer 2005 5 44 15862127 10.1186/1471-2407-5-44
Zhu WG Lakshmanan RR Beal MD Otterson GA DNA methyltransferase inhibition enhances apoptosis induced by histone deacetylase inhibitors Cancer Res 2001 61 1327 1333 11245429
Jolley ME Fluorescence polarization immunoassay for determination of therapeutic drug levels in human plasma J Anal Tox 1981 5 236 240
|
16001982
|
PMC1198251
|
CC BY
|
2021-01-04 16:36:35
|
no
|
Mol Cancer. 2005 Jul 7; 4:22
|
utf-8
|
Mol Cancer
| 2,005 |
10.1186/1476-4598-4-22
|
oa_comm
|
==== Front
Mol CancerMolecular Cancer1476-4598BioMed Central London 1476-4598-4-311609113310.1186/1476-4598-4-31Short CommunicationReduced expression of multiple gap junction proteins is a feature of cervical dysplasia Aasen Trond [email protected] Sheila V [email protected] Mike [email protected] Malcolm B [email protected] Squamous Cell Biology and Dermatology, Division of Cancer Sciences and Molecular Pathology, Robertson Building, University of Glasgow, 56 Dumbarton Road, Glasgow, G11 6NU, Scotland, UK2 Institute of Biomedical and Life Sciences, Division of Virology, University of Glasgow, Church Street, Glasgow, G11 6JR, Scotland, UK3 Centre for Cutaneous Research, Institute of Cell and Molecular Science, Queen Mary University of London, 4 Newark Street, Whitechapel, London E1 2AT, UK2005 9 8 2005 4 31 31 16 4 2005 9 8 2005 Copyright © 2005 Aasen et al; licensee BioMed Central Ltd.2005Aasen et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Cervical dysplasia is a premalignant lesion associated with human papillomavirus (HPV) infection which, over time, can turn cancerous. Previous studies have indicated that loss of gap junctions may be a feature of cervical cancer and premalignant dysplasia. Loss of the gap junction protein connexin43 has been demonstrated in dysplastic cervix, but other connexins have not been investigated. In contrast we previously showed that HPV-associated cutaneous warts – and other hyperproliferative skin conditions – display a dramatic upregulation of certain connexins, in particular connexin26. By performing immunofluorescence staining after antigen retrieval of paraffin-embedded cervical tissue samples, this study reports for the first time that connexin26 and connexin30, in addition to connexin43, are expressed in differentiating cells of normal human cervical epithelia. Moreover, in dysplastic ectocervix, all connexins studied display a dramatic loss of expression compared to adjacent normal epithelia. The role of connexins in keratinocyte differentiation and carcinogenesis is discussed.
cervical cancergap junctionsconnexinspapillomaviruskeratinocytes
==== Body
Findings
Connexins, a family of 20 trans-membrane proteins in humans, comprise the main subunits of gap junctions – specialised clusters of intercellular channels that allow adjacent cells to directly share ions and hydrophilic molecules of up to ~1 KDa in size [1]. Gap junctional intercellular communication (GJIC) is thought to control tissue homeostasis and to coordinate cellular processes such as proliferation, migration and differentiation. Disruption of GJIC or mutations in connexins is associated with several human diseases such as hearing loss, neuropathies and various skin conditions [2].
There is also substantial evidence that connexins have a tumour suppressor role [reviewed in [3]]. While reduced or aberrant GJIC or connexin expression has been found in some tumours and in many tumour cell lines [4-7], restoration of GJIC in tumour cell lines by connexin transfection can reduce growth and tumourigenicity [8-10]. However, the tumour suppressive effects may be tissue and connexin-specific [11,12] and also appear to involve non-gap junctional properties of connexins [13-15]. Moreover, it has been observed that connexin expression (especially connexin26) is often upregulated in hyperplastic tissues including psoriatic epidermis and viral warts [16], benign prostatic hyperplasia [17], and mouse papillomas [18]. While induction of connexin26 and connexin43 has also been observed in metastatic breast carcinomas [19], others have reported that connexin26 and connexin43 are downregulated in mammary carcinoma cell lines and re-expression of these connexins leads to repression of tumour-forming ability [20]. Although potent tumour promoters markedly downregulate GJIC in cultured cells [21], intact skin painted with tumour promoters such as 12-O-tetradecanoylphorbol 13-acetate (TPA) show a dramatic upregulation of connexin26 and connexin43 expression [22-24]. Moreover, several reports have shown a negative correlation between expression of connexins and cell diapedesis and or tumour metastasis, including brain tumours [25,26], melanoma [27], breast carcinoma [28] and lung squamous cell carcinomas [29]. Thus, although connexins act as tumour suppressor genes in several types of cancers, the role of connexins in metastasis are more conflicting.
The association of certain "high risk" human papillomaviruses (HPVs) with the development of cervical cancer on the other hand has been clearly demonstrated [30], with several targets and functions of the viral oncoproteins identified [31]. However, tumour progression only occurs in a very small subpopulation of HPV infected individuals; thus, it is thought that several molecular and cellular changes are required over time for malignant conversion to take place. One of these cellular changes may be loss of connexin expression and/or GJIC. Indeed, it was observed more than four decades ago, using freeze-fracture electron microscopy, that normal cervix has abundant gap junctions, and that these are deficient in cervical carcinomas [32]. Further work also demonstrated a dramatic decrease of gap junction plaques in pre-malignant conditions such as severe dysplasia [33]. More recently, immunohistochemistry of cervical biopsies showed reduced connexin43 expression in dysplastic regions compared to normal epithelia [34] and work in-vitro has suggested that loss of GJIC may be an early event in papillomavirus associated cell transformation [35-38]. However, this may be limited to certain cell types or to particular oncogenes that are not always expressed during tumour progression (for example the E5 oncogene product that is frequently deleted after viral integration). Recently, we also showed that human cervical keratinocytes harbouring HPV-16 expressed several connexins (including connexin26, connexin30 and connexin43), and displayed extensive GJIC that was only lost at a late malignant stage [39]. This, together with our observation of a dramatic upregulation of connexin26 expression in cutaneous HPV warts [16], prompted us to investigate the expression pattern of several connexins in normal and dysplastic cervical epithelia.
Six duplicate slides of HPV-16 positive dysplastic cervix (cervical intraepithelial neoplasia (CIN) I, CIN I/II, and CIN III) were obtained from Dr. John Doorbar (Dr Karl Sotlar Institute for Pathology, Tubingen Germany, as previously described [40]). Antigen retrieval using pressure cooking in citrate buffer was performed followed by immunofluorescence staining using antibodies and methods as previously described [39]. All antibodies were diluted in PBS containing 1% bovine serum albumin (BSA), 0.1% Tween-20 and 0.01% SDS, and incubated overnight at 4°C. The presence of BSA and SDS was used to improve the connexin staining as previously shown for other antibodies [41]. Sections were mounted with Vectashield mounting medium containing DAPI and all images were obtained using an upright Zeiss Axioplan 2 fluorescence microscope. After immunofluorescence imaging the slides were washed and stained with Haematoxylin and Eosin (H&E) to verify previous CIN gradings.
As seen in Figure 1A, all sections of normal ectocervix expressed connexin43 (red) mainly in the suprabasal spinous layers. Some staining was also seen in the basal layer although it was difficult to identify specific gap junction plaques. Connexin26 and connexin30 were found mainly in the upper spinous layers, but some was also detected in the lower layers similar to connexin43, particularly with connexin30 (1A and 1B). Overlapping co-localisation (yellow gap junction plaques) was seen in connexin26/connexin43 triple-immunofluorescence staining experiments (Figure 1A) as observed in palmoplantar skin [42]. The more differentiated cervical squames were negative for all three connexins. Moreover, no definite connexin membrane plaques could be identified in the basal layer although some gap junction plaques are likely to be present as previously shown in early freeze-fracture electron microscopy studies [32,33]. As expected, the proliferation marker Ki-67 was detected in basal or suprabasal keratinocytes but not in differentiated cells (Figure 1B).
Figure 1 Connexin expression in normal and dysplastic human cervical epithelium. All nuclei are stained with DAPI (blue). A: All normal ectocervical tissue sections revealed positive staining of connexin43 gap junctions (red) particularly in the spinous layers. Connexin26 was also positive (green), particularly in upper spinous and more superficial layers. Overlapping expression of connexin43 and connexin26 was observed (yellow) and these areas displayed particularly large gap junction plaques (arrow). B: Connexin30 (red) displayed a similar staining pattern to Connexin26 although gap junction plaques were more frequently observed in lower spinous layers (arrow). In these normal sections, the proliferation marker Ki-67 was detected (turquoise) in basal and immediate suprabasal layers only, as expected. (C) In all premalignant CIN III sections there was complete absence of connexin26 (green) and connexin30 (red). D: In one CIN III lesion, small amounts of connexin43 (red) were present, mainly as diffuse cytoplasmic staining, but no positive connexin26 staining was detectable (green). E: In CIN I/II lesions no clear connexin26 gap junctions were present (green) whereas several connexin43 plaques were visible (arrow), but in a less homogenous fashion compared to normal ectocervix. F: Connexin30 gap junctions were clearly detected in areas of low grade CIN I lesions, although they tended to occur in stratified regions of cell differentiation (arrow) rather than areas of abnormal cellular proliferation (Ki-67, turquoise) and atypical cellular crowding.
In contrast, all sections of dysplastic cervix displayed loss of protein expression of all three connexins investigated (Figure 1C–F). Connexin26 and connexin30 expression was completely absent in the CIN3 lesions (Figure 1C). In some areas however, particularly using high magnification power and high laser exposure, some connexin43 plaques could be seen in dysplastic areas, although it was difficult to assess accurately the dysplasia in those exact areas (Figure 1D–E). Some small connexin30 gap junction plaques were also identified in dysplastic tissue, but generally these appeared separate from cells positive for the proliferation marker Ki-67 in stratified cells of dysplastic cervix (Figure 1F). Thus, although some connexin30 and connexin43 gap junction plaques were occasionally seen in low grade CIN I/II cervical biopsies, there was always a dramatic loss of connexin expression compared to adjacent normal tissue.
In conclusion, this report demonstrates for the first time the presence of connexin26 and connexin30 gap junction plaques, as well as connexin43, in human ectocervix, displaying a similar staining pattern to that observed for connexin26 in vaginal epithelium [16]. Moreover, dysplastic areas of the biopsies show a dramatic loss of gap junction plaques, primarily due to reduced expression rather than cytoplasmic accumulation of connexin proteins. However, it cannot be excluded that cytoplasmic connexins are undetectable by the current staining protocol and antibodies available. The use of paraffin embedded tissue may reduce the detection efficiency of smaller gap junction plaques and/or cytoplasmic proteins. Some general background immunofluorescence is present in both normal and dysplastic tissue and this also hampers detection of cytoplasmic proteins. Significantly however, lack of staining in dysplastic lesions cannot be attributed to sample or patient factors, as the staining of both normal and dysplastic tissue were performed using the same tissue biopsies. It thus also appears that the lack of gap junctions in CIN lesions are likely to be controlled in an intracrine fashion rather than as an effect of paracrine or hormonal fashion, particularly since areas of gap junction plaques where observed in differentiating cells layered in between dysplastic/Ki-67 positive cells. Currently very little is known regarding how connexin gene expression is regulated and further studies, perhaps using laser capture followed by RT-PCR and/or methylation specific PCR, are required to elucidate what regulates the overexpression and loss of connexins, in cutaneous warts and CIN lesions respectively. Our previous work on model cell lines indicate that a combination of cytoplasmic connexin accumulation and loss of connexin transcription may occur [39].
Expression of connexin26 but not connexin40 and connexin43 has been shown to reduce tumourigenesis of HeLa cells (cervical tumour cells) [12]. It is also likely that other connexins are expressed in human ectocervix, for example connexin31 which is expressed in differentiating keratinocytes in the epidermis [43], and may play a different role in epidermal biology and carcinogenesis. In order to answer some of these outstanding questions, it is imperative to decipher more accurately what biological role gap junctions and individual connexins execute.
These results are clearly in disparity to observations in interfollicular epidermis, where connexin26 and connexin30 are normally not present (apart from palmoplantar skin) but becomes highly expressed in areas of hyperproliferation such as viral warts [16]. The reason for this remains unknown, but several factors may contribute. For example, the stratifying cervical epithelium is of non-keratinising nature. Unlike dysplastic cervical epithelia, cutaneous warts are typically highly proliferative, thick lesions, and there is some evidence that the presence of gap junctions may favour such stratification (perhaps due to metabolic co-operation) [39]. Conversely, it is currently difficult to explain why gap junctions are lost in HPV-16 positive CIN lesions. It is clear however that basal keratinocytes express at least tenfold fewer gap junction plaques than differentiated keratinocytes [32,33], and the lack of gap junctions in CIN lesions may simply reflect failure of keratinocyte differentiation in the stratified layers. Although there is some evidence, it remains to be seen whether connexins themselves play a direct role in regulating keratinocyte differentiation. Loss of connexin expression may also be associated with HPV infection, although our recent results suggests that such correlation requires high levels of oncogene expression and/or a further malignant progressed state [39]. A viral advantage associated with loss of GJIC has not been described, however a recent intriguing report has documented gap-junction-mediated immunological coupling allowing direct transfer of antigenic peptides [44] which may be involved in ensuring proper antigenic T-cell response against viruses such as HPV hiding in cells or expressing anti-apoptotic proteins.
Authors' contributions
TA carried out all experimental assays and drafted the manuscript. TA, MBH, ME and SVG participated in study design and coordination, data interpretation and manuscript preparation. All authors read and approved the final manuscript.
Acknowledgements
Dr. John Doorbar kindly provided the CIN-graded paraffin embedded tissue sections (Dr Karl Sotlar Institute for Pathology, Tubingen Germany, obtained as previously described [40]). We are grateful to Dr Edgar Rivedal for supplying the connexin43 antibody. TA was supported by a University of Glasgow Medical Faculty Postgraduate Research Scholarship and by a postgraduate training grant from Roche products.
==== Refs
Kumar NM Gilula NB The gap junction communication channel Cell 1996 84 381 388 8608591 10.1016/S0092-8674(00)81282-9
Kelsell DP Dunlop J Hodgins MB Human diseases: clues to cracking the connexin code? Trends Cell Biol 2001 11 2 6 11146276 10.1016/S0962-8924(00)01866-3
Trosko JE Ruch RJ Cell-cell communication in carcinogenesis Front Biosci 1998 3 D208 D236 9458335
Saito T Nishimura M Kudo R Yamasaki H Suppressed gap junctional intercellular communication in carcinogenesis of endometrium Int J Cancer 2001 93 317 323 11433394 10.1002/ijc.1350
Saitoh M Oyamada M Oyamada Y Kaku T Mori M Changes in the expression of gap junction proteins (connexins) in hamster tongue epithelium during wound healing and carcinogenesis Carcinogenesis 1997 18 1319 1328 9230274 10.1093/carcin/18.7.1319
Holden PR McGuire B Stoler A Balmain A Pitts JD Changes in gap junctional intercellular communication in mouse skin carcinogenesis Carcinogenesis 1997 18 15 21 9054584 10.1093/carcin/18.1.15
Laird DW Fistouris P Batist G Alpert L Huynh HT Carystinos GD Alaoui-Jamali MA Deficiency of connexin43 gap junctions is an independent marker for breast tumors Cancer Res 1999 59 4104 4110 10463615
Saunders MM Seraj MJ Li Z Zhou Z Winter CR Welch DR Donahue HJ Breast cancer metastatic potential correlates with a breakdown in homospecific and heterospecific gap junctional intercellular communication Cancer Res 2001 61 1765 1767 11280719
Zhang ZQ Zhang W Wang NQ Bani-Yaghoub M Lin ZX Naus CC Suppression of tumorigenicity of human lung carcinoma cells after transfection with connexin43 Carcinogenesis 1998 19 1889 1894 9854998 10.1093/carcin/19.11.1889
Huang RP Fan Y Hossain MZ Peng A Zeng ZL Boynton AL Reversion of the neoplastic phenotype of human glioblastoma cells by connexin 43 (cx43) Cancer Res 1998 58 5089 5096 9823317
Yano T Hernandez-Blazquez FJ Omori Y Yamasaki H Reduction of malignant phenotype of HEPG2 cell is associated with the expression of connexin 26 but not connexin 32 Carcinogenesis 2001 22 1593 1600 11576997 10.1093/carcin/22.10.1593
Mesnil M Krutovskikh V Piccoli C Elfgang C Traub O Willecke K Yamasaki H Negative growth control of HeLa cells by connexin genes: connexin species specificity Cancer Res 1995 55 629 639 7834634
Moorby C Patel M Dual functions for connexins: Cx43 regulates growth independently of gap junction formation Exp Cell Res 2001 271 238 248 11716536 10.1006/excr.2001.5357
Krutovskikh VA Troyanovsky SM Piccoli C Tsuda H Asamoto M Yamasaki H Differential effect of subcellular localization of communication impairing gap junction protein connexin43 on tumor cell growth in vivo Oncogene 2000 19 505 513 10698520 10.1038/sj.onc.1203340
Qin H Shao Q Curtis H Galipeau J Belliveau DJ Wang T Alaoui-Jamali MA Laird DW Retroviral delivery of connexin genes to human breast tumor cells inhibits in vivo tumor growth by a mechanism that is independent of significant gap junctional intercellular communication J Biol Chem 2002
Lucke T Choudhry R Thom R Selmer IS Burden AD Hodgins MB Upregulation of connexin 26 is a feature of keratinocyte differentiation in hyperproliferative epidermis, vaginal epithelium, and buccal epithelium J Invest Dermatol 1999 112 354 361 10084314 10.1046/j.1523-1747.1999.00512.x
Habermann H Ray V Habermann W Prins GS Alterations in gap junction protein expression in human benign prostatic hyperplasia and prostate cancer J Urol 2002 167 655 660 11792947 10.1097/00005392-200202000-00057
Sawey MJ Goldschmidt MH Risek B Gilula NB Lo CW Perturbation in connexin 43 and connexin 26 gap-junction expression in mouse skin hyperplasia and neoplasia Mol Carcinog 1996 17 49 61 8890953 10.1002/(SICI)1098-2744(199610)17:2<49::AID-MC1>3.0.CO;2-O
Jamieson S Going JJ D'Arcy R George WD Expression of gap junction proteins connexin 26 and connexin 43 in normal human breast and in breast tumours J Pathol 1998 184 37 43 9582525 10.1002/(SICI)1096-9896(199801)184:1<37::AID-PATH966>3.0.CO;2-D
Hirschi KK Xu CE Tsukamoto T Sager R Gap junction genes Cx26 and Cx43 individually suppress the cancer phenotype of human mammary carcinoma cells and restore differentiation potential Cell Growth Differ 1996 7 861 870 8809403
Brissette JL Kumar NM Gilula NB Dotto GP The tumor promoter 12-O-tetradecanoylphorbol-13-acetate and the ras oncogene modulate expression and phosphorylation of gap junction proteins Mol Cell Biol 1991 11 5364 5371 1656230
Budunova IV Carbajal S Slaga TJ The expression of gap junctional proteins during different stages of mouse skin carcinogenesis Carcinogenesis 1995 16 2717 2724 7586191
Risek B Pozzi A Gilula NB Modulation of gap junction expression during transient hyperplasia of rat epidermis J Cell Sci 1998 111 ( Pt 10) 1395 1404 9570757
Budunova IV Carbajal S Slaga TJ Effect of diverse tumor promoters on the expression of gap-junctional proteins connexin (Cx)26, Cx31.1, and Cx43 in SENCAR mouse epidermis Mol Carcinog 1996 15 202 214 8597533 10.1002/(SICI)1098-2744(199603)15:3<202::AID-MC6>3.0.CO;2-J
Lin JH Takano T Cotrina ML Arcuino G Kang J Liu S Gao Q Jiang L Li F Lichtenberg-Frate H Haubrich S Willecke K Goldman SA Nedergaard M Connexin 43 enhances the adhesivity and mediates the invasion of malignant glioma cells J Neurosci 2002 22 4302 4311 12040035
Oliveira R Christov C Guillamo JS Debouard S Palfi S Venance L Tardy M Peschanski M Contribution of gap junctional communication between tumor cells and astroglia to the invasion of the brain parenchyma by human glioblastomas BMC Cell Biol 2005 6 7 15715906 10.1186/1471-2121-6-7
Ito A Katoh F Kataoka TR Okada M Tsubota N Asada H Yoshikawa K Maeda S Kitamura Y Yamasaki H Nojima H A role for heterologous gap junctions between melanoma and endothelial cells in metastasis J Clin Invest 2000 105 1189 1197 10791993
Pollmann MA Shao Q Laird DW Sandig M Connexin 43 mediated gap junctional communication enhances breast tumor cell diapedesis in culture Breast Cancer Res 2005 7 R522 R534 15987459 10.1186/bcr1042
Ito A Koma YI Uchino K Okada T Ohbayashi C Tsubota N Okada M Increased expression of connexin 26 in the invasive component of lung squamous cell carcinoma: Significant correlation with poor prognosis Cancer Lett 2005
zur Hausen H Papillomaviruses and cancer: from basic studies to clinical application Nature Rev Cancer 2002 2 342 350 12044010 10.1038/nrc798
Longworth MS Laimins LA Pathogenesis of human papillomaviruses in differentiating epithelia Microbiol Mol Biol Rev 2004 68 362 372 15187189 10.1128/MMBR.68.2.362-372.2004
McNutt NS Weinstein RS Carcinoma of the cervix: deficiency of nexus intercellular junctions Science 1969 165 597 599 5815840
McNutt NS Hershberg RA Weinstein RS Further observations on the occurrence of nexuses in benign and malignant human cervical epithelium J Cell Biol 1971 51 805 825 4942777 10.1083/jcb.51.3.805
King TJ Fukushima LH Hieber AD Shimabukuro KA Sakr WA Bertram JS Reduced levels of connexin43 in cervical dysplasia: inducible expression in a cervical carcinoma cell line decreases neoplastic potential with implications for tumor progression Carcinogenesis 2000 21 1097 1109 10836996 10.1093/carcin/21.6.1097
Ennaji MM Schwartz JL Mealing G Belbaraka L Parker C Parentaux M Jouishomme H Arella M Whitfield JF Phipps J Alterations in cell-cell communication in human papillomavirus type 16 (HPV16) transformed rat myoblasts Cell Mol Biol (Noisy -le-grand) 1995 41 481 498 7549785
Oelze I Kartenbeck J Crusius K Alonso A Human papillomavirus type 16 E5 protein affects cell-cell communication in an epithelial cell line J Virol 1995 69 4489 4494 7769709
Tomakidi P Cheng H Kohl A Komposch G Alonso A Connexin 43 expression is downregulated in raft cultures of human keratinocytes expressing the human papillomavirus type 16 E5 protein Cell Tissue Res 2000 301 323 327 10955727 10.1007/s004410000231
Faccini AM Cairney M Ashrafi GH Finbow ME Campo MS Pitts JD The bovine papillomavirus type 4 E8 protein binds to ductin and causes loss of gap junctional intercellular communication in primary fibroblasts J Virol 1996 70 9041 9045 8971040
Aasen T Hodgins MB Edward M Graham SV The relationship between connexins, gap junctions, tissue architecture and tumour invasion, as studied in a novel in vitro model of HPV-16-associated cervical cancer progression Oncogene 2003 22 7969 7980 12970745 10.1038/sj.onc.1206709
Middleton K Peh W Southern S Griffin H Sotlar K Nakahara T El-Sherif A Morris L Seth R Hibma M Jenkins D Lambert P Coleman N Doorbar J Organization of human papillomavirus productive cycle during neoplastic progression provides a basis for selection of diagnostic markers J Virol 2003 77 10186 10201 12970404 10.1128/JVI.77.19.10186-10201.2003
Schipper RG Romain N Otten AA Tan J Lange WP Verhofstad AA Immunocytochemical detection of ornithine decarboxylase J Histochem Cytochem 1999 47 1395 1404 10544213
Rouan F White TW Brown N Taylor AM Lucke TW Paul DL Munro CS Uitto J Hodgins MB Richard G trans-dominant inhibition of connexin-43 by mutant connexin-26: implications for dominant connexin disorders affecting epidermal differentiation J Cell Sci 2001 114 2105 2113 11493646
Di WL Rugg EL Leigh IM Kelsell DP Multiple epidermal connexins are expressed in different keratinocyte subpopulations including connexin 31 J Invest Dermatol 2001 117 958 964 11676838 10.1046/j.0022-202x.2001.01468.x
Neijssen J Herberts C Drijfhout JW Reits E Janssen L Neefjes J Cross-presentation by intercellular peptide transfer through gap junctions Nature 2005 434 83 88 15744304 10.1038/nature03290
|
16091133
|
PMC1198252
|
CC BY
|
2021-01-04 16:36:35
|
no
|
Mol Cancer. 2005 Aug 9; 4:31
|
utf-8
|
Mol Cancer
| 2,005 |
10.1186/1476-4598-4-31
|
oa_comm
|
==== Front
Mol PainMolecular Pain1744-8069BioMed Central London 1744-8069-1-231611148610.1186/1744-8069-1-23Short ReportProlonged membrane potential depolarization in cingulate pyramidal cells after digit amputation in adult rats Wu MF [email protected] ZP [email protected] M [email protected] ZC [email protected] Department of Anatomy & Cell Biology, Indiana University School of Medicine, Indianapolis, IN, 46202, USA2 University of Texas Southwestern Medical Center, Dallas, TX, 75390, USA3 Department of Physiology, Faculty of Medicine, University of Toronto, University of Toronto Centre for the Study of Pain, Toronto, M5S 1A8, Canada2005 19 8 2005 1 23 23 5 5 2005 19 8 2005 Copyright © 2005 Wu et al; licensee BioMed Central Ltd.2005Wu et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
The anterior cingulate cortex (ACC) plays an important role in higher brain functions including learning, memory, and persistent pain. Long-term potentiation of excitatory synaptic transmission has been observed in the ACC after digit amputation, which might contribute to plastic changes associated with the phantom pain. Here we report a long-lasting membrane potential depolarization in ACC neurons of adult rats after digit amputation in vivo. Shortly after digit amputation of the hind paw, the membrane potential of intracellularly recorded ACC neurons quickly depolarized from ~-70 mV to ~-15 mV and then slowly repolarized. The duration of this amputation-induced depolarization was about 40 min. Intracellular staining revealed that these neurons were pyramidal neurons in the ACC. The depolarization is activity-dependent, since peripheral application of lidocaine significantly reduced it. Furthermore, the depolarization was significantly reduced by a NMDA receptor antagonist MK-801. Our results provide direct in vivo electrophysiological evidence that ACC pyramidal cells undergo rapid and prolonged depolarization after digit amputation, and the amputation-induced depolarization in ACC neurons might be associated with the synaptic mechanisms for phantom pain.
==== Body
The anterior cingulate cortex (ACC) is involved in sensory signal processing such as integration of general affect, cognition, and pain unpleasantness [1-6]. Neurons in the ACC respond to peripheral nociceptive stimulation [7-10]. It has been shown that digit amputation causes a long-lasting potentiation of the ACC responses to peripheral electrical stimulation, which might contribute to the phantom limb pain [10,11]. Phantom limb pain is experienced in a limb that is no longer present. About 50–80% amputees suffer from phantom limb pain [12,13]. Recent studies have indicated that cortical reorganization occurs following digit or limb amputation [14,15], which correlates with phantom limb pain or phantom limb sensation in most amputees [12,13]. However, less information is available about the possible early changes within the ACC after amputation. Recent studies using in vitro brain slice preparations showed that digit amputation in rats caused the loss of long-term depression (LTD) in the ACC [7]. Here we employed intracellular recordings from anesthetized adult rats and reported a prolonged membrane depolarization in ACC neurons after digit amputation of the hind paw.
Male adult Wistar rats (200–350 g) were anesthetized with 1–2% halothane. The core body temperature was maintained at 37°C with a heating pad. In vivo intracellular recording was performed as previously described [16,17]. Briefly, the animals were placed in a stereotaxic frame. A craniotomy was performed in the region above the left ACC (AP: 1.0–3.7 mm, RL: 0.5–1.0 mm, H: 1.3–3.0 mm, Bregma) [18]. The intracellular recording electrodes had a tip resistance of 40–70 M_ as filled with a solution of 4% neurobiotin (Vector, CA, USA) in 2 M potassium acetate. The microelectrode was advanced slowly into the left ACC to impale neurons. After impalement, the neurons with a stable membrane potential of -60 mV or greater and action potential amplitudes of at least 60 mV were selected for further study. After a baseline recording, the 3rd digit of the hind paw was amputated with deep anesthesia (2–3% halothane). In some experiments, extracellular recording was simultaneously conducted in the ACC with intracellular recording. The extracellular recording electrodes had a tip resistance of ~4 M_ as filled with a solution of 2 M sodium chloride. One femoral vein was cannulated in some animals for drug delivery. Data were digitized and stored with a Macintosh computer using the data acquisition program Axodata (Axon Instruments, CA, USA). Values were presented as mean ± S.E.M. Analysis of variance (ANOVA) followed by post-hoc Fisher's test was used for statistical analysis. Changes were considered significant if P < 0.05.
After each successful recording, neurobiotin was iontophoresed into the cell by passing depolarizing current pulses (2 Hz, 300 ms, 1.0–1.5 nA) for 10–20 min. At the end of the experiment, the animal was deeply anesthetized and perfused transcardially with 0.01 M phosphate-buffered saline followed by 4% paraformaldehyde. The brain was removed and stored in fixative overnight. Coronal sections were cut at 80 μm thickness using a Vibratome (Ted Pella, CA, USA) and incubated in 0.1% horseradish peroxidase-conjugated avidin-D (Vector, CA, USA) in 0.01 M potassium phosphate-buffered saline (KPBS, pH 7.4) with 0.5% Triton X-100 for 6–8 h at room temperature. After detection of peroxidase activity with 3,3'-diaminobenzidine, the sections were examined in KPBS. The sections containing labeled neurons were mounted on gelatin-coated slides and processed for light microscopy.
A total of 77 rats were used in the present study. Among these animals, intracellular and extracellular recording were performed simultaneously in 38 rats. In order to compare electrophysiological responses in ACC neurons before and after digit amputation, only one neuron was recorded from each animal. All intracellularly stained neurons were successfully recovered, among which 36 were pyramidal neurons in ACC. Two neurons were located in other regions of the brain and were excluded from the present study. An example of intracellularly stained ACC pyramidal neurons is presented in Fig 1. Two neurons were excluded from the present study because they were identified as neurons in other regions of the cortex. Because maintaining a prolonged intracellular recording is extremely difficult, extracellular recording alone was performed in the rest of animals (n = 39). The DC potential response recorded by extracellular recording exhibited the same profiles as that of intracellular recording, so the data from intracellular recordings and extracellular recordings were pooled together.
Figure 1 Prolonged membrane potential change of ACC neurons after digit amputation. A. Light photomicrograph of a pyramidal neuron (square) in the ACC intracellularly stained with neurobiotin after recording. B. High magnification of the labeled pyramidal neuron in Figure A. C. Representative recordings showing the response of ACC neuron to the contralateral third digit amputation. The upper panel is the membrane potential intracellularly recorded from a pyramidal neuron, the middle panel is the intracellularly applied hyperpolarizing current pulses (1 Hz, -0.5 nA, 200 ms), the lower panel is the simultaneous extracellular recording of DC potential in the ACC region. The arrows indicate the time of digit amputation. Approximately 2 min after amputation, the baseline membrane potentials from both recording quickly depolarize for approximately 50 mV and gradually returned to the control level in about 55 min.
After digit amputation, a dramatic membrane potential depolarization occurred in approximately 70% of the neurons (or animals) recorded in ACC (52/75). The depolarization occurred shortly after amputation with a latency of 0.52 ± 0.08 min (n = 20) and quickly reached the maximal amplitude in approximately 2 minutes. The amplitude of this amputation-induced depolarization was 53.5 ± 4.0 mV (n = 24) with the membrane potential shifted from -71.9 ± 1.2 mV to -15.1 ± 5.3 mV from intracellular recordings (n = 15) and from 0 mV to 50.1 ± 7.0 mV from extracellular recordings (n = 9). The membrane potential slowly repolarized and returned to the control levels. The duration of this depolarization was 43.0 ± 4.5 min (n = 12). During intracellular recording, hyperpolarizing current pulses were delivered to monitor the membrane input resistance, which was indicated by the amplitude of membrane potential deflection. No obvious changes in membrane input resistance were detected during the amputation-induced depolarization. Representative traces of simultaneous intracellular and extracellular recording of amputation-induced depolarization is shown in Fig 1C. The onset, time course, and the amplitude of these two traces were almost identical.
To elucidate the possible mechanisms underlying the amputation-induced depolarization, lidocaine (2%, 1 ml, s.c.) was locally injected into the right hind paw 5 min before the digit amputation. In animals with lidocaine injection, the amplitude of depolarization significantly reduced to 34.0 ± 6.5 mV (n = 10, P < 0.01) and the duration dramatically shortened to 13.2 ± 2.8 ms (n = 8, P < 0.01). However, the latency of the onset of depolarization in lidocaine treated animals remained about the same as of control ones (Table 1). These results suggest that the amputation-induced depolarization is associated with the nociceptive stimulation.
Table 1 Effects of lidocaine and glutamate receptor antagonists on membrane depolarization
Control Lidocaine MK-801
Latency 0.52 ± 0.08 (n = 20) 0.48 ± 0.08 (n = 10) 1.53 ± 0.15* (n = 10)
Amplitude (mV) 53.5 ± 4.0 (n = 24) 34.0 ± 6.5* (n = 10) 28.5 ± 2.1* (n = 10)
Duration 43.0 ± 4.5 (n = 12) 13.2 ± 2.8* (n = 8) 23.3 ± 6.0* (n = 8)
*, P < 0.01 compared with control
It has been shown that the glutamatergic activation is involved in the pain signal processing in ACC [1,19,20]. To reveal the involvement of glutamate receptors in amputation-induced depolarization, NMDA receptor or non-NMDA receptor antagonists were applied intravenously 5 min before the digit amputation. In animals received NMDA receptor blocker MK801 (1 mg/kg), the amplitude of depolarization significantly decreased to 28.5 ± 2.1 mV (n = 10, P < 0.01) and the duration reduced to 23.3 ± 6.0 min (n = 8, P < 0.01). In contrast, the latency to the onset of depolarization was significantly prolonged to 1.53 ± 0.15 min (n = 10, P < 0.01) (Table 1). These data indicate that the amputation-induced depolarization is mediated by NMDA receptors.
Accumulated evidence indicates that the ACC is involved in pain processing. Surgical ablation or electrolytic lesion of the ACC reduces pain related unpleasantness [2]. Activation of opioid receptors in the ACC is accompanied by a reduction in pain [4]. A number of functional imaging studies have shown consistently the ACC activation during application of noxious stimuli to various parts of the body [5,21,22]. Recent reports demonstrate that the digit amputation causes the enhancement of sensory responses in the ACC and the loss of LTD in in vitro ACC slices for at least 2 weeks [7,10]. Interestingly, in the present study, we observed a dramatic membrane depolarization in ACC neurons after digit amputation. Dramatic membrane depolarization has been observed during cortical spreading depression or cerebral ischemia [23,24]. Cortical spreading depression could be triggered by chemical, electrical or mechanical stimulation [25] while anoxic or ischemic depolarization is mainly caused by the failure of Na+-K+ pump due to energy depletion [26]. The depolarization in ACC neurons after amputation is most likely elicited by nociceptive stimulation as local anesthesia could significantly reduce the size of the depolarization. However, local application of lidocaine couldn't completely block the amputation-induced depolarization. Besides the possible insufficient dosage, it is possible that subcutaneous injection of lidocaine fails to block the sensory afferent fibers from the deep structures such as bones and joints.
Glutamate is a major excitatory neurotransmitter in the central nervous system including the ACC [7]. Growing evidence has recently shown that NMDA receptors are involved in synaptic plasticity related to injury [2,27]. NMDA receptors are highly expressed in the ACC [2]. It has been demonstrated that NMDA receptors play important roles in mediation of pain signal processing in the ACC [7,28]. Data from the present study suggest that activation of NMDA receptor is mainly responsible to the depolarization elicited by digit amputation. Coincided with our observation, NMDA receptor antagonist MK801 blocks the cortical spreading depression [29]. It is conceivable that membrane potential depolarization induced by amputation results in the enhancement of Ca2+ influx through voltage-dependent Ca2+ channels and NMDA receptors, which might initiate the intracellular events leading to the alteration of synaptic transmission. A recent study showed that calcium-stimulated adenylyl cyclase including AC1 and AC8 may play critical roles in inducing long-term potentiation in the ACC neurons [30], behavioral nociceptive responses to inflammation and tissue injury as well as allodynia related to nerve injury [31]. The prolonged postsynaptic membrane depolarization in the ACC pyramidal cells provides a novel induction mechanism for triggering long-term plastic changes within the ACC such as LTP and LTD. These rapid onset and prolonged changes in excitatory synaptic transmission activate a series of signaling pathways including stimulating new protein synthesis that may play important roles in amputation related cortical reorganization as reported in monkeys and humans.
List of abbreviations
NMDA: N-methyl-D-aspartate
AMPA: α-amino-3-hydroxy-5-methyl-isoxozole propionic acid
KA: kainate MK801: (+)-5-methyl-10, 11-dihydro-5H-dibenzo [a, d] cyclohepten-5, 10-imine hydrogen maleate
Acknowledgements
This work was supported in part by funding from the EJLB-CIHR Michael Smith Chair in Neurosciences and Mental Health in Canada, Canada Research Chair, NeuroScience Canada and CIHR to Dr. Min Zhuo, and NIH NINDS 42722 to Dr. Min Zhuo and Zao C. Xu.
==== Refs
Tang J Ko S Ding HK Qiu CS Calejesan AA Zhuo M Pavlovian fear memory induced by activation in the anterior cingulate cortex Mol Pain 2005 1 6 15813993 10.1186/1744-8069-1-6
Zhuo M Canadian Association of Neuroscience review: Cellular and synaptic insights into physiological and pathological pain. EJLB-CIHR Michael Smith Chair in Neurosciences and Mental Health lecture Can J Neurol Sci 2005 32 27 36 15825543
Rainville P Brain mechanisms of pain affect and pain modulation Curr Opin Neurobiol 2002 12 195 204 12015237 10.1016/S0959-4388(02)00313-6
Zubieta JK Smith YR Bueller JA Xu Y Kilbourn MR Jewett DM Meyer CR Koeppe RA Stohler CS Regional mu opioid receptor regulation of sensory and affective dimensions of pain Science 2001 293 311 315 11452128 10.1126/science.1060952
Rainville P Duncan GH Price DD Carrier B Bushnell MC Pain affect encoded in human anterior cingulate but not somatosensory cortex Science 1997 277 968 971 9252330 10.1126/science.277.5328.968
Zhuo M Central plasticity in pathological pain Novartis Found Symp 2004 261 132 45; discussion 145-54 15469048
Wei F Li P Zhuo M Loss of synaptic depression in mammalian anterior cingulate cortex after amputation J Neurosci 1999 19 9346 9354 10531439
Hutchison WD Davis KD Lozano AM Tasker RR Dostrovsky JO Pain-related neurons in the human cingulate cortex Nat Neurosci 1999 2 403 405 10321241 10.1038/8065
Sikes RW Vogt BA Nociceptive neurons in area 24 of rabbit cingulate cortex J Neurophysiol 1992 68 1720 1732 1479441
Wei F Zhuo M Potentiation of sensory responses in the anterior cingulate cortex following digit amputation in the anaesthetised rat J Physiol 2001 532 823 833 11313449 10.1111/j.1469-7793.2001.0823e.x
Zhuo M No pain, no gains ScientificWorldJournal 2001 1 204 206 12805672
Karl A Muhlnickel W Kurth R Flor H Neuroelectric source imaging of steady-state movement-related cortical potentials in human upper extremity amputees with and without phantom limb pain Pain 2004 110 90 102 15275756 10.1016/j.pain.2004.03.013
Birbaumer N Lutzenberger W Montoya P Larbig W Unertl K Topfner S Grodd W Taub E Flor H Effects of regional anesthesia on phantom limb pain are mirrored in changes in cortical reorganization J Neurosci 1997 17 5503 5508 9204932
Florence SL Taub HB Kaas JH Large-scale sprouting of cortical connections after peripheral injury in adult macaque monkeys Science 1998 282 1117 1121 9804549 10.1126/science.282.5391.1117
Merzenich M Long-term change of mind Science 1998 282 1062 1063 9841454 10.1126/science.282.5391.1062
Wei F Xu ZC Qu Z Milbrandt J Zhuo M Role of EGR1 in hippocampal synaptic enhancement induced by tetanic stimulation and amputation J Cell Biol 2000 149 1325 1334 10871275 10.1083/jcb.149.7.1325
Xu ZC Pulsinelli WA Electrophysiological changes of CA1 pyramidal neurons following transient forebrain ischemia: an in vivo intracellular recording and staining study J Neurophysiol 1996 76 1689 1697 8890285
Paxinos G Watson C The rat brain in stereotaxic coordinates 1986 2nd San Diego, Academic Press xxvi, [237] p. of plates
Johansen JP Fields HL Glutamatergic activation of anterior cingulate cortex produces an aversive teaching signal Nat Neurosci 2004 7 398 403 15004562 10.1038/nn1207
Lei LG Sun S Gao YJ Zhao ZQ Zhang YQ NMDA receptors in the anterior cingulate cortex mediate pain-related aversion Exp Neurol 2004 189 413 421 15380491 10.1016/j.expneurol.2004.06.012
Craig AD Reiman EM Evans A Bushnell MC Functional imaging of an illusion of pain Nature 1996 384 258 260 8918874 10.1038/384258a0
Kwan CL Crawley AP Mikulis DJ Davis KD An fMRI study of the anterior cingulate cortex and surrounding medial wall activations evoked by noxious cutaneous heat and cold stimuli Pain 2000 85 359 374 10781909 10.1016/S0304-3959(99)00287-0
Leao AAP Spreading depression of activity in the cerebral cortex J Neurophysiology 1944 7 359 390
Xu ZC Neurophysiological changes of spiny neurons in rat neostriatum after transient forebrain ischemia: an in vivo intracellular recording and staining study Neuroscience 1995 67 823 836 7545798 10.1016/0306-4522(95)00096-2
Bures J Burseova O Krivaned J The mechanism and applications of Leao's spreading depression of electroencephalographic activity 1974 New York, Academic
Lipton P Ischemic cell death in brain neurons Physiol Rev 1999 79 1431 1568 10508238
Zhuo M Glutamate receptors and persistent pain: targeting forebrain NR2B subunits Drug Discov Today 2002 7 259 267 11839523 10.1016/S1359-6446(01)02138-9
Wei F Wang GD Kerchner GA Kim SJ Xu HM Chen ZF Zhuo M Genetic enhancement of inflammatory pain by forebrain NR2B overexpression Nat Neurosci 2001 4 164 169 11175877 10.1038/83993
Lauritzen M Hansen AJ The effect of glutamate receptor blockade on anoxic depolarization and cortical spreading depression J Cereb Blood Flow Metab 1992 12 223 229 1312539
Liauw J Wu LJ Zhuo M Calcium-stimulated adenylyl cyclases required for long-term potentiation in the anterior cingulate cortex J Neurophysiol 2005
Wei F Qiu CS Kim SJ Muglia L Maas JW Pineda VV Xu HM Chen ZF Storm DR Muglia LJ Zhuo M Genetic elimination of behavioral sensitization in mice lacking calmodulin-stimulated adenylyl cyclases Neuron 2002 36 713 726 12441059 10.1016/S0896-6273(02)01019-X
|
16111486
|
PMC1198253
|
CC BY
|
2021-01-04 16:40:08
|
no
|
Mol Pain. 2005 Aug 19; 1:23
|
utf-8
|
Mol Pain
| 2,005 |
10.1186/1744-8069-1-23
|
oa_comm
|
==== Front
Nutr JNutrition Journal1475-2891BioMed Central London 1475-2891-4-221608079110.1186/1475-2891-4-22ResearchEvaluation of the functional efficacy of an antioxidative probiotic in healthy volunteers Songisepp Epp [email protected] Jaak [email protected] Tiiu [email protected]ändar Reet [email protected]ütt Pirje [email protected] Mihkel [email protected] Marika [email protected] Department of Microbiology, University of Tartu, 50411 Tartu, Estonia2 Department of Biochemistry, University of Tartu, 50411 Tartu, Estonia2005 4 8 2005 4 22 22 18 4 2005 4 8 2005 Copyright © 2005 Songisepp et al; licensee BioMed Central Ltd.2005Songisepp et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
In persons without clinical symptom it is difficult to assess an impact of probiotics regarding its effect on health. We evaluated the functional efficacy of the probiotic Lactobacillus fermentum ME-3 in healthy volunteers by measuring the influence of two different formulations on intestinal lactoflora, fecal recovery of the probiotic strain and oxidative stress markers of blood and urine after 3 weeks consumption.
Methods
Two 3-week healthy volunteer trials were performed. Open placebo controlled (OPC) study participants (n = 21) consumed either goat milk or by L. fermentum ME-3 fermented goat milk (daily dose 11.8 log CFU (Colony Forming Units). Double blind randomised placebo controlled (DBRP) study participants (n = 24) received either capsules with L. fermentum ME-3 (daily of dose 9.2 CFU) or placebo capsules.
The faecal lactoflora composition, faecal ME-3 recovery, effect of the consumption on intestinal lactoflora, and oxidative stress markers of blood (total antioxidative activity; total antioxidative status and glutathione red-ox ratio) was measured.
Results
ME-3 was well tolerated and a significant increase in total faecal lactobacilli yet no predominance of ME-3 was detected in all study groups. Faecal recovery of ME-3 was documented by molecular methods only in fermented milk group, however the significant improvement of blood TAA (Total Antioxidative Activity) and TAS (Total Antioxidative Status) indices was seen both in case of fermented goat milk and capsules", yet glutathione re-ox ratio values decreased only in case of fermented by ME-3 goat milk.
Conclusion
The functional efficacy of both consumed formulations of an antioxidative probiotic L. fermentum ME-3 is proved by the increase of the intestinal lactobacilli counts providing putative defence against enteric infections and by reduction of the oxidative stress indices of blood and urine of healthy volunteers. In non-diseased host the probiotic health claims can be assessed by improvement of some measurable laboratory indices of well-established physiological functions of host, e.g. markers of antioxidative defence system.
==== Body
Background
Probiotics are defined as live microbial food supplements, which beneficially influence human health [1,2]. Widely accepted probiotics contain different lactic acid producing bacteria of human origin: bifidobacteria, lactobacilli or enterococci. Nowadays the concept of functional foods, incl. probiotic food and dietary supplements implies to their ability to beneficially influence body functions in order to improve the state of well-being and health and reduce the risk of disease [2,3]. The important areas of human physiology that are relevant to functional food science according ILSI (International Life Science Institute) and FUFOSE (The European Commission Concerted Action on Functional Food Science in Europe) are besides others, the modulation of basic metabolic processes and defence against high-grade oxidative stress [4,5].
Human nutrition is clearly associated with oxidative metabolism, which beside production of energy is involved in a number of vital functions of the host. For example, under physiological conditions the reactive species (including peroxyl radicals, nitric oxide radical, superoxide anion) figure a crucial role in primary immune defense of the human body by phagocytic cells against harmful microorganisms [6,7]. On the other hand, a prolonged excess of reactive species is highly damaging for the host biomolecules and cells, resulting in dysbalance of the functional antioxidative network of the organism and leading to substantial escalation of pathological inflammation [8].
By our knowledge, no systematic studies have been performed to approve the functional efficacy of different formulations of probiotic on the antioxidative defence system of a healthy human. In our previous study Lactobacillus fermentum ME-3 (DSM 14241) [9-11], expressed strong antimicrobial activity against Gram-positive and Gram-negative entero- and uropathogens [12,13]. The cells and cell lysate of L. fermentum ME-3 possessed substantial antioxidative potency [14]. In an animal experiment ME-3 suppressed the excessive oxidative stress reaction caused by Salmonella infection in intestinal mucosa and thus improved the gut mucosal antioxidative status [15]. The antioxidative effect of L. fermentum ME-3 on human body oxidative stress markers was confirmed by our pilot study with fermented goat milk [16].
The aim present study was to evaluate the functional efficacy of the probiotic strain L fermentum ME-3 in the human gastrointestinal tract (GIT) of healthy volunteers. The faecal recovery, effect of two different formulations on total faecal lactoflora and oxidative stress markers of blood and urine were compared after 3 weeks consumption.
Methods
Formulations
The efficacy of two different formulations (experimental fermented goat milk and probiotic capsules) on the human body oxidative stress markers was evaluated. Lactobacillus fermentum ME-3, a probiotic strain of healthy human intestinal origin [17], has been identified by biochemical and molecular methods [9]. The patent application has been submitted to the Estonian Patent Agency (Application No. 0356/01PV) as well as to the International Bureau of World Intellectual Property
Organization (WIPO) (Application No. WO03002131) [11]. L. fermentum ME-3 was used as freeze-dried powder in capsulated form and in fermented milk.
Capsules
Gelatine coated capsules were manufactured by the Tallinn Pharmaceutical Company. The freshly prepared probiotic capsules contained 9.0 logs CFU of L. fermentum ME-3 per capsule in addition to 250 mg of saccharose and microcellulose. Identical placebo capsules contained only saccharose and microcellulose. All capsules were stored at +4°C.
Survival of ME-3 in capsule
Survival of ME-3 in capsule was monitored during 12 months at +4°C. The content of one capsule was dissolved aseptically in 2 ml of 0.9% NaCl solution. The suspension was vortexed, serially diluted and seeded 0.1 ml on de Man-Rogosa-Sharpe (MRS) agar medium (OXOID, U.K.) and incubated 48 hours at 37°C microaerobically (10% CO2). The number of colonies was counted and the viable cell count in capsule was calculated.
Experimental fermented milk
Three different lots of experimental fermented goat milk was prepared for the 3-week trial with healthy volunteers in order to establish the health effects of ME-3 consumption. The study group was supplied with fresh product once a week. Experimental fermented milk was prepared as described previously [16] by combining the probiotic strain with two supportive lactobacilli cultures L. plantarum LB-4 and L. buchneri S-15. L. buchneri strain S1-5 decreased the specific taste of the goat milk. L. plantarum LB-4 was included as a strong producer of exopolysaccharides, which gives the fermented milk a cream-like consistence and delightful acidity. The goat milk was inoculated with 2% mixture of Lactobacillus strains and incubated at 37°C for 24 hours. Theproduct, ready to use, was cooled and stored at 4°C.
Survival of L. fermentum ME-3 in fermented goat milk
To measure the viable cell count of ME-3 in fermented goat milk, samples were taken at the end of fermentation (before cooling the product), and after 24 h, 32 h, 48 h and 7 days from the preparation, when the product was stored at 4°C. The amount of 0.5 ml of the fermented milk was serially diluted in saline and plated on MRS agar medium and incubated for 48 h at 37°C in microaerobic conditions.
Design of human volunteer trials
Two healthy volunteer (n 45) trials, particularly open placebo controlled (OPC) study and double blind randomized placebo controlled (DBRP) study were carried on to evaluate the functional efficacy of L. fermentum ME-3 in the human body. The inclusion criteria included the wish to participate, no known health problems, and no medical conditions requiring drug therapy, no other yogurts or no special diets. The subjects with a history of GIT disease, food allergy and acute infection, use of any antimicrobial agent within the last month or use of any regular concomitant medication including antioxidant vitamins and anti-inflammatory non-steroidal drugs were excluded.
The members of the trial were daily questioned about their general welfare, intestinal function (general welfare, gut gas production, stool frequency) and putative adverse effects. The withdrawal criteria from the trials included acute infections during the study. Reasons for dropout were the unwillingness to proceed with the study or relocation to new area. The blood samples (6 ml) from the antecubital vein, faecal and urine samples were collected before and at the end of all clinical trials. Participants of all trials gave informed consent to the study protocols approved by the Ethical Committee of Tartu University.
Open placebo controlled fermented goat milk trial
The study participants were 5 men and 16 women, mean age 50 years (range 35–60). During three weeks of the trial the study group (3 males and 13 females) consumed daily 150 ml fermented goat milk. The daily dose of probiotic Lactobacillus strain was 11.2 to 11.8 log CFU per person. The control group (1 male and 4 females) consumed the same dose of fresh goat milk.
Probiotic capsule trial
A DBRP study (ISRCTN 53164826) was carried out as follows. The study group consisted of 15 men and 9 women, mean age 52 years (range 40–60) allocated according to their wish to participate and randomly divided by an independent person and computer program for two groups. The study group members (8 males and 4 females) took three probiotic containing capsules (8.4 log CFU per capsule) two times daily (the daily dose 9.2 log CFU) during three weeks. The placebo group (7 males and 5 females) received identical capsules without the probiotic strain.
Faecal samples of all participants to assess change in faecal lactoflora and the persistence of the ingested probiotic strain were collected before and at the end of trial. Several laboratory indices of blood and urine were measured before and after the consumption of ME-3. Here we report on changes in human body oxidative stress markers as total antioxidative activity (TAA), total antioxidative status (TAS) and glutathione red-ox ratio (GSH/GSSG) from blood serum.
Microbiological analyses of faeces
The total count of lactobacilli and the count of L. fermentum were evaluated in faecal samples. The faecal samples were collected at day 0 and 21 in both trials. Samples were kept at -80°C before analyzed. Serial dilutions (10-2–10-9) of the weighed faecal samples were prepared with phosphate buffer (pH 7.2) and 0.05 ml of aliquots was seeded onto MRS agar medium [17]. The plates were incubated at 37°C for 4 days microaerobically in 10% CO2 environment (incubator IG 150, Jouan, France). The catalase negative colonies were selected on the basis of typical for LAB colony morphology, cells microscopy and Gram staining.
The count of Lactobacillus species was expressed in log10 colony forming units per gram faeces (log10 CFU/g) and percentage (relative share) in the total count of lactobacilli. The detection level of lactobacilli was a 3.0 log CFU/g faeces.
The relative amount of L. fermentum, colonizing the gastrointestinal tract of persons in the study groups was expressed as a proportion of the total count (%), using the Bioquant program [18]). The program gives output data for every microorganism as an absolute count (log10 CFU/g) and their percentage in the total count with its normal values.
AP-PCR typing
The putative ME-3 isolates were typed by arbitrarily primed polymerase chain reaction (AP-PCR). Genomic DNA was extracted from 24 h old cultures, cultivated on MRS agar microaerobically with the QIAamp DNA Mini Kit 50 (QIAGEN GmbH., Hilden, Germany) according to the manufacturers instructions. AP-PCR typing was done with two primers: ERIC1R (5'-ATGTAAGCTCCT GGGGATTCAC-3') and ERIC2 (5'-AAGTAAGTGACTGGGGTGAGCG -3') (DNA Technology A/S, Aarhus, Denmark). A 30 μl volume of reaction mixture consisted of 10 × PCR buffer (Fermentas, Vilnius, Lithuania), 2.5 mM MgCl2 (Fermentas, Vilnius, Lithuania), 200 μM deoxynucleoside triphosphate mixture (dATP, dGTP, dTTP and dCTP, Amersham Pharmacia Biotech, Freiburg, Germany) 0, 60 μg of each primer and 2.5 U Taq DNA Polymerase (Fermentas, Vilnius, Lithuania,) and 5 μl of extracted DNA according to Matsumiya et al. [19]. The PCR mixture was subjected to thermal cycling 35 cycles of denaturation at 95°C for 1 min, annealing at 35°C for 1 min, and extension at 74°C for 2 min, with a final extension at 74°C for 5 min with the PTC-200 thermal cycler (Eppendorf AG, Hamburg, Germany). The PCR products were separated by electrophoresis in a horizontal 2% agarose gel containing 0.1 μl/ml ethidium bromide in Tris-acetic acid-EDTA (TAE) buffer (40 mM Tris, 20 mM boric acid, 1 mM EDTA, pH 8.3) (Bio-Rad Laboratories, Hercules, USA) at constant voltage of 120 V. A 1 kb ladder (GeneRuler, Fermentas, Vilnius, Lithuania) was used as a base pair size marker. The banding patterns of isolates were visualized with UV light and compared with that of L. fermentum ME-3 strain.
Measurement of human body oxidative stress status
Blood serum was analysed for total antioxidative activity TAA, total antioxidative status TAS and glutathione red/ox ratio (oxidized glutathione and reduced glutathione, GSSG/GSH). TAA of the serum was assessed by the linolenic acid test (LA-test) described previously [16]. This test evaluates the ability of the sample to inhibit lipid peroxidation. TAS of the serum was measured with a commercially available kit (TAS, Randox Laboratories Ltd. Ardmore, UK) as described elsewhere [16], water-soluble vitamin E (Trolox) serving as a standard. This method is based on the inhibition of the absorbance of the ferrylmyoglobin radicals of 2,2'-azinobis-ethylbenzothiazoline 6-sulfonate (ABTS+) generated by activation of metmyoglobin peroxidase with H2O2.
The cellular oxidative stress markers as total glutathione and oxidized glutathione were measured using the method of Griffith [20] as described elsewhere [16]. The glutathione content was calculated on the basis of a standard curve generated with known concentration of glutathione. Amount of GSH (μg/ml) was calculated as a difference between the total glutathione and GSSG (total glutathione – GSSG). The glutathione red/ox ratio was expressed as GSSG/GSH.
Statistical Analysis
The computer program Sigma Stat for Windows 2.0 (Jandel Corporation, USA) was applied. The counts of faecal lactoflora were compared by using Student's t-test and Mann-Whitney rank sum test. Changes in oxidative stress markers of blood sera (TAA, TAS and glutathione red-ox ratio) were evaluated by Student's t-test, paired t-test and Mann-Whitney rank sum test. The choice of tests was made automatically according to the distribution of the data. Both microbial and biochemical markers were given as mean and standard deviation.
One-way ANOVA test was performed to compare the effect of different formulation on TAA, TAS and faecal lactoflora parameters.
Differences were considered statistically significant if the value was p < 0.05.
Results
Survival of ME-3 in formulations
In capsule after approximately 1-log drop after one week from the production of the capsules, the viable count of the probiotic strain remained stable at the level of 8.4 log CFU per capsule. Additional results have shown at +4°C the stability of the freeze-dried capsulated culture at least 17 months from the production.
In fermented goat-milk the cell count of the probiotic strain varied insignificantly from 9.0 to 9.7 log CFU/ml from one preparation to the other. The viable count of ME-3 in the fermented goat milk was found to remain stable at least during 7 days of storage at 4°C.
Human volunteer trials
No dropouts were registered during volunteer trials, yet one participant was withdrawn from the probiotic capsule trial due to acute respiratory viral infection. Besides, no adverse affects in general welfare or changes in GI functionality were assessed during the trial.
Changes in total LAB count
The consumption of both ME-3 fermented milk and ME-3 capsule significantly increased the total count of lactobacilli in faeces as compared to the initial levels (Fig. 1). In opposite, in the group of volunteers consuming non-fermented goat-milk there was even a decrease in total LAB counts during the 3-week trial and no changes were found in capsule placebo group.
Figure 1 Increase of total fecal counts of lactobacilli in healthy volunteers consuming of ME-3 in fermented goat milk and probiotic capsule. 1 – Day 0, 2 – Day 21 Significantly different from pre-treatment values (Student's t-test): * p < 0.05; Significantly different from control (Student's t-test): ‡ p = 0.01
Recovery of the probiotic strain
In goat milk group L. fermentum as a species appeared in fecal samples of all individuals (n = 16) after consumption of fermented goat milk (Table 1). The AP-PCR confirmed the recovery of ME-3 in the faeces of all study group members (Fig. 2). However, in different trials the administration of ME-3 strain did not lead to the predominance of Lactobacillus fermentum species (Table 1).In the probiotic capsule trial the strain ME-3 was not detectable between L. fermentum isolates by AP-PCR.
Table 1 Changes in fecal recovery of L. fermentum during healthy human volunteer trials
L. fermentum
Groups * Prevalence (%) † Count (log10) ‡Proportion (%)
Day 0 Day 21 Day 0 Day 21 Day 0 Day 21
Goat milk trial, ME-3 (n = 16) 25 (4/16) 100 (16/16) 7.0 ± 0.7 7.3 ± 1.4** 21 13
Control (n = 5) - 20 (1/5) - 3.6 - 28
Capsule trial, ME-3 (n = 11) 16.7(2/12) 33.3 (2/12) 4.3 ± 0.5 5.8 ± 1.6 4 9
Placebo (n = 12) 25 (3/12) 16.7 (2/12) 6.3 ± 2.5 8.0 ± 1.6 11 19
* Percentage of subjects with fecal L. fermentum inside the group
** Significantly different from the pre-treatment values (paired t-test): p < 0.001
† Median value ± SD
‡ Proportion of L. fermentum among fecal LAB
Figure 2 Confirmation for the survival of L. fermentum ME-3 in GIT in subjects receiving ME-3 fermented goat milk by AP-PCR in a horizontal 2% agarose gel. a) From the left: M – DNA 1 kb Ladder, Line 1 -L. bervis ATCC 14869, Line 2 – L. buchneri ATCC 4005, Line 3 – L. reuteri DSM 20016, Line 4 – L. fermentum ATCC 14931, Line 5 – L. fermentum ME-3 b) From the left: M – molecular weight marker, Line 1 ...16 – ME-3 like profiles from feces of goat milk trial study group participants
Antioxidative health effect of ME-3
The positive effect on the blood ox stress markers as TAA and TAS was seen in the case of both formulations (Fig. 3). Particularly, the additive increase in goat milk group was 6% and 9% respectively as compared to control group, however only 4% for TAA and 2.5% for TAS in probiotic capsule study group as compared to placebo.
Figure 3 Effect of ME-3 consumption in fermented goat milk and capsules on human blood oxidative stress markers a) TAA (%) and b) TAS(mmol/l). 1 – Day 0, 2 – Day 21 Significantly different from pre-treatment values: *p < 0.05 (paired t-test); **p ≤ 0.01 (Student's t-test and paired t-test);ME-3 goat milk effect different from the effect of the ME-3 in capsule-form (ANOVA): ‡p ≤ 0.001
The effect of goat-milk consumption on the TAA and TAS values was significantlyhigher (p < 0.001) than by the consumption of the capsulated probiotic (Fig. 3). The decrease of the glutathione red-ox ratio was significant in both groups: the study Group (from 0.15 ± 0.01 to 0.11 ± 0.04 μg/ml, p < 0.01) and control (from 0.14 ± 0.03 to 0.11 ± 0.02 μg/ml, p < 0.01) in the goat milk trial. When the probiotic was consumed in capsulated form, no significant decrease was noticed in the glutathione red-ox ratio.
Discussion
Our aim was to evaluate the functional efficacy of the antimicrobial and antioxidative probiotic L. fermentum ME-3 in normal population with variable food intake. First of all the safety of the L. fermentum strain ME-3 was confirmed as no adverse side effects were registered in volunteers. Even relatively high (>1011 CFU) doses of consumed ME-3 had no negative impact on the hosts' general well-being. Lactobacillus fermentum as species, used in various food applications, has a well-established history of safe use and is evaluated as GRAS according to the Food and Drug Administration of the USA [21].
Second, a clear improvement of laboratory indices of antioxidative defence system of a healthy host was documented, using both formulations as fermented by L. fermentum ME-3 goat-milk and probiotic capsules. This effect was simultaneous with the increase of intestinal lactoflora of healthy volunteers even without necessity for faecal recovery of the strain. In the human population, persons without clinical symptoms have still a quite different health status, including stability, capacity and potency of antioxidative defence to counteract sufficiently to oxidative stress-caused adverse effects [7]. If a probiotic is able to exhibit a positive functionality on oxidative stress-related indices, it helps both to stabilize and promote the potency of the whole body antioxidative defence system in subclinical situations without disease symptoms. That in turn may have an impact for lowering the risk of atherosclerotic damage of blood vessels associated with several cardiovascular and neurodegenerative diseases [22-24].
In our study of healthy volunteers for validation of the antioxidative functionality of probiotic, four well-known oxidative stress markers of blood were chosen [25-27].
The state of the lipid fraction (including also LDL) in the antioxidative defence system of the blood is evaluated by TAA. TAS on the other hand reflects more the antioxidativity of the water-soluble fraction of the human blood. Among the measured blood sera markers both the TAA and TAS values were also improved in the two different study groups. However, there was found a significantly lower improvement of TAA and TAS values in cases of capsule than fermented goat-milk where the recovery of the strains was assessed by AP-PCR.
Similarly, the reduction of the glutathione red-ox ratio was detected after the consumption of fermented by ME-3 goat-milk but not with the capsule. The crucial non-enzymatic cellular antioxidant is GSH [28] present in the millimolar range mainly in the red blood cells, liver, pancreas, kidneys, spleen, eyes, lungs and intestinal cells [29]. The oxidized form of glutathione becomes even at low concentrations toxic, and therefore in the cells the glutathione red-ox ratio is kept as low as possible. In the case of inflammation this balance is shifted towards the oxidized form, indicating non-physiological intracellular oxidative stress.
Thus, our study shows that there is a good association between the mode of formulation of probiotic and expression of its functional properties inside the healthy host. The antioxidative potential of the food supplement containing ME-3 was excellent, as reisolates of the strain from capsule expressed significantly higher TAA in comparison with the base values of the strain in vitro (data not shown). Unexpectedly, the shifts in the antioxidativity markers in blood serum of participants of the probiotic capsule trial were less pronounced in comparison with ME-3 fermented goat milk.
Particularly, the explanation for more expressed positive shifts in oxidative stress markers of volunteers of the fermented goat milk trial could be due to the synergistic effect of the probiotic and the substrate. Milk is not just a carrier for the probiotic Lactobacillus strain, but contains natural "lactogenic" factors like lactose, minerals, vitamins and other components that enhance the metabolic activity of ingested probiotic strain in GIT. Both fermented goat milk and goat milk elevated the values of TAA and TAS The goat milk contains different biomolecules (e.g. casomorphins, lactorphins, casokinins, etc) having certain antioxidative properties, which can contribute to consumers' plasma antioxidative capacity [30-36]. This was proved by some antioxidative effect also in persons consuming non-fermented goat milk. However, the elevation these indices were remarkably more expressed in the fermented goat milk group, thus the goat milk fermentation with L. fermentum ME-3 results in additive increase of total antioxidativity. Therefore, the provisional FAO regulations [37] suggesting the need for health claims by specified formulations of probiotic seem to be of the utmost importance.
Additionally, in our study with experimental fermented milk the average daily dose of L. fermentum ME-3 being 11.5 logs CFU was clearly higher than that of capsule (max 9.5 log CFU). It is possible that the dose excesses the amount of bacteria necessary for interacting with intestinal mucosa and the unattached lactobacilli are excreted with faeces. The finding of Saxelin and colleagues confirmed that the faecal recovery of the probiotic strain started from the consumption of more than 9.0 log CFU daily doses of capsulated LGG [38]. To our surprise, in the present study the similar dose did not result in faecal recovery of the strain.
It is possible that the ME-3 strain germinated mainly in some upper parts of intestinal tract where the advantageous conditions for survival and metabolic activity of probiotic lactobacilli were present. Using molecular tools, Marteau et al. showed that lactobacilli figuring only 7% of faecal microflora performed up to 30% of microbial communities in human colon [39]. If administered in lower quantities as in case of capsule trial, ME-3 did not reach the detectable level in faecal samples. Yet, its presence in gut was proved by the positive antioxidative health effect in blood but not in urine. Therefore it is understandable that the higher load of metabolically active probiotic bacteria in goat-milk resulted also in their faecal recovery and the highest impact on the oxidative stress indices.
Moreover, in our study the positive impact of ME-3 consumption on the host lactoflora was proved by the increase of faecal lactobacilli counts in all participants of human volunteer studies. In experimental settings the high counts of intestinal lactobacilli have been shown as an important defensive factor against enteric infections [40,41]. Though up to now the period of consumption of probiotics has not been defined, the 3-week ingestion of fermented goat-milk and capsule seemed enough for reaching the aims.
It is important to mention that after consumption of ME-3, a strain with high antagonistic activity, neither the species nor the strain predominated among total lactoflora. This shows a well-granted microbial balance inside the gut, which cannot be disturbed by high load of probiotic bacteria. Apparently, the interconnected advanced metabolism of large gut microbiota keeps the proportions of different species quite stable. Some other investigators have obtained similar results showing the proportional increase of different microbial populations (bifidobacteria, coliforms) after administration of Lactobacillus sp. probiotic [42,43].
Thus, the functional efficacy of different formulations of anti-infectious and antioxidative probiotic L. fermentum ME-3 were proved both by the increase of the lactobacilli counts providing putative defence against infectious agents in gut and by reduction of the oxidative stress indices of blood and urine of healthy volunteers. Further, studies evaluating the efficacy of ME-3 as adjunct to conventional therapy in patients with atherosclerotic damages and a high-grade oxidative stress are ongoing.
Conclusion
In non-diseased host, the probiotic health claims can be assessed by improvement of some measurable laboratory indices of well-established physiological functions of organism. In our case, the possibility for augmentation of the antioxidative defence system by the probiotic L. fermentum ME-3 in normal population can be proposed.
Competing interests
Marika Mikelsaar, Mihkel Zilmer, Tiiu Kullisaar, Heidi Annuk (Hynes) and Epp Songisepp are sharing the Estonian patent application: no. EE 2001 00356 29.06.01 and International Patent application: no. WO03002131.
Authors' contributions
Epp Songisepp, and Pirje Hütt have been in charge of the microbiological analysis. The former has been also in charge of analyzing the results and writing of the manuscript. Jaak Kals was responsible for performance and management of the volunteer trials. Tiiu Kullisaar has been in charge of the biochemical analysis and writing the manuscript. Mihkel Zilmer has conducted the biochemical estimations and writing the manuscript. Reet Mändar has been in charge of the molecular analysis and revising the manuscript. Marika Mikelsaar is the main conductor of the L. fermentum ME-3 research; for this paper she has been in charge of the clinical trial design and writing the manuscript.
Acknowledgements
The study was supported by a grant from the Estonian Science Foundation (base funding 0418 and 5327) and Estonian Technology Agency funding 01103 and EU QLRT-2001-00135.
We are sincerely grateful to Dr. Irja Lutsar for critical reading of the manuscript, Dr. Heidi Hynes, Mrs. Eha-Maie Laanes and Miss Enely Alas for excellent technical assistance.
==== Refs
McFarland L A review of evidences of health claims for biotherapeutic agents Microb Ecol Health Dis 2000 12 65 76 10.1080/089106000435446
Salminen S Human studies on probiotics: aspects of scientific documentation Scand J Nutr 2001 45 8 12
Hardy G Nutraceuticals and functional foods: introduction and meaning Nutr 2000 16 688 697 10.1016/S0899-9007(00)00332-4
ILSI Europe Scientific concepts of functional foods in Europe: consensus document Br J Nutr 1999 81 1 27
Ashwell M ILSI Europe Concise monograph series Concepts of functional foods ILSI Europe Concise monograph series 2002 40 S ILSI Press 7894 10
Diplock AT Charleux J-L Crozier-Willi G Kok FJ Rice-Evans C Roberfroid M Stahl W Viňa-Ribes J Functional food science and defence against reactive oxidative species Br J Nutr 1998 80 77S 112S
Halliwell B Gutteridge JMC Free radicals in biology and medicine 1999 New York: Oxford University Press
Petrof EO Kojima K Ropeleski MJ Musch MW Tao Y De Simone C Chang EB Probiotics inhibit nuclear factor kappa-B and induce heat shock proteins in colonic epithelial cells through proteasome inhibition Gastroenterol 2004 127 1474 1487 10.1053/j.gastro.2004.09.001
Annuk H Vainumäe I Aimla K Ausmees K Kullisaar T Kask S Mikelsaar M Antagonistic activity of Lactobacilli against urinary Escherichia coli Microecol Therapy 1999 28 169 175
Mikelsaar M Annuk H Shchepetova J Mändar R Sepp E Björkstén B Intestinal Lactobacilli of Estonian and Swedish children Microb Ecol Health Dis 2002 14 75 80 10.1080/08910600260081711
Mikelsaar M Zilmer M Kullisaar T Annuk H Songisepp E Strain of microorganism Lactobacillus fermentum ME-3 as novel antimicrobial and antioxidative probiotic International Patent application 2001 WO03002131
Annuk H Selection of medicinal plants and intestinal lactobacilli as antimicrobial components for functional foods PhD thesis 2002 Tartu University, Microbiology Department
Songisepp E Kullisaar T Hütt P Elias P Brilene T Zilmer M Mikelsaar M A New Probiotic Cheese with Antioxidative and Antimicrobial Activity J Dairy Sci 2004 87 2017 2023 15328213
Kullisaar T Zilmer M Mikelsaar M Vihalemm T Annuk H Kairane C Kilk A Two antioxidative lactobacilli strains as promising probiotics Int J Food Microbiol 2002 72 215 224 11845820 10.1016/S0168-1605(01)00674-2
Truusalu K Naaber P Kullisaar T Tamm H Mikelsaar R-H Zilmer K Rehema A Zilmer M Mikelsaar M The influence of antibacterial and antioxidative probiotic lactobacilli on gut mucosa in a mouse model of Salmonella infection Micr Ecol Health Dis 2004 16 180 187 10.1080/08910600410021783
Kullisaar T Songisepp E Mikelsaar M Zilmer K Vihalemm T Zilmer M Antioxidative probiotic fermented goats milk decreases oxidative stress- mediated atherogenicity in human subjects Br J Nutr 2003 90 449 456 12908907 10.1079/BJN2003896
Mikelsaar M Lenzner A Goljanova LA Methods for the estimation of the quantitative composition of fecal microflora Laboratornoje delo 1972 1 41 45
Mändar R Mändar H Mikelsaar M Bioquant- a program for evaluation of fecal microbiocenosos Clin Chem Lookout 1992 56 46
Matsumiya Y Kato N Watanabe K Molecular epidemiological study of vertical transmission of vaginal Lactobacillus species from mothers to newborn infants in Japanese, by arbitrarily primer polymerase chain reaction J Infect Chemother 2002 8 43 49 11957119
Griffith OW Determination of glutathione and glutathione disulfide using glutathione reductase and 2-vinylpyridine Anal Biochem 1980 106 207 212 7416462 10.1016/0003-2697(80)90139-6
Food and Drug Administration Rules and regulations
Witzum JL The oxidation hypothesis of atherosclerosis Lancet 1994 344 793 795 7916078 10.1016/S0140-6736(94)92346-9
Bogdanovic N Zilmer M Zilmer K Rehema A Karelson E The Swedish APP670/671 Alzheimer's disease mutation: The first evidence for strikingly increased oxidative injury in the temporal inferior cortex Demen Geriatr Cogn Disord 2001 12 364 370 10.1159/000051282
Pihl E Zilmer K Kullisaar T Kairane C Pulges A Zilmer M High-sensitive C-reactive protein level and oxidative stress-related status in former athletes in relation to traditional cardiovascular risk factors Atherosclerosis 2003 171 321 326 14644403 10.1016/j.atherosclerosis.2003.08.015
Hata AN Breyer RM Pharmacology and signaling of prostaglandin receptors: Multiple roles in inflammation and immune modulation Pharmacol Ther 2004 103 147 66 15369681 10.1016/j.pharmthera.2004.06.003
O'Brien RC Luo M Balazs N Mecuri J In vitro and vivo antioxidant properties of gliclazide J Diabetes Complic 2000 14 201 206 10.1016/S1056-8727(00)00084-2
Sanchez-Moreno C Dashe JF Scott T Thaler D Folstein MF Martin A Decreased levels of plasma vitamin C and increased concentration of inflammatory and oxidative stress markers after sroke Stroke 2003 35 163 14671251 10.1161/01.STR.0000105391.62306.2E
Pastore A Federici G Bertini E Piemonte F Analysis of glutathione: implication in redox and detoxification Clinica Chimica Acta 2003 333 19 39 10.1016/S0009-8981(03)00200-6
Meister A Anderson ME Glutathione Ann Rev Biochem 1983 52 711 760 6137189 10.1146/annurev.bi.52.070183.003431
Alichanidis E Polychronadou A Special features of dairy products from ewe and goat milk Proceeding of the IDF 1996 Cirval seminar on production and utilization of ewe and goat milk, held in Crete (Greece) 21 43
Nakagawa K Ninomiya M Okubo T Aoi N Junoja L Kim M Yamanaka K O'Brien RC Luo M Balazs N Mecuri J In vitro and vivo antioxidant properties of gliclazide J Diabetes Complications 2000 14 201 206 11004429 10.1016/S1056-8727(00)00084-2
Hancock REW Diamon G The role of cationic antimicrobial peptides in innate host defence Trends in Microbiology 2000 8 402 410 10989307 10.1016/S0966-842X(00)01823-0
Epand RM Vogel HJ Diversity of antimicrobial peptides and their mechanisms of action Biochim Biophys Acta 1999 1462 11 28 10590300
Meisel H Bockelmann W Konings WN, Kuipers OP, Huis in't Veld J Bioactive peptides encrypted in milk proteins: proteolytic activation and tropho- functional properties Proceedings of the sixth symposium on lactic acid bacteria: genetics, metabolism and applications 1999 Veldhoven: Kluver Academic Publishers 207 215
Bos C Gaudichon C Tomé D Nutritional and physiological criteria in the assessment of milk protein quality for humans J Am Coll Nutr 2000 19 191S 205S 10759143
Boland M MacGibbon A Hill J Designer milks for the new millennium Livestock Prod Sci 2001 72 99 109 10.1016/S0301-6226(01)00270-6
Food Agriculture Organization Guidelines for the evaluation of probiotics in food
Saxelin M Elo S Salminen S Vaapatalo H Dose response colonisation of faces after oral administration of Lactobacillus casei strain GG Micr Ecol Health Dis 1991 4 209 214
Marteau P Pochart P Dore J Bera-Maillet C Bernalier A Corthier G Comparative study of bacterial groups within the cecal and fecal microbiota Appl Env Microbiol 2001 67 4939 4942 11571208 10.1128/AEM.67.10.4939-4942.2001
Kelleher SL Casas I Carbajal N Lonnerdal B Supplementation of infant formula with the probiotic Lactobacillus reuteri and zinc: impact on entericinfection and nutrition in infant rhesus monkeys J Pediatr Gastr Nutr 2002 35 162 168 10.1097/00005176-200208000-00011
Cruchet S Obregon MC Salazar G Diaz E Gotteland M Effect of the ingestion of dietary product containing Lactobacillus johnsonii La1 on Helicobacter pylori colonization in children Nutr 2003 19 716 721 10.1016/S0899-9007(03)00109-6
Sepp E Mikelsaar M Salminen S Effect of administration of Lactobacillus casei strain GG on the gastrointestinal microbiota of newborns Micr Ecol Health Dis 1993 6 309 314
Morelli L Zonenschain D Callegari ML Grossi E Maisano F Fusillo M Assessment of a new symbiotic preparation in healthy volunteers: survival, persistence of probiotic strains and its effect on the indigenous flora Nutr J 2003 2
|
16080791
|
PMC1198254
|
CC BY
|
2021-01-04 16:39:31
|
no
|
Nutr J. 2005 Aug 4; 4:22
|
utf-8
|
Nutr J
| 2,005 |
10.1186/1475-2891-4-22
|
oa_comm
|
==== Front
Reprod HealthReproductive Health1742-4755BioMed Central London 1742-4755-2-41609552510.1186/1742-4755-2-4ResearchImproving survival rates of newborn infants in South Africa Pattinson Robert [email protected] David [email protected] David [email protected] Sithembiso [email protected] MRC Maternal and Infant Health Care Strategies Research Unit, Department of Obstetrics and Gynaecology, University of Pretoria, Pretoria, South Africa2 School of Child and Adolescent Health, Groote Schuur Hospital, University of Cape Town, Cape Town, South Africa3 Department of Paediatrics, Chris Hani Baragwanath Hospital, University of the Witwatersrand, Johannesburg, South Africa2005 11 8 2005 2 4 4 1 2 2005 11 8 2005 Copyright © 2005 Pattinson et al; licensee BioMed Central Ltd.2005Pattinson et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The number, rates and causes of early neonatal deaths in South Africa were not known. Neither had modifiable factors associated with these deaths been previously documented. An audit of live born infants who died in the first week of life in the public service could help in planning strategies to reduce the early neonatal mortality rate.
Methods
The number of live born infants weighing 1000 g or more, the number of these infants who die in the first week of life, the primary and final causes of these deaths, and the modifiable factors associated with them were collected over four years from 102 sites in South Africa as part of the Perinatal Problem Identification Programme.
Results
The rate of death in the first week of life for infants weighing 1000 g or more was unacceptably high (8.7/1000), especially in rural areas (10.42/1000). Intrapartum hypoxia and preterm delivery are the main causes of death. Common modifiable factors included inadequate staffing and facilities, poor care in labour, poor neonatal resuscitation and basic care, and difficulties for patients in accessing health care.
Conclusion
Practical, affordable and effective steps can be taken to reduce the number of infants who die in the first week of life in South Africa. These could also be implemented in other under resourced countries.
==== Body
Introduction
Of the approximately 130 million infants born worldwide each year, it is estimated that four million infants die during the first month of life [1]. The vast majority of these neonatal deaths occur in poor countries where standards of both maternal and newborn care are low. One of the Millenium Development Goals is to reduce the number of childhood deaths under the age of five years by two thirds from 95 per 1000 to 31 per 1000 by 2015 [2]. In South Africa, approximately 33% of deaths of under five-year-olds, 44% of infant deaths (before one year), and 87% of neonatal deaths (in the first month) occur during the first seven days after birth [3,4]. If the Millenium Development Goal of significantly reducing childhood deaths is to be achieved, a substantial reduction in early neonatal deaths will be required, especially in poor countries. The first steps in improving early neonatal survival are to document the number and rate of deaths during the first week, identify the common causes and look for modifiable factors. Only then can a logical approach be made to plan intervention strategies.
The early neonatal death rate in South Africa is not known but a survey in health facilities in the Cape Province about live-born infants weighing 1000 g or more showed a wide range of first week deaths, between 4.4 and 17.0/1000 in different regions with higher rates in rural areas [5]. The mortality rate during the first 7 days of life increased by an additional 62% if infants weighing between 500 and 999 g at birth were included [6]. Any attempt to record early neonatal deaths in South Africa would, therefore, have to take into account regional differences and birth weight categories. In order to address the challenge of improving pregnancy outcome, a facility-based audit of perinatal deaths in South Africa was started.
The Perinatal Problem Identification Programme (PPIP) was developed in the 1990s by the Research Unit for Maternal and Infant Health Care Strategies of the South African Medical Research Council (MRC) and has been extensively field tested since 1996 [7]. The aim of the Programme was to identify the common causes of death and associated factors which could be addressed to reduce the perinatal mortality rate. Basic perinatal birth data and causes of death are recorded. Both possible and probable modifiable factors are also noted. Probable factors could be directly linked to the death. Data from various sites can be analysed separately or together. Thus perinatal care indices (perinatal mortality rate/low birth weight rate), patterns of disease and modifiable factors can be combined for various groupings of sites, e.g. regional, provincial or national; or primary, secondary and tertiary levels of care; or metropolitan, city and town, and rural areas. Although not time consuming or labour intensive, PPIP relies on the presence of regular mortality meetings to discuss perinatal deaths and the possible shortcomings in care. As this requires personal commitment to manage the system, it unfortunately cannot yet be introduced at all sites where births occur in South Africa. Therefore, these data may present a best scenario as only sites with enthusiastic staff were available for inclusion.
The classification method used in PPIP to describe the causes of perinatal death was first used in Aberdeen by Sir Dugald Baird and colleagues in the 1940s [8]. This system clearly points to where prevention can be targeted. It was later modified and modernised by Whitfield et al in 1986 [9] and adapted by Pattinson et al in 1989 [10] for use in developing countries, and again in 1995 to include the concept of modifiable factors [11]. The definition of modifiable factors used in PPIP was adapted from the confidential enquiries into maternal deaths in the United Kingdom in 1985 [12]. The primary objective of PPIP is not to conduct an inclusive nationwide survey but to provide a tool to identify ways to improve the quality of perinatal care. With large numbers of deliveries, the major causes of perinatal deaths and their order of magnitude can be determined.
The aim of this paper is to examine the PPIP data-base in order to propose ways of reducing the mortality rate during the first week of life of live born infants weighing 1000 g or more at delivery. In developing countries with limited resources, these are the early neonatal deaths most likely to be avoidable.
Methods
Data were collated from 102 sentinel sites within the public health service in South Africa. All perinatal deaths (stillbirths and neonatal deaths of 500 g or more) at these health care facilities were recorded over four years from 1st October 1999 to 30th September 2003. At each site all perinatal deaths were discussed at regular mortality audit meetings. While most first week deaths in infants weighing less than 2000 g at birth were captured, some deaths that occurred between the time of discharge from the health facility and 7 days of age in heavier infants may have been missed. Therefore, there may be an underestimation of deaths due to infections in infants weighing 2000 g or more. After review by the medical and nursing staff involved in the maternal and neonatal care, the probable primary and final causes of death were identified as well as any modifiable factors. The primary cause was defined as the underlying obstetric factor or condition which started a train of events that resulted in the death (why the death occurred, e.g. placental abruption) while the final cause was defined as the pathological process in the infant that actually caused the death (how the infant died, e.g. hypoxia). Primary causes identify pregnancy complications that can often be prevented (e.g. eclampsia) while final causes highlight areas of inadequate neonatal care (e.g. immaturity related deaths). The listed modifiable (avoidable) factors included missed opportunities for good care and examples of substandard care. These draw attention to areas of maternal and newborn care where improvements are needed.
The MRC unit contacted all services using PPIP and requested them to electronically send their data for collation, using PPIPWIN v2 (Simply Software®). This software package utilises a simple, user-friendly computer-based programme. Once basic perinatal data are entered, the programme calculates various perinatal care indices, describes the medical conditions that led to the perinatal deaths and lists modifiable factors associated with the deaths. Each site was categorised as metropolitan (the large amalgamated cities such as Cape Town), cities and towns, or rural areas. This categorisation was chosen as it grouped the hospitals and clinics into naturally comparable units, covered most of the institutional deliveries occurring in those areas and was thought to be more representative of population based data than any other combination. Most metropolitan areas are served by teaching hospitals and represent a fully functioning, tiered health care system, with all patients in the area having relatively easy access to tertiary care if needed. The city and town grouping represents areas where patients usually have easy access to primary and secondary level institutions, but where there is some difficulty in accessing tertiary hospitals. Finally the rural grouping represents primary care facilities, with the patients having to be referred for either secondary or tertiary care. Often health care facilities in cities and towns and rural areas had to provide levels of care beyond their means due to an inability to refer these patients. It was decided not to combine the data by levels of care across the country because of the very different referral patterns. All data were therefore categorised by site of delivery and not area of residence. The number of unrecorded deaths after home births is unknown, but is estimated not to be large. Only probable modifiable factors related to deaths are included in this analysis. Data on live born infants weighing 500 to 999 g were excluded as the reliability of these data were questionable. Some very small infants were coded as stillbirths in error, or their data were not recorded, as they were regarded as non-viable. Late neonatal deaths were also not considered as many infants are not closely followed by the health care services after the first week of life. Data were not available from the private sector where most of the community with health insurance receive care.
Infants with low-weight unrelated to maternal hypertension or any other identified obstetric cause were coded as "idiopathic SGA". Infection as a final cause of death included both prenatal infections (e.g. syphilis) and infections after delivery (e.g. necrotising enterocolitis).
Patient, administration and health worker related modifiable factors, which probably lead directly to the death, were considered. Sometimes more than one modifiable factor could be identified for a single death.
Ethics approval for the initial studies using PPIP was obtained from the University of Pretoria's Faculty of Health Sciences. The programme has since been taken over by the national and provincial Departments of Health and is approved by all health institutional Chief Executive Officers where it is used. Patient anonymity is assured at all times.
As this paper specifically addresses infant deaths in the first week of life, data on stillbirths have been excluded but are available on the PPIP website [7].
Results
During the four year period, data on 452 927 live born infants weighing 1000 g or more were recorded from the 102 study sites (Table 1). These sites were grouped into metropolitan (200 738 infants), cities and towns (146 999 infants) and rural (105 190 infants). The low birth weight rates (percentages) at these sites were 15.1%, 15.5% and 11.4% respectively. While the annual number of live births in the public sector in South Africa is not known, it is estimated at 800 000. Therefore, the study sample consisted of approximately 20% of all live births in the public health service. The hospitals and clinics where these births took place may present the best scenario as they were the health centres willing and enthusiastic to join the project. Data were not available for the approximately 20% of births which occurred in the private sector.
Table 1 Number of live births by birth weight and place of birth in South Africa over four years 1999–2003
Metropolitan Cities & Towns Rural All sites
1000–1499 g 4068 2520 765 7353
1500–1999 g 7928 5204 2449 15581
2000–2499 g 18353 15073 8743 42169
2500 g+ 170389 124202 93233 387824
Total 200738 146999 105190 452927
During the study period, 3916 live born infants, who weighed 1000 g or more and died in the first seven days, were identified. A further 17 were excluded as they were stillbirths incorrectly categorised as early neonatal deaths. The most common primary causes of death were "spontaneous preterm birth" (35.1%) and "intrapartum asphyxia and birth trauma" (33.4%) while the most common final causes of death were "hypoxia" (35.3%) and "immaturity related" (35.1%). Most of the deaths due to "hypoxia" (1187) were in infants weighing 2000 g or more. Mechanical trauma was uncommonly coded as a final cause of death, suggesting that the primary cause of "intrapartum asphyxia and trauma" reflected mainly foetal hypoxia in labour. There were more deaths without identified causes in rural than other areas.
The number, percentage and rates of primary and final causes of early neonatal deaths in the three regions are given in Table 2 and Table 3 respectively. The rates (per 1000 live births) for "intrapartum asphyxia and birth trauma" as a primary cause of death were highest in city and town (2.99) and rural (3.75) areas where death rates associated with "spontaneous preterm birth" were 4.33 and 4.0 respectively.
"Immaturity related" rates as a final cause of death were lowest in metropolitan areas (1.61) and higher in city and town (4.37) and rural (3.88) areas. "Hypoxia" rates were also higher in city and town (3.5) and rural (3.85) areas than metropolitan areas (2.3). Mortality rates due to infection would probably have been higher if late neonatal deaths were also considered.
Table 2 Primary causes of death in the three geographical groups (number, percentage and rate/1000 live births).
Primary Causes
Metropolitan Cities and Towns Rural
Number % Rate Number % Rate Number % Rate
Spontaneous preterm birth 324 25.47 1.61 636 40.64 4.33 421 38.41 4.00
Intrapartum asphyxia/birth trauma 362 28.46 1.80 439 28.05 2.99 394 35.95 3.75
Congenital abnormality 169 13.29 0.84 123 7.86 0.84 59 5.38 0.56
Hypertensive disorders 114 8.96 0.57 109 6.96 0.74 43 3.92 0.41
Infections 77 6.05 0.38 109 6.96 0.74 43 3.92 0.41
Abruptio placentae 89 7.00 0.44 53 3.39 0.36 23 2.10 0.22
Idiopathic SGA 46 3.62 0.23 25 1.60 0.17 14 1.28 0.13
Other antepartum haemorrhage 13 1.02 0.06 17 1.09 0.12 16 1.46 0.15
Pre-existing maternal disease 15 1.18 0.07 9 0.58 0.06 9 0.82 0.09
Other 58 4.95 0.31 35 2.88 0.31 72 6.75 0.70
Misclassified intra-uterine death 5 10 2
Total 1272 100.00 6.34 1565 100.00 10.65 1096 100.00 10.42
Table 3 Final causes of death in the three geographical groups (number, percentage and rate/1000 live births).
Final Causes
Metropolitan Cities and Towns Rural
Number % Rate Number % Rate Number % Rate
Hypoxia 462 36.46 2.30 515 33.12 3.50 405 37.02 3.85
Immaturity related 324 25.57 1.61 642 41.29 4.37 408 37.29 3.88
Infection 175 13.81 0.87 163 10.48 1.11 93 8.50 0.88
Congenital abnormalities 209 16.50 1.04 133 8.55 0.90 85 7.77 0.81
Trauma 13 1.03 0.06 11 0.71 0.07 11 1.01 0.10
Other 84 6.63 0.42 91 5.85 0.62 92 8.41 0.87
Total 1267 100.00 6.31 1555 100.00 10.58 1094 100.00 10.40
The top ten modifiable factors associated with death in the first week are given in Tables 4, 5 and 6 for metropolitan, city and town, and rural areas respectively. The most commonly identified modifiable factors could be grouped into poor care in labour, inadequate staffing and facilities, poor neonatal care, and patient difficulties in accessing care. In metropolitan areas, the most frequent problems were inadequate foetal monitoring in labour, insufficient staff, a delay in women seeking medical help in labour followed by poor management of the second stage of labour, inadequate neonatal care facilities and insufficient transport to move patients for tertiary care. In cities and towns the major problems were a delay in women seeking medical attention in labour, inadequate antenatal care, poor intrapartum foetal monitoring followed by inadequate facilities for neonatal care, a delay in medical personnel calling for assistance when needed, and a delay in referring patients for secondary or tertiary care from primary care clinics. In rural areas the most common problems were inadequate facilities for neonatal care, a lack of antenatal care, and poor intrapartum foetal monitoring followed by a delay in women seeking care in labour, and poor management in the second stage of labour.
Table 4 The top ten modifiable factors in deaths in metropolitan areas.
Description Number % of total deaths
Poor intrapartum foetal monitoring 49 3.8
Insufficient nurses/doctors on duty to manage the patient adequately 38 3.0
Patient delay in seeking medical attention during labour 28 2.2
Prolonged 2nd stage with no intervention 24 1.9
Inadequate facilities/equipment in neonatal unit/nursery 19 1.5
Lack of institution to institution transport 19 1.5
Inadequate monitoring of the newborn infant 14 1.1
Never initiated antenatal care 11 0.9
Delay in referring patient for secondary/tertiary treatment 10 0.8
No response to apparent post-term pregnancy 8 0.6
Table 5 The top ten modifiable factors in deaths in cities and towns.
Description Number % of total deaths
Patient delay in seeking medical attention during labour 84 5.3
No or infrequent antenatal care 60 3.8
Poor intrapartum foetal monitoring 58 3.7
Inadequate facilities/equipment in neonatal unit/nursery 30 1.9
Delay in medical personnel calling for expert assistance 28 1.8
Delay in referring patient for secondary/tertiary treatment 26 1.7
Prolonged 2nd stage with no intervention 16 1
Neonatal management plan inadequate 16 1
Poor progress in labour and partogram not used correctly 16 1
Medical personnel underestimated foetal size 13 1
Table 6 The top ten modifiable factors in deaths in rural areas.
Description Number % of total deaths
Inadequate facilities/equipment in neonatal unit/nursery 44 4.0
No or poor antenatal care 39 3.5
Poor intrapartum foetal monitoring 35 3.2
Patient delay in seeking medical attention during labour 27 2.4
Prolonged 2nd stage with no intervention 16 1.4
Inappropriate response to rupture of membranes 13 1.2
Lack of home to institution transport 13 1.2
No accessible neonatal ICU bed with ventilator 10 0.9
Poor progress in labour and partogram not used correctly 10 0.9
Delay in medical personnel calling for expert assistance 9 0.8
Neonatal management plan inadequate 9 0.8
Discussion
This is the first attempt to systematically document and identify the factors related to early neonatal deaths in South Africa. It is not an epidemiological study representing the whole country but rather an attempt to improve the care of infants in hospitals and clinics by identifying the main causes of death and common modifiable factors. It also only includes births in public hospitals and clinics willing to participate in the project. However, the numbers of births and deaths are substantial and the data were collected prospectively from both urban and rural sites after each death was discussed at a perinatal audit meeting. The findings are not surprising and probably apply to many other under resourced countries, especially in southern Africa. The higher percentage of unidentified caused of death in rural areas probably reflects a lack of clinical skills and knowledge.
The two most important causes of death were "intrapartum asphyxia and birth trauma" (intrapartum hypoxia) which resulted in neonatal "hypoxia", and "spontaneous preterm labour" leading to "immaturity related" births. The number of deaths coded as being due to the latter would have been even higher if infants weighing 500 to 999 g at birth were included. Death rates in preterm infants were particularly high in cities and towns and rural areas where neonatal high care facilities are very limited. The large number of deaths associated with perinatal hypoxia in all three groups suggested problems and inadequacies in care of women in labour and the resuscitation of newborn infants. Many of these deficiencies were identified as modifiable factors in all regions.
Almost half of the neonatal deaths due to "intrapartum asphyxia and trauma" were thought to be probably preventable. The commonest areas of suboptimal care were in foetal monitoring, monitoring the progress of labour and in managing the second stage of labour. Protocols for managing all aspects of labour are widely available and the appropriate use of partograms is strongly encouraged. The high recording rate of suboptimal care in labour is probably an indication that the knowledge on how to manage labour correctly is available, but not being adequately applied. It is disturbing that the mortality rate of term infants due to intrapartum hypoxia is so high. These deaths should be prevented and efforts to improve labour management must be intensified.
Relatively few modifiable factors were recorded, when related to the clinical management of the newborn infant. This is surprising as the large differences in neonatal death rates between metropolitan and other areas indicate that the assessment and management of newborn infants are not being performed adequately.
Specifically, only a few infants were recorded as having died as a result of poor neonatal resuscitation and care. Given the large number of neonatal deaths due to "hypoxia", it is inconceivable that this is a true reflection of the actual circumstances. There is probably poor insight into the deficiencies in the basic management of newborn infants as well as a lack of knowledge on neonatal resuscitation and care compared to intrapartum care. Time allocated to neonatal care during the training of both medical and nursing students is very limited in South Africa. If neonatal care is to be improved, more time and attention to its teaching is urgently required. A culture of questioning the neonatal management of infants who die in the first week of life must be included in perinatal mortality meetings. This will bring aspects of poor neonatal resuscitation to the fore and also act as a vehicle for training of all health workers involved in perinatal care. Currently a project to provide training in neonatal resuscitation to all health care workers is being launched by the South African Paediatric Association. Basic neonatal resuscitation is a simple skill that can be easily taught and applied. It should be made a requirement with the appropriate registration authority that all health care workers who conduct deliveries are able to provide basic neonatal resuscitation and immediate newborn care to prevent hypoxia, hypothermia and hypoglycaemia.
Improving the survival of low birth weight infants is a particular challenge in poor countries. Even in developed countries, efforts to lower the rate of preterm labour have not been successful. Therefore, as every site conducting deliveries will be faced at some time with a woman in advanced preterm labour, emphasis must be placed on excellent basic care of these small infants, often in a climate of inadequate facilities, equipment and well trained health care workers. Kangaroo mother care provides a cheap and effective answer to many of the problems posed by low birth weight infants. Recent projects in South Africa have shown that KMC can be implemented at all levels of care [13]. However, KMC must be incorporated into a package of good neonatal care and not be viewed as a stand alone intervention. Exclusive breastfeeding, clear management protocols and referral criteria, and respiratory support with nasal continuous positive airways pressure can be effectively introduced. These measurements can contribute to save infants weighing 1500 g to 2500 g.
The surprising finding of a lower rate of low birth weight infants in rural areas is unexplained but may reflect referral patterns of high risk mothers for delivery in towns and cities or a more secure food supply in rural areas than periurban slums.
A critical staff shortage was the second most common recorded modifiable factor directly involved in the death of a newborn infant in the metropolitan areas. Adequate numbers of well-trained nurses in both maternal and newborn care is probably the greatest need if the early neonatal mortality rate is to be substantially reduced in South Africa. As standards are not available to judge adequate staffing ratios, chronic understaffing is often accepted as the norm, especially outside metropolitan areas. Basic two-year courses for large numbers of primary care nurses and midwives are urgently required. Ongoing in-service support and training are also needed. However, with the current financial constraints, only a limited form of outreach training programme could realistically be made available to help staff in rural areas. Traditional methods of continuing education, where health workers are brought to regional centres for formal teaching, are no longer practical or affordable. Distance learning courses, aimed at enabling groups of practicing midwives to manage their own continuing education, have provided an alternative approach and had a major impact on the quality of maternal and perinatal care in South Africa. The Perinatal Education Programme (PEP) has been used as a self-help method of continuing training by more than 30 000 nurses over the past ten years in South Africa [14]. A number of studies have shown that PEP can significantly improve the knowledge [15], clinical skills [16], attitudes [17] and patient care practices [18] of midwives. Courses in maternal care, newborn care, primary newborn care, perinatal HIV/AIDS, mother and baby friendly care, and birth defects are available. An additional PEP course specifically addresses the knowledge and skills needed to manage perinatal mortality audits and use these data to improve perinatal care. These courses are available on an open website . However, facilitators to promote and support distance training programmes within health care regions throughout the country are still needed. Other sources of implementing best practice, such as the WHO Reproductive Health Library [19], could also help to transfer evidence into practice.
A new project in South Africa is tackling the development of appropriate, robust and cheap wind-up equipment, such as foetal heart rate and oxygen saturation monitors, to facilitate the perinatal care of infants in under resourced areas which often do not have a reliable energy supply and technological back up. The provision of adequate health infrastructures within an integrated, regional health system, better communication, and appropriate transport to improve care in rural as well as town and city areas needs to be accelerated. Specific problems such as the early identification and management of maternal syphilis are being tackled with the use of on-site screening. With the rapidly expanding AIDS pandemic, the cry for appropriate treatment for these mothers and their infants and a well planned and strengthened perinatal service is essential to support programmes that introduce the use of antiretroviral therapy and prophylaxis. With poverty alleviation and education of the general public to the importance of good, early antenatal care, many of the patient related modifiable factors could be reduced. These steps can best be achieved within the broader framework of social uplifting and job creation by the government with greater community involvement in public projects to provide a better life for all.
We now know what to do to lower the first week mortality rates of infants weighing 1000 g or more in South Africa. We also know how this can be done. What is currently lacking is not the required funding but the vision and political will to translate knowledge into practice. These challenges facing perinatal care in South Africa are mirrored in many other developing countries, especially in sub-Saharan Africa. Problems identified and lessons learned should be shared with others as we strive to improve the care of women and children and achieve the Millenium Development Goal of significantly reducing the number of childhood deaths under the age of five years in the next decade.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
Article written by R Pattinson and D Woods reviewed by D Greenfield and S Velaphi. All authors played an active role in collecting and analysing the data.
Acknowledgements
We would like to thank all the users of PPIP who submitted their data, Dr JohanCoetzee for creating the PPIPWIN programme, the MRC of South Africa, and the Saving Newborn Lives initiative of Save the Children (USA).
==== Refs
Saving Newborn Lives The state of the world's newborns: a report from Saving Newborn Lives 2001 Washington DC: Save the Children 14
United Nations, General Assembly, 56th session Road map towards the implementation of the United Nations Millenium Declaration: report of the Secretary-General 2001 New York: United Nations
Makubalo L (ed) Infant and Child Mortality South African Demographic and Health Survey 1998: Full Report 2000 Government Printer Pretoria 100 107
Pattinson RC (ed) Why Babies Die Saving Babies: Perinatal Care Survey of South Africa 2000 2001 Government Printer Pretoria 13 30
Louw HH Khan MBM Woods DL Power M Thompson MC Perinatal mortality in the Cape Province: 1989–1991 S Afr Med J 1995 85 352 355 7638682
Woods D Khan M Louw H A comparison of the perinatal mortality rates for infants at or above 500 g and 1000 g S Afr Med J 2001 91 323 324 11402904
The Perinatal Problem Identification Programme (PPIP) 2001
Baird D Thompson AM Butler NR, Alberman ED The survey perinatal deaths re-classified by special clinico-pathological assessment Perinatal Problems: the Second Report of the 1958 British Perinatal Mortality Survey 1969 Churchill Livingstone Edinburgh 200 210
Whitfield CR Smith NC Cockburn F Gibson AAM Perinatally related wastage – a proposed classification of primary obstetric factors Br J Obstet Gynaecol 1986 93 694 703 3730339
Pattinson RC De Jonge G Theron GB Primary causes of total perinatally related wastage at Tygerberg Hospital S Afr Med J 1989 75 50 53 2643836
Pattinson RC Makin JD Shaw A Delport SD The value of incorporating modifiable factors into perinatal deaths S Afr Med J 1995 85 145 147 7777959
Report on Confidential Enquiries Into Maternal Deaths in the United Kingdom 1985–87. Government Printer 1990 xiv 2343629
Bergh AM Arsalo I Phillips N Patrick M Malan A Pattinson RC Pattinson Implementation of kangaroo mother care: a successful case study in KwaZulu-Natal Saving babies 2002 2003 Third perinatal care survey of South Africa. Government Printers, Preoria 31 37
Woods DL An innovative programme for training in maternal and newborn care Semin Neonatal 1999 4 209 216 10.1016/S1084-2756(99)90090-8
Woods DL Theron GB The impact of the Perinatal Education Programme on cognitive knowledge of midwives S Afr Med J 1995 85 150 153 7777961
Theron GB Improving practical skills of midwives practicing in the Eastern Cape Province of the Republic of South Africa through the study of a self-educational manual J Perinatol 2000 3 184 188 10802845 10.1038/sj.jp.7200334
Theron GB The effect of the maternal care manual of the Perinatal Education Programme on the attitudes of midwives towards their work Curationis 2000 22 63 68 11051935
Theron GB Effect of the maternal care manual of the Perinatal Education Programme on the ability of midwives to interpret antenatal cards and partograms J Perinatol 1999 19 432 435 10685273 10.1038/sj.jp.7200225
The WHO Reproductive Health Library 2005 8 Update Software, Oxford
|
16095525
|
PMC1198255
|
CC BY
|
2021-01-04 16:38:16
|
no
|
Reprod Health. 2005 Aug 11; 2:4
|
utf-8
|
Reprod Health
| 2,005 |
10.1186/1742-4755-2-4
|
oa_comm
|
==== Front
Reprod HealthReproductive Health1742-4755BioMed Central London 1742-4755-2-41609552510.1186/1742-4755-2-4ResearchImproving survival rates of newborn infants in South Africa Pattinson Robert [email protected] David [email protected] David [email protected] Sithembiso [email protected] MRC Maternal and Infant Health Care Strategies Research Unit, Department of Obstetrics and Gynaecology, University of Pretoria, Pretoria, South Africa2 School of Child and Adolescent Health, Groote Schuur Hospital, University of Cape Town, Cape Town, South Africa3 Department of Paediatrics, Chris Hani Baragwanath Hospital, University of the Witwatersrand, Johannesburg, South Africa2005 11 8 2005 2 4 4 1 2 2005 11 8 2005 Copyright © 2005 Pattinson et al; licensee BioMed Central Ltd.2005Pattinson et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The number, rates and causes of early neonatal deaths in South Africa were not known. Neither had modifiable factors associated with these deaths been previously documented. An audit of live born infants who died in the first week of life in the public service could help in planning strategies to reduce the early neonatal mortality rate.
Methods
The number of live born infants weighing 1000 g or more, the number of these infants who die in the first week of life, the primary and final causes of these deaths, and the modifiable factors associated with them were collected over four years from 102 sites in South Africa as part of the Perinatal Problem Identification Programme.
Results
The rate of death in the first week of life for infants weighing 1000 g or more was unacceptably high (8.7/1000), especially in rural areas (10.42/1000). Intrapartum hypoxia and preterm delivery are the main causes of death. Common modifiable factors included inadequate staffing and facilities, poor care in labour, poor neonatal resuscitation and basic care, and difficulties for patients in accessing health care.
Conclusion
Practical, affordable and effective steps can be taken to reduce the number of infants who die in the first week of life in South Africa. These could also be implemented in other under resourced countries.
==== Body
Introduction
Of the approximately 130 million infants born worldwide each year, it is estimated that four million infants die during the first month of life [1]. The vast majority of these neonatal deaths occur in poor countries where standards of both maternal and newborn care are low. One of the Millenium Development Goals is to reduce the number of childhood deaths under the age of five years by two thirds from 95 per 1000 to 31 per 1000 by 2015 [2]. In South Africa, approximately 33% of deaths of under five-year-olds, 44% of infant deaths (before one year), and 87% of neonatal deaths (in the first month) occur during the first seven days after birth [3,4]. If the Millenium Development Goal of significantly reducing childhood deaths is to be achieved, a substantial reduction in early neonatal deaths will be required, especially in poor countries. The first steps in improving early neonatal survival are to document the number and rate of deaths during the first week, identify the common causes and look for modifiable factors. Only then can a logical approach be made to plan intervention strategies.
The early neonatal death rate in South Africa is not known but a survey in health facilities in the Cape Province about live-born infants weighing 1000 g or more showed a wide range of first week deaths, between 4.4 and 17.0/1000 in different regions with higher rates in rural areas [5]. The mortality rate during the first 7 days of life increased by an additional 62% if infants weighing between 500 and 999 g at birth were included [6]. Any attempt to record early neonatal deaths in South Africa would, therefore, have to take into account regional differences and birth weight categories. In order to address the challenge of improving pregnancy outcome, a facility-based audit of perinatal deaths in South Africa was started.
The Perinatal Problem Identification Programme (PPIP) was developed in the 1990s by the Research Unit for Maternal and Infant Health Care Strategies of the South African Medical Research Council (MRC) and has been extensively field tested since 1996 [7]. The aim of the Programme was to identify the common causes of death and associated factors which could be addressed to reduce the perinatal mortality rate. Basic perinatal birth data and causes of death are recorded. Both possible and probable modifiable factors are also noted. Probable factors could be directly linked to the death. Data from various sites can be analysed separately or together. Thus perinatal care indices (perinatal mortality rate/low birth weight rate), patterns of disease and modifiable factors can be combined for various groupings of sites, e.g. regional, provincial or national; or primary, secondary and tertiary levels of care; or metropolitan, city and town, and rural areas. Although not time consuming or labour intensive, PPIP relies on the presence of regular mortality meetings to discuss perinatal deaths and the possible shortcomings in care. As this requires personal commitment to manage the system, it unfortunately cannot yet be introduced at all sites where births occur in South Africa. Therefore, these data may present a best scenario as only sites with enthusiastic staff were available for inclusion.
The classification method used in PPIP to describe the causes of perinatal death was first used in Aberdeen by Sir Dugald Baird and colleagues in the 1940s [8]. This system clearly points to where prevention can be targeted. It was later modified and modernised by Whitfield et al in 1986 [9] and adapted by Pattinson et al in 1989 [10] for use in developing countries, and again in 1995 to include the concept of modifiable factors [11]. The definition of modifiable factors used in PPIP was adapted from the confidential enquiries into maternal deaths in the United Kingdom in 1985 [12]. The primary objective of PPIP is not to conduct an inclusive nationwide survey but to provide a tool to identify ways to improve the quality of perinatal care. With large numbers of deliveries, the major causes of perinatal deaths and their order of magnitude can be determined.
The aim of this paper is to examine the PPIP data-base in order to propose ways of reducing the mortality rate during the first week of life of live born infants weighing 1000 g or more at delivery. In developing countries with limited resources, these are the early neonatal deaths most likely to be avoidable.
Methods
Data were collated from 102 sentinel sites within the public health service in South Africa. All perinatal deaths (stillbirths and neonatal deaths of 500 g or more) at these health care facilities were recorded over four years from 1st October 1999 to 30th September 2003. At each site all perinatal deaths were discussed at regular mortality audit meetings. While most first week deaths in infants weighing less than 2000 g at birth were captured, some deaths that occurred between the time of discharge from the health facility and 7 days of age in heavier infants may have been missed. Therefore, there may be an underestimation of deaths due to infections in infants weighing 2000 g or more. After review by the medical and nursing staff involved in the maternal and neonatal care, the probable primary and final causes of death were identified as well as any modifiable factors. The primary cause was defined as the underlying obstetric factor or condition which started a train of events that resulted in the death (why the death occurred, e.g. placental abruption) while the final cause was defined as the pathological process in the infant that actually caused the death (how the infant died, e.g. hypoxia). Primary causes identify pregnancy complications that can often be prevented (e.g. eclampsia) while final causes highlight areas of inadequate neonatal care (e.g. immaturity related deaths). The listed modifiable (avoidable) factors included missed opportunities for good care and examples of substandard care. These draw attention to areas of maternal and newborn care where improvements are needed.
The MRC unit contacted all services using PPIP and requested them to electronically send their data for collation, using PPIPWIN v2 (Simply Software®). This software package utilises a simple, user-friendly computer-based programme. Once basic perinatal data are entered, the programme calculates various perinatal care indices, describes the medical conditions that led to the perinatal deaths and lists modifiable factors associated with the deaths. Each site was categorised as metropolitan (the large amalgamated cities such as Cape Town), cities and towns, or rural areas. This categorisation was chosen as it grouped the hospitals and clinics into naturally comparable units, covered most of the institutional deliveries occurring in those areas and was thought to be more representative of population based data than any other combination. Most metropolitan areas are served by teaching hospitals and represent a fully functioning, tiered health care system, with all patients in the area having relatively easy access to tertiary care if needed. The city and town grouping represents areas where patients usually have easy access to primary and secondary level institutions, but where there is some difficulty in accessing tertiary hospitals. Finally the rural grouping represents primary care facilities, with the patients having to be referred for either secondary or tertiary care. Often health care facilities in cities and towns and rural areas had to provide levels of care beyond their means due to an inability to refer these patients. It was decided not to combine the data by levels of care across the country because of the very different referral patterns. All data were therefore categorised by site of delivery and not area of residence. The number of unrecorded deaths after home births is unknown, but is estimated not to be large. Only probable modifiable factors related to deaths are included in this analysis. Data on live born infants weighing 500 to 999 g were excluded as the reliability of these data were questionable. Some very small infants were coded as stillbirths in error, or their data were not recorded, as they were regarded as non-viable. Late neonatal deaths were also not considered as many infants are not closely followed by the health care services after the first week of life. Data were not available from the private sector where most of the community with health insurance receive care.
Infants with low-weight unrelated to maternal hypertension or any other identified obstetric cause were coded as "idiopathic SGA". Infection as a final cause of death included both prenatal infections (e.g. syphilis) and infections after delivery (e.g. necrotising enterocolitis).
Patient, administration and health worker related modifiable factors, which probably lead directly to the death, were considered. Sometimes more than one modifiable factor could be identified for a single death.
Ethics approval for the initial studies using PPIP was obtained from the University of Pretoria's Faculty of Health Sciences. The programme has since been taken over by the national and provincial Departments of Health and is approved by all health institutional Chief Executive Officers where it is used. Patient anonymity is assured at all times.
As this paper specifically addresses infant deaths in the first week of life, data on stillbirths have been excluded but are available on the PPIP website [7].
Results
During the four year period, data on 452 927 live born infants weighing 1000 g or more were recorded from the 102 study sites (Table 1). These sites were grouped into metropolitan (200 738 infants), cities and towns (146 999 infants) and rural (105 190 infants). The low birth weight rates (percentages) at these sites were 15.1%, 15.5% and 11.4% respectively. While the annual number of live births in the public sector in South Africa is not known, it is estimated at 800 000. Therefore, the study sample consisted of approximately 20% of all live births in the public health service. The hospitals and clinics where these births took place may present the best scenario as they were the health centres willing and enthusiastic to join the project. Data were not available for the approximately 20% of births which occurred in the private sector.
Table 1 Number of live births by birth weight and place of birth in South Africa over four years 1999–2003
Metropolitan Cities & Towns Rural All sites
1000–1499 g 4068 2520 765 7353
1500–1999 g 7928 5204 2449 15581
2000–2499 g 18353 15073 8743 42169
2500 g+ 170389 124202 93233 387824
Total 200738 146999 105190 452927
During the study period, 3916 live born infants, who weighed 1000 g or more and died in the first seven days, were identified. A further 17 were excluded as they were stillbirths incorrectly categorised as early neonatal deaths. The most common primary causes of death were "spontaneous preterm birth" (35.1%) and "intrapartum asphyxia and birth trauma" (33.4%) while the most common final causes of death were "hypoxia" (35.3%) and "immaturity related" (35.1%). Most of the deaths due to "hypoxia" (1187) were in infants weighing 2000 g or more. Mechanical trauma was uncommonly coded as a final cause of death, suggesting that the primary cause of "intrapartum asphyxia and trauma" reflected mainly foetal hypoxia in labour. There were more deaths without identified causes in rural than other areas.
The number, percentage and rates of primary and final causes of early neonatal deaths in the three regions are given in Table 2 and Table 3 respectively. The rates (per 1000 live births) for "intrapartum asphyxia and birth trauma" as a primary cause of death were highest in city and town (2.99) and rural (3.75) areas where death rates associated with "spontaneous preterm birth" were 4.33 and 4.0 respectively.
"Immaturity related" rates as a final cause of death were lowest in metropolitan areas (1.61) and higher in city and town (4.37) and rural (3.88) areas. "Hypoxia" rates were also higher in city and town (3.5) and rural (3.85) areas than metropolitan areas (2.3). Mortality rates due to infection would probably have been higher if late neonatal deaths were also considered.
Table 2 Primary causes of death in the three geographical groups (number, percentage and rate/1000 live births).
Primary Causes
Metropolitan Cities and Towns Rural
Number % Rate Number % Rate Number % Rate
Spontaneous preterm birth 324 25.47 1.61 636 40.64 4.33 421 38.41 4.00
Intrapartum asphyxia/birth trauma 362 28.46 1.80 439 28.05 2.99 394 35.95 3.75
Congenital abnormality 169 13.29 0.84 123 7.86 0.84 59 5.38 0.56
Hypertensive disorders 114 8.96 0.57 109 6.96 0.74 43 3.92 0.41
Infections 77 6.05 0.38 109 6.96 0.74 43 3.92 0.41
Abruptio placentae 89 7.00 0.44 53 3.39 0.36 23 2.10 0.22
Idiopathic SGA 46 3.62 0.23 25 1.60 0.17 14 1.28 0.13
Other antepartum haemorrhage 13 1.02 0.06 17 1.09 0.12 16 1.46 0.15
Pre-existing maternal disease 15 1.18 0.07 9 0.58 0.06 9 0.82 0.09
Other 58 4.95 0.31 35 2.88 0.31 72 6.75 0.70
Misclassified intra-uterine death 5 10 2
Total 1272 100.00 6.34 1565 100.00 10.65 1096 100.00 10.42
Table 3 Final causes of death in the three geographical groups (number, percentage and rate/1000 live births).
Final Causes
Metropolitan Cities and Towns Rural
Number % Rate Number % Rate Number % Rate
Hypoxia 462 36.46 2.30 515 33.12 3.50 405 37.02 3.85
Immaturity related 324 25.57 1.61 642 41.29 4.37 408 37.29 3.88
Infection 175 13.81 0.87 163 10.48 1.11 93 8.50 0.88
Congenital abnormalities 209 16.50 1.04 133 8.55 0.90 85 7.77 0.81
Trauma 13 1.03 0.06 11 0.71 0.07 11 1.01 0.10
Other 84 6.63 0.42 91 5.85 0.62 92 8.41 0.87
Total 1267 100.00 6.31 1555 100.00 10.58 1094 100.00 10.40
The top ten modifiable factors associated with death in the first week are given in Tables 4, 5 and 6 for metropolitan, city and town, and rural areas respectively. The most commonly identified modifiable factors could be grouped into poor care in labour, inadequate staffing and facilities, poor neonatal care, and patient difficulties in accessing care. In metropolitan areas, the most frequent problems were inadequate foetal monitoring in labour, insufficient staff, a delay in women seeking medical help in labour followed by poor management of the second stage of labour, inadequate neonatal care facilities and insufficient transport to move patients for tertiary care. In cities and towns the major problems were a delay in women seeking medical attention in labour, inadequate antenatal care, poor intrapartum foetal monitoring followed by inadequate facilities for neonatal care, a delay in medical personnel calling for assistance when needed, and a delay in referring patients for secondary or tertiary care from primary care clinics. In rural areas the most common problems were inadequate facilities for neonatal care, a lack of antenatal care, and poor intrapartum foetal monitoring followed by a delay in women seeking care in labour, and poor management in the second stage of labour.
Table 4 The top ten modifiable factors in deaths in metropolitan areas.
Description Number % of total deaths
Poor intrapartum foetal monitoring 49 3.8
Insufficient nurses/doctors on duty to manage the patient adequately 38 3.0
Patient delay in seeking medical attention during labour 28 2.2
Prolonged 2nd stage with no intervention 24 1.9
Inadequate facilities/equipment in neonatal unit/nursery 19 1.5
Lack of institution to institution transport 19 1.5
Inadequate monitoring of the newborn infant 14 1.1
Never initiated antenatal care 11 0.9
Delay in referring patient for secondary/tertiary treatment 10 0.8
No response to apparent post-term pregnancy 8 0.6
Table 5 The top ten modifiable factors in deaths in cities and towns.
Description Number % of total deaths
Patient delay in seeking medical attention during labour 84 5.3
No or infrequent antenatal care 60 3.8
Poor intrapartum foetal monitoring 58 3.7
Inadequate facilities/equipment in neonatal unit/nursery 30 1.9
Delay in medical personnel calling for expert assistance 28 1.8
Delay in referring patient for secondary/tertiary treatment 26 1.7
Prolonged 2nd stage with no intervention 16 1
Neonatal management plan inadequate 16 1
Poor progress in labour and partogram not used correctly 16 1
Medical personnel underestimated foetal size 13 1
Table 6 The top ten modifiable factors in deaths in rural areas.
Description Number % of total deaths
Inadequate facilities/equipment in neonatal unit/nursery 44 4.0
No or poor antenatal care 39 3.5
Poor intrapartum foetal monitoring 35 3.2
Patient delay in seeking medical attention during labour 27 2.4
Prolonged 2nd stage with no intervention 16 1.4
Inappropriate response to rupture of membranes 13 1.2
Lack of home to institution transport 13 1.2
No accessible neonatal ICU bed with ventilator 10 0.9
Poor progress in labour and partogram not used correctly 10 0.9
Delay in medical personnel calling for expert assistance 9 0.8
Neonatal management plan inadequate 9 0.8
Discussion
This is the first attempt to systematically document and identify the factors related to early neonatal deaths in South Africa. It is not an epidemiological study representing the whole country but rather an attempt to improve the care of infants in hospitals and clinics by identifying the main causes of death and common modifiable factors. It also only includes births in public hospitals and clinics willing to participate in the project. However, the numbers of births and deaths are substantial and the data were collected prospectively from both urban and rural sites after each death was discussed at a perinatal audit meeting. The findings are not surprising and probably apply to many other under resourced countries, especially in southern Africa. The higher percentage of unidentified caused of death in rural areas probably reflects a lack of clinical skills and knowledge.
The two most important causes of death were "intrapartum asphyxia and birth trauma" (intrapartum hypoxia) which resulted in neonatal "hypoxia", and "spontaneous preterm labour" leading to "immaturity related" births. The number of deaths coded as being due to the latter would have been even higher if infants weighing 500 to 999 g at birth were included. Death rates in preterm infants were particularly high in cities and towns and rural areas where neonatal high care facilities are very limited. The large number of deaths associated with perinatal hypoxia in all three groups suggested problems and inadequacies in care of women in labour and the resuscitation of newborn infants. Many of these deficiencies were identified as modifiable factors in all regions.
Almost half of the neonatal deaths due to "intrapartum asphyxia and trauma" were thought to be probably preventable. The commonest areas of suboptimal care were in foetal monitoring, monitoring the progress of labour and in managing the second stage of labour. Protocols for managing all aspects of labour are widely available and the appropriate use of partograms is strongly encouraged. The high recording rate of suboptimal care in labour is probably an indication that the knowledge on how to manage labour correctly is available, but not being adequately applied. It is disturbing that the mortality rate of term infants due to intrapartum hypoxia is so high. These deaths should be prevented and efforts to improve labour management must be intensified.
Relatively few modifiable factors were recorded, when related to the clinical management of the newborn infant. This is surprising as the large differences in neonatal death rates between metropolitan and other areas indicate that the assessment and management of newborn infants are not being performed adequately.
Specifically, only a few infants were recorded as having died as a result of poor neonatal resuscitation and care. Given the large number of neonatal deaths due to "hypoxia", it is inconceivable that this is a true reflection of the actual circumstances. There is probably poor insight into the deficiencies in the basic management of newborn infants as well as a lack of knowledge on neonatal resuscitation and care compared to intrapartum care. Time allocated to neonatal care during the training of both medical and nursing students is very limited in South Africa. If neonatal care is to be improved, more time and attention to its teaching is urgently required. A culture of questioning the neonatal management of infants who die in the first week of life must be included in perinatal mortality meetings. This will bring aspects of poor neonatal resuscitation to the fore and also act as a vehicle for training of all health workers involved in perinatal care. Currently a project to provide training in neonatal resuscitation to all health care workers is being launched by the South African Paediatric Association. Basic neonatal resuscitation is a simple skill that can be easily taught and applied. It should be made a requirement with the appropriate registration authority that all health care workers who conduct deliveries are able to provide basic neonatal resuscitation and immediate newborn care to prevent hypoxia, hypothermia and hypoglycaemia.
Improving the survival of low birth weight infants is a particular challenge in poor countries. Even in developed countries, efforts to lower the rate of preterm labour have not been successful. Therefore, as every site conducting deliveries will be faced at some time with a woman in advanced preterm labour, emphasis must be placed on excellent basic care of these small infants, often in a climate of inadequate facilities, equipment and well trained health care workers. Kangaroo mother care provides a cheap and effective answer to many of the problems posed by low birth weight infants. Recent projects in South Africa have shown that KMC can be implemented at all levels of care [13]. However, KMC must be incorporated into a package of good neonatal care and not be viewed as a stand alone intervention. Exclusive breastfeeding, clear management protocols and referral criteria, and respiratory support with nasal continuous positive airways pressure can be effectively introduced. These measurements can contribute to save infants weighing 1500 g to 2500 g.
The surprising finding of a lower rate of low birth weight infants in rural areas is unexplained but may reflect referral patterns of high risk mothers for delivery in towns and cities or a more secure food supply in rural areas than periurban slums.
A critical staff shortage was the second most common recorded modifiable factor directly involved in the death of a newborn infant in the metropolitan areas. Adequate numbers of well-trained nurses in both maternal and newborn care is probably the greatest need if the early neonatal mortality rate is to be substantially reduced in South Africa. As standards are not available to judge adequate staffing ratios, chronic understaffing is often accepted as the norm, especially outside metropolitan areas. Basic two-year courses for large numbers of primary care nurses and midwives are urgently required. Ongoing in-service support and training are also needed. However, with the current financial constraints, only a limited form of outreach training programme could realistically be made available to help staff in rural areas. Traditional methods of continuing education, where health workers are brought to regional centres for formal teaching, are no longer practical or affordable. Distance learning courses, aimed at enabling groups of practicing midwives to manage their own continuing education, have provided an alternative approach and had a major impact on the quality of maternal and perinatal care in South Africa. The Perinatal Education Programme (PEP) has been used as a self-help method of continuing training by more than 30 000 nurses over the past ten years in South Africa [14]. A number of studies have shown that PEP can significantly improve the knowledge [15], clinical skills [16], attitudes [17] and patient care practices [18] of midwives. Courses in maternal care, newborn care, primary newborn care, perinatal HIV/AIDS, mother and baby friendly care, and birth defects are available. An additional PEP course specifically addresses the knowledge and skills needed to manage perinatal mortality audits and use these data to improve perinatal care. These courses are available on an open website . However, facilitators to promote and support distance training programmes within health care regions throughout the country are still needed. Other sources of implementing best practice, such as the WHO Reproductive Health Library [19], could also help to transfer evidence into practice.
A new project in South Africa is tackling the development of appropriate, robust and cheap wind-up equipment, such as foetal heart rate and oxygen saturation monitors, to facilitate the perinatal care of infants in under resourced areas which often do not have a reliable energy supply and technological back up. The provision of adequate health infrastructures within an integrated, regional health system, better communication, and appropriate transport to improve care in rural as well as town and city areas needs to be accelerated. Specific problems such as the early identification and management of maternal syphilis are being tackled with the use of on-site screening. With the rapidly expanding AIDS pandemic, the cry for appropriate treatment for these mothers and their infants and a well planned and strengthened perinatal service is essential to support programmes that introduce the use of antiretroviral therapy and prophylaxis. With poverty alleviation and education of the general public to the importance of good, early antenatal care, many of the patient related modifiable factors could be reduced. These steps can best be achieved within the broader framework of social uplifting and job creation by the government with greater community involvement in public projects to provide a better life for all.
We now know what to do to lower the first week mortality rates of infants weighing 1000 g or more in South Africa. We also know how this can be done. What is currently lacking is not the required funding but the vision and political will to translate knowledge into practice. These challenges facing perinatal care in South Africa are mirrored in many other developing countries, especially in sub-Saharan Africa. Problems identified and lessons learned should be shared with others as we strive to improve the care of women and children and achieve the Millenium Development Goal of significantly reducing the number of childhood deaths under the age of five years in the next decade.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
Article written by R Pattinson and D Woods reviewed by D Greenfield and S Velaphi. All authors played an active role in collecting and analysing the data.
Acknowledgements
We would like to thank all the users of PPIP who submitted their data, Dr JohanCoetzee for creating the PPIPWIN programme, the MRC of South Africa, and the Saving Newborn Lives initiative of Save the Children (USA).
==== Refs
Saving Newborn Lives The state of the world's newborns: a report from Saving Newborn Lives 2001 Washington DC: Save the Children 14
United Nations, General Assembly, 56th session Road map towards the implementation of the United Nations Millenium Declaration: report of the Secretary-General 2001 New York: United Nations
Makubalo L (ed) Infant and Child Mortality South African Demographic and Health Survey 1998: Full Report 2000 Government Printer Pretoria 100 107
Pattinson RC (ed) Why Babies Die Saving Babies: Perinatal Care Survey of South Africa 2000 2001 Government Printer Pretoria 13 30
Louw HH Khan MBM Woods DL Power M Thompson MC Perinatal mortality in the Cape Province: 1989–1991 S Afr Med J 1995 85 352 355 7638682
Woods D Khan M Louw H A comparison of the perinatal mortality rates for infants at or above 500 g and 1000 g S Afr Med J 2001 91 323 324 11402904
The Perinatal Problem Identification Programme (PPIP) 2001
Baird D Thompson AM Butler NR, Alberman ED The survey perinatal deaths re-classified by special clinico-pathological assessment Perinatal Problems: the Second Report of the 1958 British Perinatal Mortality Survey 1969 Churchill Livingstone Edinburgh 200 210
Whitfield CR Smith NC Cockburn F Gibson AAM Perinatally related wastage – a proposed classification of primary obstetric factors Br J Obstet Gynaecol 1986 93 694 703 3730339
Pattinson RC De Jonge G Theron GB Primary causes of total perinatally related wastage at Tygerberg Hospital S Afr Med J 1989 75 50 53 2643836
Pattinson RC Makin JD Shaw A Delport SD The value of incorporating modifiable factors into perinatal deaths S Afr Med J 1995 85 145 147 7777959
Report on Confidential Enquiries Into Maternal Deaths in the United Kingdom 1985–87. Government Printer 1990 xiv 2343629
Bergh AM Arsalo I Phillips N Patrick M Malan A Pattinson RC Pattinson Implementation of kangaroo mother care: a successful case study in KwaZulu-Natal Saving babies 2002 2003 Third perinatal care survey of South Africa. Government Printers, Preoria 31 37
Woods DL An innovative programme for training in maternal and newborn care Semin Neonatal 1999 4 209 216 10.1016/S1084-2756(99)90090-8
Woods DL Theron GB The impact of the Perinatal Education Programme on cognitive knowledge of midwives S Afr Med J 1995 85 150 153 7777961
Theron GB Improving practical skills of midwives practicing in the Eastern Cape Province of the Republic of South Africa through the study of a self-educational manual J Perinatol 2000 3 184 188 10802845 10.1038/sj.jp.7200334
Theron GB The effect of the maternal care manual of the Perinatal Education Programme on the attitudes of midwives towards their work Curationis 2000 22 63 68 11051935
Theron GB Effect of the maternal care manual of the Perinatal Education Programme on the ability of midwives to interpret antenatal cards and partograms J Perinatol 1999 19 432 435 10685273 10.1038/sj.jp.7200225
The WHO Reproductive Health Library 2005 8 Update Software, Oxford
|
16105174
|
PMC1198256
|
CC BY
|
2021-01-04 16:38:16
|
no
|
Reprod Health. 2005 Aug 16; 2:6
|
latin-1
|
Reprod Health
| 2,005 |
10.1186/1742-4755-2-6
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-671600197910.1186/1465-9921-6-67ResearchIL-4 increases type 2, but not type 1, cytokine production in CD8+ T cells from mild atopic asthmatics Stanciu Luminita A [email protected] Kevan [email protected] Nikolaos G [email protected] Sang-Heon [email protected] Stephen T [email protected] Anthony J [email protected] Sebastian L [email protected] Department of Respiratory Medicine, National Heart and Lung Institute & Wright Fleming Institute for Infection and Immunity, Imperial College London, Norfolk Place, London, UK2 Respiratory Cell and Molecular Biology Research Division, University of Southampton, Southampton, UK3 MedImmune, Gaithersburg, USA2005 7 7 2005 6 1 67 67 7 2 2005 7 7 2005 Copyright © 2005 Stanciu et al; licensee BioMed Central Ltd.2005Stanciu et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Virus infections are the major cause of asthma exacerbations. CD8+ T cells have an important role in antiviral immune responses and animal studies suggest a role for CD8+ T cells in the pathogenesis of virus-induced asthma exacerbations. We have previously shown that the presence of IL-4 during stimulation increases the frequency of IL-5-positive cells and CD30 surface staining in CD8+ T cells from healthy, normal subjects. In this study, we investigated whether excess IL-4 during repeated TCR/CD3 stimulation of CD8+ T cells from atopic asthmatic subjects alters the balance of type 1/type 2 cytokine production in favour of the latter.
Methods
Peripheral blood CD8+ T cells from mild atopic asthmatic subjects were stimulated in vitro with anti-CD3 and IL-2 ± excess IL-4 and the expression of activation and adhesion molecules and type 1 and type 2 cytokine production were assessed.
Results
Surface expression of very late antigen-4 [VLA-4] and LFA-1 was decreased and the production of the type 2 cytokines IL-5 and IL-13 was augmented by the presence of IL-4 during stimulation of CD8+ T cells from mild atopic asthmatics.
Conclusion
These data suggest that during a respiratory virus infection activated CD8+ T cells from asthmatic subjects may produce excess type 2 cytokines and may contribute to asthma exacerbation by augmenting allergic inflammation.
==== Body
Background
Asthmatic airways are characterised by high levels of IL-4 due to excess Th2 responses to common aeroallergens [1-3]. Respiratory viruses are associated with 80 to 85% of asthma exacerbations in children and with 60% in adults and rhinovirus (RV) is common in all age groups [4-6]. We have found that atopic asthmatic subjects suffer from more severe and longer lasting lower respiratory tract symptoms when infected with RV than do normal subjects [7], but the mechanisms are not known. We proposed that one mechanism could be that immune responses to virus infection are type 1 deficient/type 2 augmented if they take place in a milieu rich in the type 2 cytokine IL-4 as occurs in atopic asthma [8]. Optimal CD8+ T cell responses to respiratory viruses are predominantly type 1 and over exuberant type 2 responses are detrimental [9]. Animal studies suggest that immune responses to virus infections are characterized by an increase in the frequency of type 2 cytokine-producing T cells if they take place in an allergic environment [10,11]. Data from human studies also show that an excessive type 2/deficient type 1 cytokine response to RSV-infection in the respiratory tract is associated with a more severe clinical course [12].
We have published data showing that human CD8+ T cells from both normal and asthmatic subjects have the capacity to produce type 2 cytokines [13,14]. We have recently shown that stimulation of CD8+ T cells from normal, healthy subjects in an IL-4 rich milieu significantly increased the number of IL-5-positive CD8+ T cells [15]. Due to excess Th2 immune response to allergens, CD8+ T cells from atopic asthmatic subjects are continuous exposed to IL-4 [1]. Here we investigated whether exogenous IL-4 present during TCR/CD3 stimulation modulates cell growth, surface phenotype (activation markers and adhesion molecules) and type 1 (IL-2 and IFN-γ) and type 2 (IL-4, -5, -13) cytokine production in CD8+ T cells from atopic asthmatic subjects. We found that the presence of exogenous IL-4 during the activation of mild atopic asthmatic CD8+ T cells enhanced cell growth and increased IL-5 and IL-13 production.
Methods
Subjects
Peripheral blood (40 to 50 mL) was obtained from 10 patients fulfilling clinical diagnostic criteria for asthma [16] and showing hyperreactivity to inhaled methacholine (Table 1). Asthmatic patients were classified as having mild-to-moderate asthma according to the standards of American Thoracic Society [16] and were treated with selective beta2-agonists as required. All asthmatic patients were atopic with high levels of total serum IgE (>80 IU/mL) and with positive skin prick test responses (greater than 3 mm skin wheal response) to one or more of a series of common allergens: house dust mite, mixed grass pollens, mixed tree pollens, mixed feathers, cat fur and dog hair (ALK, Denmark). None of the subjects were smokers or had a history of respiratory tract infections within the previous two months. The study was approved by Southampton Joint University and Hospital Ethics Committee.
Table 1 Characteristics of subjects
Asthma subjects (n = 10)
Age (years)
Mean (Range) 33 (20–48)
Sex (male/female) 5/5
PC20 methacholine (mg/mL) *
Geometric mean (range) 5.33 (1.5 to 7.9)
Atopy# 10
Serum IgE (IU/mL)
Median (range) 100 (81 to 442)
*Provocative concentration of methacholine required to cause a 20% decrease in baseline forced expiratory volume in 1 second (FEV1).
# Positive skin reactivity to a panel of common allergen (see methods)
Antibodies and other reagents
"CD8+ T cell isolation" kits were from Miltenyi (Miltenyi Biotec GmbH, Germany). Mouse anti-human CD3 Abs was purified from the OKT-3 hybridoma (American Type Culture Collection, Rockville, USA). rIL-2 and rIL-4 were purchased from Genzyme (Genzyme, West Malling, Kent, UK). Leu 4 (CD3)-Peridinin Chlorophyll Protein conjugate, Leu 3a (CD4)-FITC and -phycoerythrin (PE), Leu 2a (CD8)-FITC and -PE, CD14-PE, CD16-PE, CD19-PE, CD25-FITC, CD11a (LFA-1)-FITC and CD49d (VLA-4)-PE were from Becton Dickinson (Becton Dickinson, Mountain View, CA). CD54 (ICAM-1)-FITC was from Serotec (Serotec Ltd, Oxford, UK). CD30-FITC and CD154 (CD40 ligand, CD40L)-PE were from PharMingen (PharMingen, San Diego, CA). ELISA kits were obtained from Biosource (Biosource Europe S.A., Fleurus, Belgium).
CD8+ T cell purification and culture
CD8+ T cells were enriched by negative immunomagnetic selection using the MACS system [17]. After passage through MACS columns, between 0.25 and 2.3 × 106 CD8+ T cells were recovered from 1 mL of blood. The resulting cell populations were routinely >98% CD3+ and >97% CD8+ and contained no detectable CD4+, CD16+, CD19+, CD14+ cells as determined by flow cytometry.
Purified CD8+ T cells were cultured in an antigen presenting cell-free system as previously described [15]. In brief, 0.5 × 106 cells/mL were stimulated for 3 days with plate-bound anti-CD3 Abs (10 μg/mL) in 96-well microtiter plates in RPMI 1640 culture medium supplemented with 10% FCS, 2 mM L-glutamine, 1 mM sodium pyruvate, antibiotics (all from Gibco BRL, Life Technologies, Uxbridge, UK), IL-2 (50 U/mL) with or without added IL-4 (10 ng/mL) (stimulation). After 3 days, the cells were harvested, washed, adjusted to 0.2 × 106/mL and cultured for 4 days in culture medium containing IL-2 (50 U/mL) (expansion). After every cycle of stimulation or expansion the cells were harvested and the number of live cells was determined by trypan blue exclusion and counting. After the second cycle of stimulation and expansion the cells were harvested, washed, adjusted to 1 × 106/mL and analysed.
Cell surface markers
Following the second cycle of stimulation/expansion, the cells were harvested and washed. PerCP-, FITC-, and PE-conjugated antibodies were added to 50 μl of suspensions (0.05 × 106 cells) and incubated for 30 min at 4°C. After washing in PBS, stained cells were resuspended in 500 μl PBS. Flow cytometry was performed on a FACScan (Becton Dickinson, San Jose, CA). The FACScan program was used to acquire data of 10,000 events and the data were analysed by using Lysys II software. Live lymphocytes were gated by the forward scatter and side scatter pattern. Percentage in dot plots and mean fluorescence intensity (MFI) in histograms were analysed on CD3+CD8+ cells for the molecules of interest.
Cytokine production
After two cycles of stimulation/expansion, CD8+ T cells (1 × 106/mL) were restimulated for 24 hours with phorbol myristate acetate (PMA) (20 ng/mL) and ionomycin (2 μg/mL) or with immobilised anti-CD3 Ab. Supernatants were microfuged for 5 minutes at 400 g to remove cell debris and IL-2, IL-4, IL-5, IL-13, and IFN-γ levels were measured by ELISA using commercial assay kits (Biosource, Europe SA). The limits of sensitivity of these assays were 3 pg/mL for IL-4, 4 pg/mL for IL-5 and IFN-γ, 5 pg/mL for IL-2, and 12 pg/mL for IL-13.
Statistical analysis
Paired Student t-test was used to analyse normally distributed data such as growth of CD8+ T cells cultured with or without IL-4. The Wilcoxon signed rank test for paired data was used for comparison of surface markers and for cytokine levels in culture supernatants. P values less than 0.05 were chosen for rejection of the null hypothesis.
Results
IL-4 amplifies the growth of CD8+ T cells from atopic asthmatic subjects
In order to analyse the effect of IL-4 on CD8+ T cell growth we examined cell recovery after the first and second cycles of stimulation and expansion. At the end of the first cycle of stimulation/expansion there was no difference in the number of CD8+ T cells cultured with or without IL-4 (data not shown), but after the second cycle the number of cells stimulated in the presence of IL-4 was significantly higher than number of cells stimulated in the absence of IL-4 (4.5 ± 0.7 fold increase relative to the initial population in the presence of exogenous IL-4 versus 2.9 ± 0.7 fold increase without exogenous IL-4 (Fig. 1, P < 0.02).
Figure 1 Growth kinetics (fold increase) of CD8+ T cell numbers. Data are means ± SEM (n = 10 subjects). * p < 0.02
IL-4 decreases surface expression of adhesion molecules in CD8+ T cells from atopic asthmatic subjects
As we were interested in airway responses of CD8+ T in the context of viral infection, we then investigated whether levels of expression of surface antigens associated with activation, adhesion and transmigration capabilities were altered by the presence of IL-4 during CD8+ T cell stimulation (Table 2). There was a trend to increased numbers of CD30-positive CD8+ T cells in culture with exogenous IL-4 but the difference did not reach significance (P < 0.07). However, the presence of exogenous IL-4 during stimulation induced a significant decrease in the expression (assessed by MFI) on CD8+ T cells of the adhesion molecules VLA-4 (CD29/CD49d) (P < 0.02) and LFA-1 (P < 0.05). There was a trend towards a similar reduction in ICAM-1 expression, though this did not reach statistical significance (P < 0.06).
Table 2 Surface molecules on repeatedly stimulated CD8+ T cells*
IL-4 - + P†
CD3+ 97 ± 1 98 ± 0.5 NS
CD3+CD8+CD4- 97 ± 1 97 ± 1 NS
CD3+CD4+CD8- 0.2 ± 0.2 0.3 ± 0.2 NS
CD3+CD4-CD8- 2 ± 1 2 ± 1 NS
CD3+CD4+CD8+ 7 ± 6 4 ± 1 NS
CD3+CD8+
CD25 53 ± 9 68 ± 9 NS
CD30 23 ± 7 40 ± 6 0.07
CD40L 14.5 ± 5 14.5 ± 6 NS
VLA-4# 444 ± 97 283 ± 72 0.02
LFA-1# 280 ± 51 194 ± 25 0.05
ICAM-1# 37 ± 10 27 ± 7 0.06
*frequency of positive cells or #mean fluorescence intensity of surface antigens after two rounds of stimulation of CD8+ T cells from atopic asthmatic subjects (mean ± SEM; n = 7; NS, nonsignificant).
† comparison between IL-4-treated and -untreated cells (paired Student's t test)
IL-4 enhances type 2 cytokine production by CD8+ T cells from atopic asthmatic subjects
To investigate whether the presence of IL-4 during CD8+ T cell stimulation alters the cytokine production pattern, CD8+ T cells harvested after two cycles of stimulation/expansion were restimulated with PMA/ionomycin or immobilized anti-CD3 Abs for 24 h to investigate cytokine synthesis and release into supenatants. The patterns of alteration in cytokine secretion following 24 h restimulation with anti-CD3 Abs (data not shown) were similar to that observed after PMA and ionomycin restimulation (Fig. 2).
Figure 2 Exogenous IL-4 during TCR/CD3 stimulation increased type 2 cytokine production in culture supernatants of CD8+ T cells from atopic asthmatic subjects. CD8+ T cells were stimulated with anti-CD3 Abs+IL-2 in the presence or absence of IL-4 for 3 days followed by 4 days expansion with IL-2 alone- this cycle of stimulation/expansion was then repeated. Cells were then harvested and restimulated (1 × 106/mL) with PMA and ionomycin. Supernatants were collected after 24 hours, and cytokines measured by ELISA. Cytokine data are median (quartiles 75) of 10 atopic asthmatic subjects. * p < 0.05
Type 2 cytokine production by repeatedly stimulated CD8+ T cells from atopic asthmatic subjects was increased by the presence of IL-4 during stimulation (Fig. 2). There was a trend towards an increase in IL-4 (median and [range]) production by CD8+ T cells stimulated in the presence of IL-4 (without IL-4, 146 pg/mL [20 to 793 pg/mL]; with IL-4, 389 pg/mL [45 to 4309 pg/mL], P < 0.1). IL-5 levels were significantly increased by the presence of IL-4 (without IL-4, 494 pg/mL [56 to 1612 pg/mL]; with IL-4, 1251 pg/mL [391 to 10936 pg/mL], P < 0.02). IL-13 levels were also significantly increased by the presence of IL-4 during CD8+ T cell stimulation (without IL-4, 1045 pg/mL [378 to 1392 pg/mL]; with IL-4, 1455 pg/mL [580 to 5820 pg/mL], P < 0.01).
Excess IL-4 during repeated stimulation of CD8+ T cells did not significantly alter the levels of IL-2 (without IL-4, 2000 pg/mL [1000 to 4000 pg/mL]; with IL-4, 2135 pg/mL [1000 to 2900 pg/mL]) and IFN-γ (without IL-4, 22 ng/mL [2 to 44 ng/mL]; with IL-4, 23 ng/mL [1 to 44 ng/mL] released in culture by CD8+ T cells.
Discussion
In this study, we show that excess exogenous IL-4 during in vitro CD8+ T cell stimulation amplified the proliferation of CD8+ T cells from atopic asthmatic subjects. Furthermore, excess IL-4 resulted in a decrease in the expression of the adhesion molecules VLA-4 and LFA-1 on CD8+ T cells, along with increased production of the type 2 cytokines IL-5 and IL-13. These data indicate that the activation of CD8+ T cells from asthmatic subjects in the presence of IL-4 increases their capacity to produce type 2 cytokines but not IFN-γ or IL-2. These effects may be important in promoting adverse CD8+ T cell responses to virus infections in asthmatic subjects.
The presence of exogenous IL-4 during TCR/CD3 stimulation of peripheral blood CD8+ T cells induced expansion of cell numbers. It has previously been reported that IL-4 may act as a growth factor for human CD4+ and CD8+ T cell clones and that there is a synergy between IL-2 and IL-4 during CD8+ T cell proliferation [18]. This suggests that stimulation of CD8+ T cells in an environment rich in IL-4 may result in the expansion of CD8+ T cells when exogenous IL-2 is present.
CD8+ T cells exposed to repeated stimulation in the presence of exogenous IL-4 decreased their expression of the α4β1 integrin (VLA-4, CD49d/CD29) and αLβ2 integrin (LFA-1, CD11a/CD18). VLA-4 is a surface marker associated with the capacity of T cells to migrate towards sites of antigen stimulation [19] and it has been previously reported that CD49d membrane expression on Th1 and Th2 clones is markedly reduced by activation [20]. The reduction in the levels of adhesion molecules as a result of activation of CD8+ T cells may play a role in the retention of activated cells at sites of antigenic stimulation. Recently, in an animal model it was reported that type 2 or type 0 CD8+ T cells had decreased surface VLA-4 density and deficient activation-induced LFA-1/ICAM-1-dependent homotypic adhesion in vitro and consequently reduced antiviral activity in vivo [21]. The presence of exogenous IL-4 during stimulation of CD8+ T cells also tended to increase the frequencies of CD30-positive cells. The expression of CD30, a member of the TNF receptor family, by T cells has been previously reported to be primarily regulated by IL-4 [22] and preferentially expressed by type 2 and type 0 cells [20]. These data collectively suggest that the presence of IL-4 in the microenvironment induces a phenotype of cells similar to tissue-resident cells and induces patterns of expression of adhesion molecules associated with adoption of type 2 function. These data suggest our in vitro observations on induction of type 2 functions more likely to truly reflect events occurring in the lungs in vivo.
We have found that peripheral blood CD8+ T cells from atopic asthmatic individuals, when stimulated in the presence of exogenous IL-4, released higher levels of IL-5 and IL-13 as compared with CD8+ T cells stimulated in the absence of IL-4. It has previously been reported that exogenous IL-4 increased production of IL-4 and IL-5 in stimulated neonatal CD8+ T cells [23]. However, it is now known that neonatal T cells are skewed towards a type 2 phenotype and may, therefore, not be representative of adult human CD8+ T cells [24]. We have recently reported induction of type 2 activities in CD8+ T cells from healthy adult subjects during TCR stimulation by the presence of exogenous IL-4 [15]. However, we observed only a tendency to increased IL-5-production and no increased IL-13-production in CD8+ T cells from normal subjects [[15] and unpublished data]. In the present study we extend these findings to confirm that induction by IL-4 of type 2 activities also occurs in CD8+ T cells from asthmatic subjects and demonstrate the deviation in phenotype towards type 2 is more extensive with increases in IL-5 and IL-13 production.
Respiratory virus infections are frequently associated with asthma exacerbations and CD8+ T cells are important in anti-viral immune responses. Given the above data in our in vitro model, we suggest that stimulation of CD8+ T cells during viral infections in an environment enriched in IL-4 as occurs in asthma is likely to favour the induction of type 2 cytokine synthesis by CD8+ T cells. Production of type 2 cytokines by CD8+ T cells during immune responses to viral infections in asthma has the potential to augment allergic inflammation by a number of different mechanisms including each of the type 2 cytokines measured in this study. This hypothesis is lent credence by our recent observations that virus infection and allergen exposure are synergistic risk factors for acute exacerbations of asthma [25].
By antagonising the production of IFN-γ, type 2 cytokines may also have effects on the capacity of CD8+ T cells to clear virus. In our culture conditions, even though exogenous IL-4 did not significantly decrease the levels of IFN-γ, the levels remained stable and the ratio of IL-4/IFN-γ was increased (data not shown). It is quite possible that during respiratory viral infections in asthmatic relative to normal subjects in vivo, where the stimulation conditions are different, IFN-γ production is depressed as a result of these mechanisms. This hypothesis is supported by the observation in mice that elevation of IL-4 levels concurrent with viral infections suppresses or delays activation of virus-specific CD8+ T cells and leads to delayed viral clearance [26-28]. We have previously reported a significant increase in submucosal CD4+ and CD8+ lymphocytes and in epithelial eosinophils in both normal and asthmatic subjects with experimental rhinovirus infections, and in asthmatic subjects eosinophil numbers remained elevated during convalescence [29]. These data suggest that modulation by IL-4 of CD8+ virus-specific immune responses towards a type 2 phenotype may occur in humans in vivo, and that this may be an important mechanism in virus-induced exacerbations of asthma acting both by augmenting allergic inflammation and by diminishing viral clearance. A recent rhinovirus experimental infection study has supported this hypothesis by observing impaired virus clearance and increased symptoms in subjects with increased IL-5/IFN-γ mRNA ratios in induced sputum [30].
Conclusion
In conclusion, our data demonstrate induction of type 2 cytokine production and a tissue-resident phenotype by exogenous IL-4 in CD8+ T cells from atopic asthmatic subjects. Given the excess IL-4 observed in asthma we believe that these mechanisms may play an important role in impairing virus clearance and augmenting allergic inflammation in virus-induced asthma. We are currently testing this possibility in vivo in human experimental virus infection models.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
LAS participated in the design of the study, performed the studies and drafted the manuscript. KR participated in the design of the study. NGP participated in the design of the study. SHC participated in the design of the study. STH conceived of the study. AJC conceived of the study. SLJ conceived of the study, participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript.
Acknowledgements
S.L. Johnston and L.A. Stanciu thank the National Asthma Campaign and British Lung Foundation for their generous support.
==== Refs
Robinson DS Hamid Q Yin S Tsicopoulos A Barkans J Bentley AM Corrigan C Durham SR Kay AB Predominant Th2-like bronchoalveolar T-lymphocyte population in atopic asthma N Engl J Med 1992 326 298 304 1530827
Walker C Bauer W Braun RK Menz G Braun P Schwarz F Hansel TT Villiger B Activated T cells and cytokines in bronchoalveolar lavages from patients with various lung diseases associated with eosinophilia Am J Respir Crit Care Med 1994 150 1038 1048 7921434
Shahid SK Kharitonov SA Wilson NM Bush A Barnes PJ Increased interleukin-4 and decreased interferon-gamma in exhaled breath condensate of children with asthma Am J Respir Crit Care Med 2002 165 1290 1293 11991881 10.1164/rccm.2108082
Johnston SL Pattemore PK Sanderson G Smith S Lampe F Josephs L Symington P O'Toole S Myint SH Tyrrell DAJ Holgate ST Community study of role of viral infections in exacerbations of asthma in 9–11 year old children Br Med J 1995 310 1225 1229 7767192
Nicholson KG Kent J Ireland DC Respiratory viruses and exacerbations of asthma in adults Br Med J 1993 307 982 986 8241910
Wark PA Johnston SL Moric I Simpson JL Hensley MJ Gibson PG Neutrophil degranulation and cell lysis is associated with clinical severity in virus-induced asthma Eur Respir J 2002 19 68 75 11852895 10.1183/09031936.02.00226302
Corne JM Marshall C Smith S Schreiber J Sanderson G Holgate ST Johnston SL Frequency, severity, and duration of rhinovirus infections in asthmatic and non-asthmatic individuals: a longitudinal cohort study Lancet 2002 359 831 834 11897281 10.1016/S0140-6736(02)07953-9
Papadopoulos NG Stanciu LA Papi A Holgate ST Johnston SL A defective type 1 response to rhinovirus in atopic asthma Thorax 2002 57 328 332 11923551 10.1136/thorax.57.4.328
Woodland DL Cell-mediated immunity to respiratory virus infections Curr Opin Immunol 2003 15 430 435 12900275 10.1016/S0952-7915(03)00067-0
Coyle AJ Erard F Bertrand C Walti S Pircher H Le Gros G Virus-specific CD8+ cells can switch to interleukin 5 production and induce airway eosinophilia J Exp Med 1995 181 1229 1233 7869040 10.1084/jem.181.3.1229
Makela MJ Tripp R Dakhama A Park JW Ikemura T Joetham A Waris M Anderson LJ Gelfand EW Prior airway exposure to allergen increases virus-induced airway hyperresponsiveness J Allergy Clin Immunol 2003 112 861 869 14610471 10.1016/S0091-6749(03)02020-7
Legg JP Hussain IR Warner JA Low JL Johnston SL Warner JO Type 1 and type 2 cytokine imbalance in acute respiratory syncytial virus bronchiolitis Am J Respir Crit Care Med 2003 168 633 639 12773328 10.1164/rccm.200210-1148OC
Stanciu LA Shute J Holgate ST Djukanovic R Production of IL-8 and IL-4 by positively and negatively selected CD4+ and CD8+ human T cells following a four-step cell separation method including magnetic cell sorting (MACS) J Immunol Methods 1996 189 107 115 8576572 10.1016/0022-1759(95)00240-5
Stanciu LA Shute J Promwong C Holgate ST Djukanovic R Increased levels of IL-4 in CD8+ T cells in atopic asthma J Allergy Clin Immunol 1997 100 373 378 9314350
Stanciu LA Roberts K Lau LCK Coyle AJ Johnston SL Induction of type 2 activity in adult human CD8 Tcells by repeated stimulation an IL-4 Int Immunol 2001 13 341 348 11222503 10.1093/intimm/13.3.341
American Thoracic Society Standards for the diagnosis and care of patients with chronic obstructive pulmonary disease and asthma AmRev Respir Dis 1987 136 225 244
Stanciu LA Djukanovic R Kearse KP Isolation of T-cell subsets by magnetic cell sorting (MACS) Methods in molecular biology T cell protocols: Development and activation 1999 134 New York: Humana Press 133 141
Spits H Yssel H Takebe Y Arai N Yokota T Lee F Arai K Banchereau J de Vries JE Recombinant interleukin 4 promotes the growth of human T cells J Immunol 1987 139 1142 7 3497196
Nakajima H Sano H Nishimura T Yoshida S Iwamoto I Role of vascular cell adhesion molecule 1/very late activation antigen 4 and intercellular adhesion molecule 1/lymphocyte function-associated antigen 1 interactions in antigen-induced eosinophil and T cell recruitment into the tissue J Exp Med 1994 179 1145 1154 7511681 10.1084/jem.179.4.1145
Almerigogna F D'Elios MM De Carli M Del Prete G Markers of Th1 and Th2 cells Chem Immunol 1996 63 30 50 8934830
Wirth S van der Broek M Frossard CP Hugin AW Leblond I Pircher H Hauser C CD8+ T cells secreting type 2 lymphokines are defective in protection against viral infection Cell Immunol 2000 202 13 22 10873302 10.1006/cimm.2000.1639
Nakamura T Lee RK Nam SY Al-Ramadi BK Koni PA Bottomly K Podack ER Flavell RA Reciprocal regulation of CD30 expression on CD4 J Immunol 1997 158 2090 2098 9036953
Byun D-G Demeure CE Yang LP Shu U Ishihara H Vezzio N Gately MK Delespesse G In vitro maturation of neonatal human CD8 T lymphocytes into IL-4 and IL-5-producing cells J Immunol 1994 153 4862 4871 7963550
Prescott SL Macaubas C Smallacombe T Holt BJ Sly PD Holt PG Development of allergen-specific T-cell memory in atopic and normal children Lancet 1999 353 196 200 9923875 10.1016/S0140-6736(98)05104-6
Green RM Custovic A Sanderson G Hunter J Johnston SL Woodcock A Synergism between allergens and viruses and risk of hospital admission with asthma: case-control study BMJ 2002 324 763 11923159 10.1136/bmj.324.7340.763
Moran TM Isobe H Fernandez-Sesma A Schulman JL Interleukin-4 causes delayed virus clearance in influenza virus-infected mice J Virol 1996 70 5230 5235 8764032
Fischer JE Johnson JE Kuli-Zade RK Johnson TR Aung S Parker RA Graham BS Overexpression of interleukin-4 delays virus clearance in mice infected with respiratory syncytial virus J Virol 1997 71 8672 8677 9343225
Bot A Holz A Christen U Wolfe T Temann A Flavell R von Herrath M Local IL-4 expression in the lung reduces pulmonary influenza-virus-specific secondary cytotoxic T cell responses Virology 2000 269 66 77 10725199 10.1006/viro.2000.0187
Fraenkel DJ Bardin PG Sanderson G Lampe F Johnston SL Holgate ST Lower airways inflammation during rhinovirus colds in normal and in asthmatic subjects Am J Respir Crit Care Med 1995 151 879 886 7881686
Gern JE Vrtis R Grindle KA Swenson C Busse WW Relationship of upper and lower airway cytokines to outcome of experimental rhinovirus infection Am J Respir Crit Care Med 2000 162 2226 2231 11112143
|
16001979
|
PMC1198257
|
CC BY
|
2021-01-04 16:23:25
|
no
|
Respir Res. 2005 Jul 7; 6(1):67
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-67
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-881606096210.1186/1465-9921-6-88ResearchPilot study of losartan for pulmonary hypertension in chronic obstructive pulmonary disease Morrell Nicholas W [email protected] Matthew A [email protected] Peter G [email protected] B Haleema [email protected] Paul J [email protected] Ray J [email protected] Department of Respiratory Medicine, Imperial College School of Medicine, London, UK2 Medical Department, Merck Sharp & Dohme Limited, Hoddesdon, UK2005 1 8 2005 6 1 88 88 11 4 2005 1 8 2005 Copyright © 2005 Morrell et al; licensee BioMed Central Ltd.2005Morrell et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Morbidity in COPD results from a combination of factors including hypoxia-induced pulmonary hypertension, in part due to pulmonary vascular remodelling. Animal studies suggest a role of angiotensin II and acute studies in man concur. Whether chronic angiotensin-II blockade is beneficial is unknown. We studied the effects of an angiotensin-II antagonist losartan, on haemodynamic variables, exercise capacity and symptoms.
Methods
This was a double-blind, randomized, parallel group, placebo- controlled study of 48 weeks duration. Forty patients with COPD and pulmonary hypertension (Tran tricuspid pressure gradient (TTPG) = 30 mmHg) were randomised to losartan 50 mg or placebo. Changes in TTPG were assessed at 3, 6 and 12 months.
Results
There was a trend for TTPG to increase in the placebo group (baseline 43.4 versus 48.4 mmHg at endpoint) and stay constant in the losartan group (baseline 42.8 versus 43.6 mmHg). More patients in the losartan group (50%) than in the placebo group (22%) showed a clinically meaningful reduction in TTPG at any timepoint; these effects seemed more marked in patients with higher baseline TTPG. There were no clear improvements in exercise capacity or symptoms.
Conclusion
In this 12-month pilot study, losartan 50 mg had no statistically significant beneficial effect on TTPG, exercise capacity or symptoms in pulmonary hypertension secondary to obstructive disease. A sub-group of patients with higher TTPG may benefit.
==== Body
Background
Pulmonary hypertension is the main cardiovascular complication of chronic obstructive pulmonary disease (COPD) and is associated with a substantial increase in morbidity and mortality [1-3]. The major characteristic of COPD is chronic airflow limitation that progresses slowly over a period of years and is, by definition, largely irreversible [4]. We previously reported echocardiographic evidence of pulmonary hypertension in over 50% of patients attending a hospital clinic for COPD [5], with the highest prevalence in those with severe disease.
Alveolar hypoxia undoubtedly contributes to pulmonary hypertension in COPD and there is a correlation, albeit poor, between the pulmonary artery pressure and the degree of arterial hypoxaemia [6,7]. Structural remodelling of the pulmonary vasculature as a consequence of chronic alveolar hypoxia is probably the main contributor to the pathogenesis of pulmonary hypertension in COPD [3,8] although a number of other factors may be involved, including hyperinflation and loss of alveolar capillaries in emphysema [9]. In addition, a degree of structural remodelling of the distal pulmonary circulation has been demonstrated in smokers with mild airflow obstruction [10]. Although overt right heart failure is unusual in stable severe COPD, electrocardiographic signs of right heart dysfunction predict a higher mortality [11]. Moreover, pulmonary arterial pressure rises markedly in COPD patients on minimal exertion and it is likely that pulmonary hypertension contributes partly to exercise limitation in these patients. At present, there is no specific pharmacological therapy for pulmonary hypertension in COPD. Long-term oxygen therapy improves survival but does not reverse the pulmonary vascular changes [12].
A possible therapeutic role for losartan, a selective angiotensin-II antagonist, in hypoxic pulmonary hypertension has been suggested by results from animal models [13] and acute studies in man [14-17]. Losartan, which selectively binds to the AT1 receptor, is currently used for reducing central pressures in patients with heart failure [18] and is effective in reducing systemic hypertension. It has also been shown to regress carotid artery hypertrophy [19] and left ventricular hypertrophy [20] and to reduce cardiovascular mortality/morbidity [20] in hypertension.
We hypothesized that long-term administration of losartan may benefit patients with pulmonary hypertension secondary to COPD, using echo-Doppler derived measurements of Tran tricuspid pressure gradient (TTPG) as an index of pulmonary hypertension. We, therefore, undertook a study to evaluate the effects of losartan on TTPG, exercise capacity, quality of life, arterial blood gases and safety in patients with cor pulmonale secondary to severe COPD.
Methods
Patient Selection
Male or female patients, aged 50–80 years, were included. Patients had a clinical history of COPD, evidence of obstructive spirometry (FEV1/FVC ratio ≤ 70%), echocardiographic evidence of pulmonary arterial hypertension (TTPG ≥ 30 mmHg) and sitting systolic blood pressure ≥ 100 mmHg. Exclusion criteria included left ventricular dysfunction (ejection fraction <35%), myocardial infarction, significant renal impairment, recent infective exacerbation of COPD, or concomitant use of vasodilators, β-blockers or potassium-sparing diuretics. Patients were permitted to continue on their regular COPD therapy.
Study design
This pilot study was conducted as a single centre, double-blind, randomized, parallel group comparison of losartan and placebo in patients with pulmonary hypertension secondary to COPD. Following the initial 4-week run-in phase, eligible patients were randomized (week 0) to receive either losartan (Cozaar, Merck & Co, NJ) or placebo for 48 weeks. Losartan 25 mg or a placebo tablet was administered once daily for 1 week. The dose was then increased to 50 mg daily (or placebo equivalent), providing the patient's systolic blood pressure remained ≥ 100 mmHg. The dose could be down titrated once (to 25 mg) if necessary.
The study protocol was approved by the hospital's research ethics committee, and all patients provided written informed consent to study participation.
Study procedures
At the initial visit (week -4) patients underwent physical examination, spirometry, a practice exercise test (see below), echocardiography and safety blood/urine tests. For all eligible patients, tests were repeated at the baseline visit (week 0) when arterial blood gases were also measured.
Patients returned to the clinic at weeks 1, 4, 12, 24, 36 and 48 when blood pressure measurements and safety tests were repeated. Spirometry, echocardiography and exercise testing using a symptom-limited 10 m shuttle walk test [21] were repeated at weeks 12, 24 and at the end of the study (week 48) when arterial gases were sampled again. Echocardiographic assessments were carried out using a Toshiba Powervision model SSA380 ultrasound scanner and a multifrequency probe with a range of 2.5–3.7 MHz (Toshiba Medical Systems, West Sussex, UK). Mean maximum tricuspid valve regurgitation velocity (V) was recorded in m.s-1 and used to calculate the TTPG in mmHg, as previously described [5]. Pulmonary hypertension was defined as a TTPG = 30 mmHg. Right atrial pressure (RAP) was estimated clinically from the height of the jugular venous pressure above the sternal angle, plus 5 cm (mean distance from RA to sternal angle), divided by 1.3 to convert to mmHg. Right ventricular systolic pressure could then be derived (TTPG + RAP).
A quality of life questionnaire (St George's Hospital Respiratory Questionnaire) [22] and Patient Health Survey (SF-36) were performed at weeks 0, 12, 24 and 48. Adverse experiences were monitored throughout the study.
Outcome measures
The primary endpoint was change from baseline (week 0) in TTPG. Pre-specified secondary endpoints included change from baseline in other echocardiographic parameters including right ventricular systolic pressure, peak tricuspid regurgitant velocity, left ventricular fractional shortening and right ventricular wall thickness.
Change from baseline in exercise capacity, breathlessness score after exercise (on a 10 point visual analogue scale: 0 = best, 10 = worst); and quality of life assessments were specified as secondary outcomes.
Statistical methods
This pilot study was designed with a sample size of 44 patients, to give a power of 85% to detect a 25% reduction in TTPG, assuming a baseline TTPG of 46 mmHg increasing by 3 mmHg in the placebo group during the year of the study and a standard deviation of 16 mmHg. This power calculation was based on studies of pulmonary arterial pressure estimated from measurement of TTPG in patients of a similar type [23], and the mean annual rate of increase in pulmonary hypertension in patients affected by COPD [3,24] A reduction of 25% from baseline was chosen as clinically relevant, based on previous studies on vasodilators [25,26] in pulmonary hypertension and therefore a magnitude worthy of further study. Analysis was based on intention-to-treat with last post-randomisation observation carried forward to study end where data were missing. Continuous efficacy variables were analysed by ANOVA. The validity of the assumptions for the ANOVA was confirmed from a review of plots of the residuals against predicted values.
Results
A total of 73 patients with COPD underwent screening echocardiography. Of these, 48 patients entered the run-in phase of the study and of these, 20 were randomized to losartan and 20 to placebo (Figure 1). The 21 females and 19 males ranged in age from 52–79 years (mean 67 years). All had a clinical history of COPD of at least 1-year duration (mean 8.4 years); the mean baseline FEV1 was 0.83 L (range 0.28–1.95 L), with the mean percent of predicted FEV1 being 35.3%. The mean FEV1/FVC ratio was 36% (range 14 to 69%) The distribution of patient characteristics, symptoms and clinical data at baseline (Table 1) revealed no clinically meaningful differences between treatment groups.
Figure 1 Overview of study design, randomization and drop-out rates.
Table 1 Patient characteristics at baseline – mean (SD)
Losartan (n = 20) Placebo (n = 20)
Demography
Male:Female 9:11 10:10
Age (years) 68 (8.4) 66 (7.3)
Spirometry
FEV1 (litres) 0.85 (0.40) 0.82 (0.38)
FVC (litres) 2.5 (0.9) 2.5 (0.9)
Percent predicted FEV1 (%) 37 (19) 33 (14)
Echo findings
TTPG (mmHg) 42.8 (8.8) 43.4 (9.9)
Estimated right atrial pressure (mmHg) 4.0 (0.8) 4.11 (0.9)
Exercise capacity
Number of shuttles 18.3 (9.3) 18.2 (7.6)
Breathlessness score 6.4 (1.9) 6.7 (2.0)
Transtricuspid pressure gradient
Measurements of TTPG were similar at -4 weeks and at baseline and all patients had evidence of tricuspid regurgitation and pulmonary hypertension; the mean TTPG at -4 weeks was 44.6 (11.9) mmHg and at baseline was 43.1 (9.2) mmHg. Based on all available data over the course of the study, the TTPG tended to increase in the placebo group and stay constant in the losartan group with mean increases of 5.03 mmHg and 0.84 mmHg, respectively, (Figure 2). The change from baseline in the losartan group minus that in the placebo group (point estimate of treatment difference) at the end of the study was -4.19 mmHg (95% confidence interval (CI): -13.88 to 5.50 mmHg; p = 0.39). The greatest apparent difference between treatment groups was observed at week 12, when the point estimate was -7.49 (95% CI -15.98, 0.99 mmHg) with a p-value of 0.08.
Figure 2 Transtricuspid pressure gradient in placebo and losartan groups. Data points represent means (95% confidence interval).
Amongst the losartan-treated group there was considerable variability in the TTPG response to treatment between individual patients, with some patients demonstrating a clinically meaningful reduction in TTPG and in others no response or worsening over the 1-year period. The number of patients completing all assessments in the protocol was less than planned. Taking all available post-randomisation echo's available, the frequency of a reduction in TTPG > 25% at any time point was greater in the losartan group, with 8/16 (50%) than in the placebo group, 4/18 (22%), (p = 0.09).
There were no significant differences between the treatment groups in the changes from baseline in the following secondary efficacy endpoints: right atrial pressure, peak tricuspid regurgitant velocity, right ventricular pressure, left ventricular fractional shortening and right ventricular wall thickness (Table 2).
Table 2 Summary of secondary endpoints at baseline and week 48
Losartan Placebo
Baseline
(n = 20) End of study
(n = 16) Change Baseline
(n = 20) End of study
(n = 18) Change p-value
Right atrial pressure (mmHg) 4.0 (0.8) 4.0 (0.6) -0.1 (1.0) 4.1 (0.9) 3.8 (0.0) -0.4 (1.0) 0.56
Peak tricuspid regurgitant velocity (m/s) 3.3 (0.3) 3.2 (0.6) +0.0 (0.7) 3.3 (0.4) 3.5 (0.5) +0.2 (0.4) 0.37
RV systolic pressure (mmHg) 46.9 (8.9) 47.3 (14.5) +0.7 (16.8) 47.5 (10.0) 52.5 (13.4) +4.7 (10.9) 0.41
LV fractional shortening (%) 33 (8) 36 (7) +2.5 (8.6) 31 (6) 32 (9) +1.8 (9.7) 0.84
RV wall thickness (mm) 4.1 (0.8) 4.4 (1.3) +0.4 (1.1) 4.0 (0.6) 4.2 (0.6) +0.3 (0.6) 0.69
Between group p-value for the difference in change from baseline.
LV= left ventricular. RV = right ventricular.
Exploratory analysis (Figure 3) suggested a greater treatment effect on TTPG in those patients with baseline >40 mmHg (16% fall on losartan, 4% rise on placebo) than in those patients with baseline TTPG <40 mmHg (30% rise on losartan, 25% rise on placebo).
Figure 3 Transtricuspid pressure gradient in placebo and losartan groups, split by baseline TTPG. Data points represent means (95% confidence interval).
Exercise capacity and symptom scores
Symptom-limited exercise capacity tended to stay constant over the course of the study in losartan-treated patients (n = 15); there was a mean decrease from baseline to week 48 of 0.2 shuttles completed (Figure 4). For placebo treated patients (n = 17) there was a mean decrease of 2.6 shuttles completed. Data were available for 32 patients only. No significant treatment differences were found; the between-groups point estimate of treatment difference at the end of the study was 2.39 (95% confidence interval: -1.26 to 6.04) shuttles completed (p = 0.19). Exploratory analysis in those patients with baseline TTPG >40 mmHg failed to suggest a differential effect on exercise capacity.
Figure 4 Changes in exercise tolerance measured in placebo and losartan groups by the mean number of shuttles completed in the shuttle walk test. Data points represent means (95% confidence interval).
There were no significant changes from baseline in either treatment group for breathlessness after exercise. For losartan-treated patients, the mean breathlessness score increased from 6.4 (1.9) to 6.8 (1.2) between baseline and week 48; in the placebo group there was a corresponding decrease from 6.7 (2.0) to 6.4 (2.1). The p-value for the point estimate of treatment difference was 0.42. Exploratory analysis in those patients with baseline TTPG >40 mmHg failed to suggest a differential effect on symptom score.
Quality of life
The St George's Hospital Respiratory Questionnaire produces three component scores (symptom, activity and impact) and an overall score; improvement in quality of life is denoted by a reduction in score. There was a mean decrease of 3.61 in overall score for losartan-treated patients and a mean increase of 1.70 for placebo-treated patients (Table 3). There was no significant treatment difference in the overall score; however, the point estimate of treatment difference approached statistical significance (p = 0.06) in favour of losartan for the activity component of the score.
Table 3 Quality of life as assessed on the St George's Hospital Respiratory
Losartan Placebo
Baseline
(n = 19) End of study
(n = 15) Change Baseline
(n = 20) End of study
(n = 18) Change p-value*
Overall score 66 (18) 60 (15) -3.6 (9.2) 65 (14) 66 (19) +1.7 (9.2) 0.11
Symptom component 72 (17) 67 (22) -2.7 (15.5) 71 (16) 66 (20) -5.6 (16.3) 0.61
Activity component 76 (17) 74 (13) -1.2 (10.8) 77 (17) 83 (19) +7.5 (14.1) 0.06
Impact component 57 (23) 49 (18) -5.2 (12.6) 56 (17) 56 (22) +0.7 (13.1) 0.20
*Between group p-value for the difference in change from baseline.
NB. Improvement in QoL is denoted by a reduction in score.
No significant differences were found in any of the dimensions (social, physical, emotional, mental, energy, pain or general health) of the Patient Health Survey (SF-36).
Safety
Thirty-five of the forty patients reported at least one adverse event during the study, 16 (80%) in the losartan group and 19 (95%) in the placebo group. One patient (in the losartan group) died from peritonitis due to diverticular disease. Adverse experiences determined by the investigator to be possibly, probably or definitely drug related, were reported for seven patients (35%) in the losartan group and ten patients (50%) in the placebo group. Treatment was discontinued because of a drug-related adverse event by four patients, one in the losartan group (nausea with rash and hypotension) and three in the placebo group (one case each of rash, orthostatic hypotension, dizziness with tremor).
Blood pressure control was not a significant problem. One patient in the losartan group had asymptomatic hypotension (and discontinued treatment); two patients in the placebo group experienced orthostatic symptoms (one discontinued therapy – see above).
There were no adverse changes in arterial blood gases. Mean (sd) PaO2for losartan-treated patients was 9.1 (1.2) kPa at baseline and 8.1 (1.3) kPa at study end; the corresponding figures for placebo-treated patients were 8.9 (1.8) and 8.6(1.5) kPa, respectively. For PaCO2, mean values were 5.6 (0.8) and 5.8 (0.7) kPa for losartan- and 5.5 (0.7) and 5.5 (1.0) kPa for placebo-treated patients at baseline and study end, respectively.
Discussion
Despite the encouraging data in animal models of chronic hypoxic pulmonary hypertension [27,28], there have been no previous placebo-controlled long-term studies of angiotensin-receptor antagonists in patients with hypoxic lung disease and pulmonary hypertension. In this pilot study, we were unable to demonstrate any statistically significant beneficial effects of losartan in terms of TTPG, exercise capacity, symptoms or quality of life in the treatment group as a whole. There was an early trend towards improvement in pressure gradient and maintenance of exercise capacity in losartan-treated patients; these changes were not sustained throughout the year-long study. There was a trend to deterioration in the placebo group but differences between treatments did not reach statistical significance. In addition, there was a trend towards an improvement (p = 0.06) in the activity component of the St. George's Hospital Respiratory Quality of Life Questionnaire. Exploratory analysis suggests that patients with more severe pulmonary hypertension (TTPG >40 mmHg) may benefit more than the group as a whole in terms of TTPG reduction, but this did not clearly translate into a clinical benefit. Treatment with losartan in this group of patients was well tolerated with a safety profile comparable to placebo. No safety issue specific to patients with pulmonary hypertension secondary to COPD was identified in this pilot study, in particular no significant effect on arterial blood gases.
The rationale behind this study was based on the findings that angiotensin converting enzyme and angiotensin II are both involved in the pulmonary vascular remodelling associated with hypoxic pulmonary hypertension [13,29] Angiotensin converting enzyme expression is increased in the remodelled arteries of patients with plexogenic pulmonary hypertension [30]. In the hypoxic rat model both captopril, an angiotensin converting enzyme inhibitor, and losartan have been shown to prevent the haemodynamic and structural changes of pulmonary hypertension without inhibiting acute hypoxic vasoconstriction [13,31] In addition, right ventricular remodelling in the chronically hypoxic rat is associated with increased angiotensin converting enzyme expression and activity [32]. Furthermore, angiotensin II stimulates hypertrophy of human pulmonary artery smooth muscle cells in culture [33]. In man, there have been but a few studies of the effect of these agents on pulmonary hypertension; almost all have involved acute administration and variable findings have been reported [14-17,34,35]. The only reported study using losartan in pulmonary hypertension secondary to COPD was also in the acute setting [14]. Oral dosing with losartan (50 mg) produced a significant reduction in mean pulmonary artery pressure and total pulmonary vascular resistance; in addition, plasma aldosterone was significantly lower after treatment with losartan compared to placebo.
It is well recognized that the increased pulmonary vascular resistance in COPD may be due to a combination of reduced cross-sectional area of the pulmonary vascular bed in emphysema, hyperinflation, and hypoxic pulmonary vascular remodelling. We would expect losartan to target only the latter. The relative contribution of these factors almost certainly differs between COPD patients, and there may be a sub-population of patients who stand to benefit more than others.
A number of methodological issues need to be considered before rejecting a beneficial effect. Firstly, fewer patients than planned completed the study, largely due to adverse events unrelated to study medication and more related to the elderly nature of the group. This may have reduced our power. Nonetheless, our pre-study assumptions (baseline TTPG of 46 mmHg, placebo group increase of 3 mmHg per annum and a standard deviation of the change from baseline of 16 mmHg) were not far from those observed (baseline TTPG of 43 mmHg, placebo increase of 5 mmHg per annum and standard deviation of 16.8 mmHg) We conclude therefore that a change of this magnitude is unlikely in the group as a whole. A much larger study would be needed to confidently detect or exclude a smaller effect. There does appear to be a greater proportionate effect on TTPF in patients with higher baseline values raising the possibility of further study in this group. Secondly, the choice of assessments requires consideration. Studies of agents shown to be beneficial such as Iloprost [36] and Bosentan [37] have used right heart catheterisation to assess haemodynamics. The use of echocardiography may be less precise than invasive measurements but affords the opportunity of repeated measurements. We were unable to show an increase in exercise tolerance using the shuttle test which is a maximal test, whereas others have shown benefit using the sub-maximal 6-minute walk. The latter may be more relevant to daily life.
Finally, whilst our study was ongoing, other losartan studies [20,38] using higher doses (50–100 mg) have shown beneficial effects in both diabetic nephropathy and hypertension with left ventricular hypertrophy, raising the possibility that more marked effects may have been seen with a higher dose.
Conclusion
In conclusion, this pilot study of the effect of losartan 50 mg on pulmonary hypertension secondary to COPD showed no significant sustained differences between control and treatment groups over the course of one year. There was a trend to early benefit in terms of a slowing of the rate of decline of TTPG and exercise capacity which may warrants further study, particularly in patients with more severe disease, and at higher doses.
List of Abbreviations
AT1 Angiotensin II type 1 receptor
COPD Chronic Obstructive pulmonary disease
FEV1 Forced expiratory volume in 1 second
FVC Forced Vital Capacity
LV Left ventricle
RAP Right atrial pressure
RA Right atrium
RV Right ventricle
TTPG Transtricuspid pressure gradient
Conversion Factor
To convert KPa to mmHg, multiply by 7.5
Competing interests
This study was funded by Merck Sharp & Dohme Ltd. At the time of the work, PR & RB were employees of Merck Sharp & Dohme and may own stock/stock options.
Authors' contributions
NM, PR, RB were involved in the design of the study. NM, PP, MH, BHS were involved in the acquision of clinical data. NM & PR wrote the manuscript. All authors read and approved the final manuscript.
Acknowledgements
We acknowledge the assistance of Brenda Mullinger, ScopeMedical Ltd. in drafting the manuscript.
==== Refs
Cooper R Ghali J Simmons BE Castaner A Elevated pulmonary artery pressure. An independent predictor of mortality Chest 1991 99 1 112 20 1824624
Pride NB Burrows B Calverley, Pride Development of impaired lung function: Natural history and risk factors Chronic obstructive pulmonary disease 1996 Longman
Weitzenblum E Chaouat A Oswald M Peacock AJ Pulmonary hypertension due to chronic hypoxic lung disease Pulmonary circulation 1996 London: Chapman and Hall 157 170
Siafakas NM Vermeire P Pride NB Paoletti P Gibson J Howard P Yernault JC Decramer M Higenbottam T Postma DS Optimal assessment and management of chronic obstructive pulmonary disease (COPD). The European Respiratory Society Task Force Eur Respir J 1995 8 8 1398 420 7489808 10.1183/09031936.95.08081398
Higham MA Dawson D Joshi J Nihoyannopoulos P Morrell NW Utility of echocardiography in assessment of pulmonary hypertension secondary to COPD Eur Respir J 2001 17 3 350 5 11405510 10.1183/09031936.01.17303500
Whitaker W Pulmonary hypertension in congestive heart failure complicating chronic lung disease Q J Med 1954 23 89 57 72 13134461
Yu PN Lovejoy FW JrJoos HA Nye RE JrMcCann WS Studies of pulmonary hypertension. I. Pulmonary circulatory dynamics in patients with pulmonary emphysema at rest J Clin Invest 1953 32 2 130 7 13022771
Heath D Peacock AJ Remodeling of the pulmonary vasculature in hypoxic lung disease Pulmonary circulation 1996 London: Chapman and Hall 171 179
Burrows B Kettel LJ Niden AH Rabinowitz M Diener CF Patterns of cardiovascular dysfunction in chronic obstructive lung disease N Engl J Med 1972 286 17 912 8 5013974
Santos S Peinado VI Ramirez J Melgosa T Roca J Rodriguez-Roisin R Barbera JA Characterization of pulmonary vascular remodelling in smokers and patients with mild COPD Eur Respir J 2002 19 4 632 8 11998991 10.1183/09031936.02.00245902
Incalzi RA Fuso L De Rosa M Di Napoli A Basso S Pagliari G Pistelli R Electrocardiographic signs of chronic cor pulmonale: A negative prognostic finding in chronic obstructive pulmonary disease Circulation 1999 99 12 1600 5 10096937
Wilkinson M Langhorne CA Heath D Barer GR Howard P A pathophysiological study of 10 cases of hypoxic cor pulmonale Q J Med 1988 66 249 65 85 3174923
Morrell NW Atochina EN Morris KG Danilov SM Stenmark KR Angiotensin converting enzyme expression is increased in small pulmonary arteries of rats with hypoxia-induced pulmonary hypertension J Clin Invest 1995 96 4 1823 33 7560074
Kiely DG Cargill RI Wheeldon NM Coutie WJ Lipworth BJ Haemodynamic and endocrine effects of type 1 angiotensin II receptor blockade in patients with hypoxaemic cor pulmonale Cardiovasc Res 1997 33 1 201 8 9059545 10.1016/S0008-6363(96)00180-0
Peacock AJ Matthews A Transpulmonary angiotensin II formation and pulmonary haemodynamics in stable hypoxic lung disease: the effect of captopril Respir Med 1992 86 1 21 6 1565813
Pison CM Wolf JE Levy PA Dubois F Brambilla CG Paramelle B Effects of captopril combined with oxygen therapy at rest and on exercise in patients with chronic bronchitis and pulmonary hypertension Respiration 1991 58 1 9 14 1649486
Zielinski J Hawrylkiewicz I Gorecka D Gluskowski J Koscinska M Captopril effects on pulmonary and systemic hemodynamics in chronic cor pulmonale Chest 1986 90 4 562 5 3530649
Crozier I Ikram H Awan N Cleland J Stephen N Dickstein K Frey M Young J Klinger G Makris L Losartan in heart failure. Hemodynamic effects and tolerability. Losartan Hemodynamic Study Group Circulation 1995 91 3 691 7 7828295
Olsen M Wachtell K Neland K Rokkedal J Ibsen H Dige-Petersen H The effect of losartan versus atenolol on carotid artery hypertrophy in essential hypertension. A LIFE substudy Circulation 2002 106 19 11574
Dahlof B Devereux RB Kjeldsen SE Julius S Beevers G Faire U Fyhrquist F Ibsen H Kristiansson K Lederballe-Pedersen O Cardiovascular morbidity and mortality in the Losartan Intervention For Endpoint reduction in hypertension study (LIFE): a randomised trial against atenolol Lancet 2002 359 9311 995 1003 11937178 10.1016/S0140-6736(02)08089-3
Singh SJ Morgan MD Hardman AE Rowe C Bardsley PA Comparison of oxygen uptake during a conventional treadmill test and the shuttle walking test in chronic airflow limitation Eur Respir J 1994 7 11 2016 20 7875275
Jones PW Quirk FH Baveystock CM Littlejohns P A self-complete measure of health status for chronic airflow limitation. The St. George's Respiratory Questionnaire Am Rev Respir Dis 1992 145 6 1321 7 1595997
Tramarin R Torbicki A Marchandise B Laaban JP Morpurgo M Doppler echocardiographic evaluation of pulmonary artery pressure in chronic obstructive pulmonary disease. A European multicentre study. Working Group on Noninvasive Evaluation of Pulmonary Artery Pressure. European Office of the World Health Organization, Copenhagen Eur Heart J 1991 12 2 103 11 2044542
MacNee W Pathophysiology of cor pulmonale in chronic obstructive pulmonary disease. Part One Am J Respir Crit Care Med 1994 150 3 833 52 8087359
McLaughlin VV Genthner DE Panella MM Rich S Reduction in pulmonary vascular resistance with long-term epoprostenol (prostacyclin) therapy in primary pulmonary hypertension N Engl J Med 1998 338 5 273 7 9445406 10.1056/NEJM199801293380501
Stumpe KO Schmengler K Bette L Overlack A Kolloch R Persistent hemodynamic and clinical improvement after captopril in patients with pulmonary hypertension Herz 1986 11 4 217 25 3093346
Morrell NW Morris KG Stenmark KR Role of angiotensin-converting enzyme and angiotensin II in development of hypoxic pulmonary hypertension Am J Physiol 1995 269 4 Pt 2 H1186 94 7485548
Hales CA Rouse ET Kazemi H Failure of saralasin acetate, a competitive inhibitor of angiotensin II, to diminish alveolar hypoxic vasoconstriction in the dog Cardiovasc Res 1977 11 6 541 6 603879
Chassagne C Eddahibi S Adamy C Rideau D Marotte F Dubois-Rande JL Adnot S Samuel JL Teiger E Modulation of angiotensin II receptor expression during development and regression of hypoxic pulmonary hypertension Am J Respir Cell Mol Biol 2000 22 3 323 32 10696069
Orte C Polak JM Haworth SG Yacoub MH Morrell NW Expression of pulmonary vascular angiotensin-converting enzyme in primary and secondary plexiform pulmonary hypertension J Pathol 2000 192 3 379 84 11054722 10.1002/1096-9896(2000)9999:9999<::AID-PATH715>3.0.CO;2-Q
Nong Z Stassen JM Moons L Collen D Janssens S Inhibition of tissue angiotensin-converting enzyme with quinapril reduces hypoxic pulmonary hypertension and pulmonary vascular remodeling Circulation 1996 94 8 1941 7 8873672
Morrell NW Danilov SM Satyan KB Morris KG Stenmark KR Right ventricular angiotensin converting enzyme activity and expression is increased during hypoxic pulmonary hypertension Cardiovasc Res 1997 34 2 393 403 9205554 10.1016/S0008-6363(97)00049-7
Morrell NW Upton PD Kotecha S Huntley A Yacoub MH Polak JM Wharton J Angiotensin II activates MAPK and stimulates growth of human pulmonary artery smooth muscle via AT1 receptors Am J Physiol 1999 277 3 Pt 1 L440 8 10484450
Bertoli L Fusco M Lo Cicero S Micallef E Busnardo I Influence of ACE inhibition on pulmonary haemodynamics and function in patients in whom beta-blockers are contraindicated Postgrad Med J 1986 62 Suppl 1 47 51 3022273
Patakas D Georgopoulos D Rodini H Christaki P Effects of captopril in patients with chronic obstructive pulmonary disease and secondary pulmonary hypertension Postgrad Med J 1988 64 749 193 5 3050941
Hoeper MM Schwartz M Ehlerding S Long term treatment of primary hypertension with aerosolized Iloprost, a postacyclin analogue N Eng J Med 2000 342 1866 70 10.1056/NEJM200006223422503
Channick RN Simonneau G Sitbon O Effects of the dual endothelin-antagonist bosentan in patients with pulmonary hypertension: a randomised placebo-controlled study Lancet 2001 358 1119 23 11597664 10.1016/S0140-6736(01)06250-X
Brenner BM Cooper ME de Zeeuw D Keane WF Mitch WE Parving HH Remuzzi G Snapinn SM Zhang Z Shahinfar S Effects of losartan on renal and cardiovascular outcomes in patients with type 2 diabetes and nephropathy N Engl J Med 2001 345 12 861 9 11565518 10.1056/NEJMoa011161
|
16060962
|
PMC1198258
|
CC BY
|
2021-01-04 16:23:25
|
no
|
Respir Res. 2005 Aug 1; 6(1):88
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-88
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-941609823310.1186/1465-9921-6-94ResearchIdentification of HLA-DRPheβ47 as the susceptibility marker of hypersensitivity to beryllium in individuals lacking the berylliosis-associated supratypic marker HLA-DPGluβ69 Amicosante Massimo [email protected] Floriana [email protected] Milton [email protected] Richard H [email protected] Paola [email protected] den Berg-Loonen Ella [email protected] Cesare [email protected] Department of Internal Medicine, University of Rome "Tor Vergata", Rome, Italy2 Pulmonary, Allergy and Critical Care Division, Department of Medicine, Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania, USA3 Institute of Cell Biology, National Research Council, Monterotondo (Rome), Italy4 Tissue Typing Laboratory, University Hospital Maastricht, Maastricht, The Netherlands2005 14 8 2005 6 1 94 94 31 3 2005 14 8 2005 Copyright © 2005 Amicosante et al; licensee BioMed Central Ltd.2005Amicosante et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Susceptibility to beryllium (Be)-hypersensitivity (BH) has been associated with HLA-DP alleles carrying a glutamate at position 69 of the HLA-DP β-chain (HLA-DPGlu69) and with several HLA-DP, -DQ and -DR alleles and polymorphisms. However, no genetic associations have been found between BH affected subjects not carrying the HLA-DPGlu69 susceptibility marker.
Methods
In this report, we re-evaluated an already described patient populations after 7 years of follow-up including new 29 identified BH subjects. An overall population 36 berylliosis patients and 38 Be-sensitization without lung granulomas and 86 Be-exposed controls was analysed to assess the role of the individual HLA-class II polymorphisms associated with BH-susceptibility in HLA-DPGlu69 negative subjects by univariate and multivariate analysis.
Results
As previously observed in this population the HLA-DPGlu69 markers was present in higher frequency in berylliosis patients (31 out of 36, 86%) than in Be-sensitized (21 out of 38, 55%, p = 0.008 vs berylliosis) and 41 out of 86 (48%, p < 0.0001 vs berylliosis, p = 0.55 vs Be-sensitized) Be-exposed controls.
However, 22 subjects presenting BH did not carry the HLA-DPGlu69 marker. We thus evaluated the contribution of all the HLA-DR, -DP and -DQ polymorphisms in determining BH susceptibility in this subgroup of HLA-Glu69 subjects. In HLA-DPGlu69-negatives a significant association with BH was found for the HLA-DQLeu26, for the HLA-DRB1 locus residues Ser13, Tyr26, His32, Asn37, Phe47 and Arg74 and for the HLA-DRB3 locus clusterized residues Arg11, Tyr26, Asp28, Leu38, Ser60 and Arg74. HLA-DRPhe47 (OR 2.956, p < 0.05) resulting independently associated with BH. Further, Be-stimulated T-cell proliferation in the HLA-DPGlu69-negative subjects (all carrying HLA-DRPhe47) was inhibited by the anti-HLA-DR antibody (range 70–92% inhibition) significantly more than by the anti-HLA-DP antibody (range: 6–29%; p < 0.02 compared to anti-HLA-DR) while it was not affected by the anti-HLA-DQ antibody.
Conclusion
We conclude that HLA-DPGlu69 is the primary marker of Be-hypersensitivity and HLA-DRPhe47 is associated with BH in Glu69-negative subjects, likely playing a role in Be-presentation and sensitization.
==== Body
Background
Due to its unique chemical-physical properties, beryllium (Be) compounds continue to be used in aerospace, ceramics, defence, electronics and telecommunication industries where inhalation of Be dust is the cause of Be-hypersensitivity (BH) in susceptible individuals [1]. Among subjects developing Be-hypersensitivity, all show sensitization, i.e. T-cell reactivity to Be revealed by either a blood or a bronchalveolar lavage cell test. Less than 50% of subjects with BH present which chronic disease [1-3] i.e., with chronic granuloma formation in the lung maintained by the accumulation in the lower respiratory tract of CD4+ T-cells responding to Be as a specific antigen/hapten [4], presenting an effector-memory phenotype [5,6] and producing Th1 cytokines upon Be stimulation [4-6].
The observation that beryllium disease affects only 1 to 16% of Be-exposed individuals led to the hypothesis that genetic susceptibility may play an important role in the pathogenesis of this disease [1]. In 1993, the HLA-DP supratypic variant characterized by a glutamic acid at position 69 of the HLA-DP molecule β chain (DPGlu69) was identified as a genetic marker of susceptibility to BH, an observation subsequently confirmed by seven independent studies [7-14]. Two independent studies have also identified the HLA-DPGlu69 marker as the immune response gene responsible for presentation of Be to Be-specific T-cells [15,16] and an immunochemical study has suggested that the structural basis for Be presentation by the HLA-DPGlu69 positive molecule is in its unique ability to bind beryllium with high affinity possibly in the context of a coordination bond formed by the contribution of other electron donor groups present in the fourth pocket of the peptide binding groove of the HLA-DP molecule [17]. Further, antibody inhibition studies have shown that Be-presentation to blood and lung T-cells in DPGlu69-positive subjects is inhibited almost exclusively by anti-HLA-DP antibodies [16,18], strongly indicating HLA-DPGlu69 as the immune response gene used by DPGlu69-positive subjects i.e., about 80% of the BH affected population [7-14].
In contrast, the HLA gene which might function as the immune response gene in DPGlu69-negative BH-affected subjects i.e., in the remaining 20% of the BH affected population, has not yet been determined.
Previous studies have identified the HLA-DRB1 alleles belonging to the *01 group [13] as negatively associated with berylliosis, while the HLA-DRB1 variants Ser11 [12], Tyr26 [10], Asn37 [12], Glu71 [12] and Arg74 [10] and the HLA-DQ variant Gly86 [12] were positively associated with BH. Analysis of the role of these markers has, however, been hampered by the small size of the populations examined in most studies. In all studies published so far, the putative susceptibility markers covered only 40 to 50% of the DPGlu69-negative subjects. In this context, our previous study [10] on 45 individuals affected by beryllium sensitization with or without demonstrable lung granulomas, showed that HLA-DR Arg74 and Tyr26 were associated with sensitization without lung granulomas, and HLA-DP Glu69 with sensitization accompanied by lung granulomas, thereby suggesting a different role for Glu69 and these markers [10]. However, in the HLA-DPGlu69 negative subjects reported in the Saltini et al. study population, HLA-DR Arg74 and Tyr26 were expressed only by 11 out of 19 DPGlu69 negative sensitized subjects 10 of which without and one with demonstrable lung granulomas [10]. In another study in the field conducted by Rossman et al. [12] evaluating 56 BH affected subjects, four out of seven DPGlu69 negative patients carried either DRAsn37, DRGlu71 or DQGly86 [12]. Finally, Maier and co-workers [13], in 19 HLA-DP Glu69-negative BH subjects, found that HLA-DRB1*13 alleles were associated with BH susceptibility; however, they were only expressed by 12 of these subjects.
In order to search for this(these) disease associated immune response gene(s), we re-evaluated a previously described population [10] after a follow up of 7 years that allowed us to extend the study to other 29 newly identified BH subjects (14 with biopsy proven lung granulomas and 15 with Be-sensitization without lung granulomas) for a total number of 74 BH subjects and 86 Be-exposed controls. This panel included a sufficiently large number of DPGlu69-negative subjects to analyze phenotypic frequencies of all aminoacid variants of the HLA-DPB1, -DQB1, -DRB1, -DRB3, -DRB4 and DRB5 genes in BH affected and Be exposed controls.
Methods
Study population
The study population, already described in part in a previous study [10], includes 86 Be exposed healthy controls and 74 subjects affected by beryllium hypersensitivity, all working in the same beryllium manufacturing plant, 45 of whom (23 Be-sensitized subjects and 22 berylliosis affected i.e., Be-sensitized subjects with biopsy proven lung granulomas) have been already described in the Saltini et al. report [10]. Study subjects are categorized as (i) Be-exposed controls, when having negative blood Be-LPT test, (ii) Be-sensitized, when having 2 blood BeLPT positive tests and (iii) berylliosis-affected when having 2 blood BeLPT positive tests and/or biopsy proven lung granulomas [10]. While 4 control subjects were diagnosed with beryllium sensitization and 3 with berylliosis during the 7-years follow up, none of the subjects in the previous study progressed from sensitization to berylliosis.
Overall the population in this report included 36 berylliosis (age 40 ± 7 years; 33 Caucasians, 2 African-Americans and 1 Asian; 32 males and 4 females; mean duration of Be-exposure 11 ± 7 years) and 38 showed Be-sensitization without lung granulomas detected by trans-bronchial biopsy (age 43 ± 9 years; 37 Caucasians and 1 Afro-American; 31 males and 7 females; mean age of Be-exposure 17 ± 9 years) and 86 Be-exposed controls (age 44 ± 9 years; 81 Caucasians, 2 African-American, 2 Hispanics, 1 Asian; 71 males and 15 females; mean duration of Be-exposure 16 ± 11 years).
High resolution HLA class II typing
High resolution HLA-class II typing for the HLA-DPB1, DQB1, DRB1, DRB3, DRB4, DRB5 loci were performed by standard protocols as already reported [10].
Beryllium lymphocyte proliferation test (Be-LPT)
The Be-LPT as measures of the T-cell lymphocytes response against Be in peripheral blood has been performed by standard method [19]. Briefly, peripheral blood mononuclear cells (PBMC) were tested against three doses beryllium sulfate (BeSO4*4H2O) at 1, 10, and 100 μM at 3 and 5 days. T-cell proliferation were evaluated by tritium (3H) labelled thymidine incorporation and a stimulation index (SI) calculated as the ratio of the radioactivity (counts per minute) of beryllium-exposed cell cultures to the count rate of unstimulated cultures. A test was defined as abnormal when two or more stimulation index values of six possible values exceeded the normal standard ratio of 3.0 (stimulated to unstimulated) as the cut-off.
Lymphocyte proliferation to Be salt and monoclonal antibody (MoAb) inhibition of lymphocyte activation
T-cell proliferation in response to BeSO4 and inhibition by anti-HLA class II MoAbs were performed as previously described [18]. Briefly, PBMCs obtained from patients with BH were isolated from heparinized whole blood by density centrifugation on Ficoll Hypaque gradient. PBMCs (2 × 105 cells/well) were then cultured in 96-well flat-bottomed microtiter plates in RPMI 1640 tissue culture medium supplemented with 2 mM L-glutamine, 10% fetal bovine serum, 100 U/ml penicillin, and 100 μg/ml streptomycin in the presence of beryllium sulfate (BeSO4*4H2O) at 10–50–100 μM (all reagents form Sigma Co., St. Louise, MO). Phytohemoagglutinin (PHA, 5 μg/ml, Sigma) and Candida albicans (10 μg/ml) were used as positive controls. T lymphocyte proliferation was measured by [3H]TdR incorporation. Cells were pulsed with 1 μCi of [3H]TdR (Amersham International, Amersham, UK) after 5 days of culture and harvested onto glass fiber filters 18 hours later. Proliferation was measured as 3H-TdR incorporation by liquid scintillation spectroscopy and the test was scored as positive in the presence of a greater than twofold proliferation index. Protein-A sepharose purified MoAb directed against HLA-DR (L243) [15], HLA-DP (B7/21) [15], HLA-DQ (L2) [15], HLA-class I (W6/32) [15] were used at increasing concentrations (10, 20 and 50 μg/ml) to inhibit antigen presentation and lymphocyte proliferation as previously described [4,15,18]. The 19 kDa Mycobacterium tuberculosis (MTB19) protein monoclonal antibody HYT6 [20] was used as control.
Statistical analysis
Statistical analysis was carried out as previously described [10,21,22]. Phenotypic frequency data are expressed as percentages with Odds Ratio (OR) with respect to the Be-exposed control group when appropriate. Comparisons between phenotypic frequencies in the study groups were done by χ2 test with the Yates correction where necessary. Linkage disequilibrium analysis was carried out as previously described [20]. Forward and stepwise multiple logistic regression multivariate analysis were applied for identifying independent parameter(s) in multiple comparisons. Be-stimulated lymphocyte proliferation data are expressed as mean ± standard deviation of the mean (SD). Comparisons between groups in Be-lymphocyte proliferation data were done using the Student's t test with Welch's correction when appropriate. All the statistical analysis were carried out with the SPSS (SPSS inc., Chicago, IL) and GraphPad Prism (GraphPad Software Inc., San Diego, CA) packages.
Results
The allelic frequencies for HLA-DPB1, DQB1 and DRB1, 3, 4 and 5 in general population are reported in the tables 1–4 of the additional file (supported material.pdf).
Similarly to the previous study on this population [10] the HLA-DPGlu69 marker was carried with higher frequency by berylliosis affected (31 out of 36, 86%) than subjects with Be-sensitization without granuloma (21 out of 38, 55%, p = 0.008 vs berylliosis affected) and Be-exposed controls (41 out of 86, 48%, p < 0.0001 vs berylliosis affected, p = 0.55 vs Be-sensitized).
The HLA-DPGlu69 has been previously proposed as a marker of progression from the sensitization state to the lung granulomatous reaction of chronic beryllium disease [10,13]. In this study population there were no cases of progression from systemic sensitization to lung disease notwithstanding the substantial follow-up period of 7.0 ± 3.7 years from the first positive Be-LPT test, while 4 control subjects were diagnosed with beryllium sensitization and 3 with berylliosis during the 7-years follow up. Hence, we could not directly look at the HLA-DPGlu69 association with disease progression. In addition, the frequency of the HLA-DPGlu69 homozygosity, another marker associated with disease progression [13] was higher in the disease affected population compared to the sensitized and the Be-exposed control population [5 out of 86 healthy exposed controls (5.8%), 3 out of 38 sensitized without disease (7.9%; p = 0.97 compared to controls) and 8 out of 36 disease affected (22.2%; p = 0.06 compared to controls; p = 0.10 compared to the sensitized)] although the difference was not statistically significant.
A total number of 22 BH subjects (17 Be sensitized and 5 berylliosis affected) and 45 Be-exposed controls were HLA-DPGlu69 negative. They did not differ from the DPGlu69-positive (neither the BH affected nor the Be-exposed controls) in terms of gender, ethnicity, age or length of Be-exposure (p > 0.05, all comparisons). The allelic frequencies for HLA-DPB1, DQB1 and DRB1, 3, 4 and 5 in HLA-DPGlu69 negative subjects are reported in the tables 5–8 of the additional file (supported material.pdf).
This subgroup of HLA-DPGlu69 negative subjects was analyzed for the distribution of all HLA class II polymorphic aminoacid residues, with the exception of HLA-DPGlu69, by univariate analysis. No associations were found between any of the HLA-DP polymorphic residues and BH. Strikingly, among the polymorphic residues of the HLA-DR β-chain coded for by the HLA-DRB1 locus, residues Ser13, Tyr26, His32, Asn37, Phe47 and Arg74 were associated with BH (Table 1). Similarly, the HLA-DR β-chain HLA-DRB3 locus polymorphic residues Arg11, Tyr26, Asp28, Leu38, Ser60 and Arg74 were found associated to BH (Table 1). No polymorphisms associated with BH were found in the HLA-DRB4 and DRB5 loci. Finally, a statistically significant association with BH in HLA-DPGlu69 negatives was found for the HLA-DQB1 gene polymorphic residue Leu26 (Table 1).
Table 1 Phenotypic frequencies of the polymorphisms found associated with Be-hypersensitivity in HLA-DPGlu69 negative subjects.
Be-exposed controls (n = 45) Be-Hypersensitive (n = 22)
HLA-DRB1 polymorphisms1 N positive subjects (%) N positive subjects (%) OR2 p3
Ser13 28 (62.2%) 20 (90.9) 6.07 0.015
Tyr26 7 (15.6%) 10 (45.5%) 4.52 0.009
His32 21 (46.7%) 17 (77.3%) 3.89 0.017
Asn37 17 (37.8%) 16 (72.7%) 4.39 0.007
Phe47 30 (66.7%) 21 (95.5%) 10.50 0.011
Arg74 7 (15.6%) 10 (45.5%) 4.52 0.009
HLA-DQB1 polymorphism4
Leu26 32 (71.1%) 21 (95.5%) 8.53 0.021
HLA-DRB3 polymorphisms5 N = 30 N = 17
Arg11 8 (26.7%) 12 (70.6%) 6.60 0.008
Tyr26 8 (26.7%) 12 (70.6%) 6.60 0.008
Asp28 8 (26.7%) 12 (70.6%) 6.60 0.008
Leu38 8 (26.7%) 12 (70.6%) 6.60 0.008
Ser60 8 (26.7%) 12 (70.6%) 6.60 0.008
Arg74 8 (26.7%) 12 (70.6%) 6.60 0.008
1. HLA-DRB1 polymorphisms found associated with Be-hypersensitivity in HLA-DPGlu69 negative subjects among the overall HLA-DRB1 polymorphic variants analyzed at positions: 9, 10, 11, 12, 13, 16, 26, 28, 30, 32, 37, 38, 47, 57, 58, 60, 67, 70, 71, 73, 74, 77, 85, 86.
2. Odds ratio with respect to Be-exposed controls.
3. p value (χ2 analysis) with respect to Be-exposed controls.
4. HLA-DQB1 polymorphisms found associated with Be-hypersensitivity in HLA-DPGlu69 negative subjects among the overall HLA-DQB1 polymorphic variants analyzed at positions: 9, 13, 14, 23, 26, 28, 30, 37, 38, 45, 46, 47, 52, 53, 55, 56, 57, 66, 67, 70, 71, 74, 75, 77, 84, 85, 86, 87, 89, 90.
5. HLA-DRB3 polymorphisms found associated with Be-hypersensitivity in HLA-DPGlu69 negative subjects among the overall HLA-DRB3 polymorphic variants analyzed at positions: 8, 11, 26, 28, 30, 37, 38, 39, 51, 57, 58, 60, 67, 74, 77, 86. No polymorphisms were found associated to BH in HLA-DPGlu69 negatives in the HLA-DRB4 and DRB5 loci and HLA-DP locus.
However, as a linkage disequilibrium could exist between the HLA-DRB1 gene coded residues and other HLA-DRB1, -DRB3 and -DQB1 loci, in order to identify the independently associated residue(s) in the HLA-DPGlu69 negative BH subjects, multiple logistic regression models were carried out on all the HLA variants above. As a result, only HLA-DRPhe47 (OR 2.956, p < 0.05) was identified as independently associated with BH in the HLA-DPGlu69-negative subgroup, hence suggesting that the other HLA-DRB1, -DRB3 and -DQB1 loci shown in table 1 could be associated with BH due to linkage disequilibrium with HLA-DRPhe47. Of the 22 HLA-DPGlu69 negative subjects with BH, 21 were HLA-DRPhe47 (16 Be sensitized and 5 berylliosis affected) and only one, a Be sensitized individual, was HLA-DPGlu69 and HLA-DRPhe47 negative.
Further, in order to identify which of the HLA isotypic molecules associated with BH could function as the restriction elements of Be-stimulated T-cell proliferation in HLA-DPGlu69-negative subjects, we analyzed PBMC proliferation in response to BeSO4 in a subgroup of 15 BH affected subjects, using antibodies directed against HLA-DR, HLA-DQ and HLA-DP as probes to identify the antigen presentation restricting molecule.
In four HLA-DPGlu69-negatives, Be-stimulated T-cell proliferation was inhibited by the anti-HLA-DR MoAb (range 70–92% inhibition) significantly more than by the anti-HLA-DP MoAb (range: 6–29%; p < 0.02 compared to anti-HLA-DR) while it was not affected at all by the anti-HLA-DQ, anti-HLA class I or the anti-MTB19 control antibody (Figure 1), suggesting a role for the HLA-DR molecule in the presentation of Be in the HLA-DPGlu69-negative subjects. All the four HLA-DPGlu69-negative subjects carried the HLA-DRPhe47 polymorphism.
Figure 1 Inhibition of beryllium (BeSO4)-induced proliferation, by MoAbs directed against HLA-DR, HLA-DP, HLA-DQ, HLA-class I and the 19 kDa M. tuberculosis protein in PBMC from BH subjects carrying or not the HLA-DPGlu69 and the HLA-DRPhe47 markers. They were 4 HLA-DPGlu69 negative/HLA-DRPhe47 positive (3 sensitized and 1 berylliosis affected), 3 HLA-DPGlu69 positive/HLA-DRPhe47 negative (0 sensitized and 3 berylliosis affected) and 8 HLA-DPGlu69 positive/HLA-DRPhe47 positive (3 sensitized and 5 berylliosis affected). On the ordinate is shown the percentage of inhibition (with respect to the MoAbs untreated cells) of T-cell proliferation obtained by co-culturing the PBMC from berylliosis patients with BeSO4 in the presence of each MoAb reported on the abscissa (anti-HLA-DR: DR, anti-HLA-DP: DP, anti-HLA-DQ: DQ, anti-HLA-class I: C.I, anti-19 kDa M. tuberculosis: Mtb).
In contrast, in the three subjects who were HLA-DPGlu69 positive and HLA-DRPhe47 negative, proliferation was completely inhibited by the anti-HLA-DP MoAb (range 68–100%) but not by anti-HLA-DR (range 7–14%, p < 0.05 compared to anti-HLA-DP), nor by anti-HLA-DQ, anti-HLA class I or the anti-MTB19 control antibody (Figure 1). Finally, in the eight subjects carrying both HLA-DPGlu69 and HLA-DRPhe47, the proliferative response to BeSO4 was always inhibited by the anti-HLA-DP (range: 63–100%) and variable inhibited by anti-HLA-DR antibodies (range: 0–94%), with the inhibition by anti HLA-DP being significantly stronger than the anti HLA-DR (paired t-test, p < 0.01).
Discussion
Similarly to our previous report [10] and consistently with the more recent study by McCanlies et al. [14], the re-evaluation of this patient population shows a higher prevalence of HLA-DPGlu69 among the subjects with lung granulomas compared to the Be-sensitized without lung involvement (86% vs 55%, p = 0.008). However, not having identified any case of disease progression from beryllium sensitization to lung disease during the 7.0 years follow-up, we could not formally assess the association of the HLA-DPGlu69 marker with progression from sensitization to disease, although the knowledge that HLA-DPGlu69 is the primary immune response gene of beryllium hypersensitivity [15-17] makes it attractive to hypothesize that the gene might induce a stronger immune reaction hence inducing granuloma formation as suggested by Maier et al. [13] in which 11 out of 12 subjects progressed from sensitization to disease status were HLA-DPGlu69 positives [13], as well as in more recent publication in which Be-sensitized subjects progress to berylliosis at a rate of 6–8% per year [23]. However, it could be also considered that all these data could be inferred by the possibility of misdiagnosis in the identification of the Beryllium induced granuloma.
The finding that a sizeable fraction of BH affected subjects, varying from 3 to 27% in published reports [7-14], do not carry HLA-DPGlu69 has indicated that other HLA molecules may provide the restriction element of Be-stimulated T-cell proliferation and may be implicated in the pathogenesis of susceptibility to BH.
In this regard, with all the limitations imposed by the need of performing specific antigen presentation studies using specific reagents such as HLA-DRPhe47 restricted antigen presenting cells and/or HLA-DRPhe47-engineered transfectants as already made for HLA-DPGlu69 [15], the T-cell studies using isotype-specific inhibition of BeSO4-stimulated T-cell proliferation with anti-HLA isotype specific antibodies in HLA typed subjects support the notion that HLA-DR genes are implicated in beryllium presentation in HLA-DPGlu69-negative. A role for HLA-DRPhe47 in Be presentation is conceivable, and the data presented suggest it.
In fact, multiple analysis has indicated that the HLA-DR aminoacid variants Ser13, Tyr26, His32, Asn37, Arg74 are indirectly associated with beryllium hypersensitivity due to linkage disequilibrium with Phe47. It is worth noticing that this analysis accounts for the association of HLA-DRArg74 and HLA-DRTyr26 with beryllium sensitization found in a previous study [10]. These markers were identified for their positive and negative association of HLA-DR alleles with disease or sensitization using univariate analysis in the overall population analyzed [10], while in this re-evaluation of the same study population after 7-years of follow up including more subjects with Be-sensitization and disease we could use multiple logistic regression multivariate analysis on HLA-DPGlu69-negative subjects only. The fact that the alleles of the HLA-DRB1*03 group (alleles found associated with Be-sensitization and not disease in the previous evaluation of this study population [10]) carrying almost exclusively the HLA-DRArg74 and HLA-DRTyr26 are also carrying HLA-DRPhe47 support the notion of the linkage disequilibrium between them. Further, only a fraction of HLA-DPGlu69-negative carry only HLA-DRArg74 and -DRTyr26 (10 out of 22; 45.5%) while all except one (21 out of 22, 95.5%) carry HLA-DRPhe47 suggesting that more than an allele or set of them is a residue and the role play in the HLA-class II peptide binding pocket where he is mapping involved in the Be-presentation to T-cell determining Be-susceptibility.
Consistent with the T-cell antibody inhibition study, multiple regression analysis also indicates that the association between the HLA-DQ marker Leu26 and BH is attributable to linkage disequilibrium between HLA-DR and HLA-DQ loci [24]. It is well known that the HLA-DQLeu26 residue is expressed by all the HLA-DQB1*02 alleles and most of *03 and *06 alleles that are in linkage with the HLA-DRB1*03, 11, 12, 13 or 15 alleles which, in turn, express HLA-DRPhe47. Similar to our data, Maier and coworkers [13] obtained evidence for an association of HLA-DQB1*06, a Leu26 expressing group of alleles, with BH in HLA-DPGlu69-negative subjects. They too attributed the increased frequency of this HLA-DQ marker to linkage disequilibrium with HLA-DR and in particular to HLA-DR*13 alleles, a group of alleles expressing Phe47 [13].
The data of this study take the above observations [10,13] a step further by suggesting that the HLA-DR gene, possibly the HLA-DRPhe47 supratypic variant, ought to play a functional role in Be presentation, as this (Tyr/Phe 47) polymorphism is known to be important for peptide binding and presentation to T-cells [25,26]. Interestingly, HLA-DRPhe47 has been also implicated in susceptibility to the histopathological alike of berylliosis, sarcoidosis, in a very large case control population study [21].
How might the HLA-DRPhe47 molecule bind beryllium? Potolicchio et al. have shown that cobalt binds directly to polymorphic residue(s), likely Glu69 and/or Glu56, of the HLA-DP molecule [27], and this model may apply to beryllium interaction with HLA-DP. On the other hand, Lu et al. have recently reported that nickel interacts both with the HLA-DR backbone, with the non-polymorphic His in position 81 of the HLA-DR β-chain, and with a bound peptide [28]. This latter mechanism could also be envisioned for beryllium interaction with HLA-DRPhe47. Residue 47 of the HLA-DR β-chain is located in pocket 7, together with residues 65 and 69 of the α-chain and residues 28, 30, 61, 67, 70 of the β-chain, involving to some extent residues 71 and 74 of the β-chain which however do primarily contribute to pocket 4 [29-32]. In pocket 7, besides residue 47, only residues 28, 67 and 70 of the β chain are polymorphic. Interestingly, all of the alleles carrying the HLA-DRPhe47 in the HLA-DPGlu69-negative BH subjects, expressed always an aspartic acid residue 28 of the β-chain and either an aspartic acid or a glutamine, both residues which, together with the asparagine and tryptophan present at the non polymorphic residues 69 of the α-chain and 61 of the β-chain, could coordinate the positive charge of Be for presentation to Be-specific T-cells. The spatial relationships between the residues potentially involved may be defined precisely from the crystal structure of HLA-DR3 and HLA-DR15 [30,32] (Figure 2, panels A-B). As seen from the crystal structure, when Phe47 in pocket 7 is substituted for by Tyr47 in the HLA-DR1 and HLA-DR4 molecules, the hydrogen of the Tyr47 hydroxyl is engaged in a hydrogen bound network with the aspartic acid at position 28 and glutamine at position 70 of the β-chain [31,33] (Figure 2, panels C-D), thereby preventing their participation in coordinating the charge of the Be ion. Thus, the presence of Phe47 in pocket 7 of the HLA-DR molecule could favor Be binding. Furthermore, the HLA-DR molecules carrying Phe47 could coordinate Be together with a bound peptide with a mechanism similar to that described for nickel [28]. A number of considerations are likely to support this hypothesis. First, the distance between the different electron donor groups in the crystal structure of HLA-DR molecules carrying Phe47 lies between 6.7 and 12.6 Å [30,32] i.e., very close to the upper limits of a coordination bond with Be. Second, the two electron donor (non-polymorphic) residues α69 (Asn) and β61 (Trp) of HLA-DR pocket 7 are known to be involved in the formation of H-bonds with the backbone of the peptide antigen [29] and would therefore be unable to directly coordinate Be. Finally, the pocket 7 of HLA-DR allelic variants carrying Phe47 are capable of binding, with higher affinity than pocket 7 carrying Tyr47, aminoacid side chains with electron donor groups such as Asn, His, Met, Trp and Tyr [34].
Figure 2 Analysis of the H-bond network in the pocket 7, the peptide binding pocket where the HLA-DR residue β47 is mapping, of HLA-DR molecules carrying HLA-DRPhe47 (Panel A: HLA-DR3 and Panel B: HLA-DR15) or its counterpart Tyr47 (Panel C: HLA-DR1 and Panel D: HLA-DR4). Molecular modelling of the PDB entry crystal structures (HLA-DR3: 1A6A; HLA-DR15: 1BX2; HLA-DR1: 1AQD; HLA-DR4: 2SEB) have been evaluated with the SwissPDB viewer v3.7b2 software (free available at ). The HLA-DR α-chain backbone is reported in red colored ribbon style, while the HLA-DR β-chain backbone is reported in grey colored ribbon. Aminoacids are colored in CPK style (C: light blue; O: red; N: blue) and residue names are reported in red. H-bonds were computed with the SwissPDB viewer (H-bond detection threshold: 1.20–2.76 A when Hydrogen is present and 2.19–3.30 A when Hydrogen is absent) and are shown as green dashed lines. All the aminoacids presenting electron donor groups in the pocket 7, of HLA-DR1, -DR3, -DR4 and -DR15, putatively capable of coordinating Be are shown (residues α69, β28, β61, β70 and β71). In the HLA-DR3 crystal structure (Panel A) with the presence of Phe47 only one of the two terminal oxygens of Aspβ28 is engaged in a H-bond network with Lys71, leaving four other contacts points for co-ordinating Be (specifically residues αAsn69, βAsp28, βTrp61, βGln70). A similar pattern is present in HLA-DR15 (Panel B) where, with the presence of Phe47, no H-bonds are present leaving 5 electron donor groups available for Be coordination (specifically one electron donor group for each residue αAsn69, βTrp61, βGln70 and two electron donor groups for βAsp28). When Tyr47, the HLA-DRPhe47 counterpart, is present in HLA-DR molecules as in HLA-DR1 (panel C) and HLA-DR4 (Panel D), the H-bond network of pocket 7 results dramatically modified. Specifically, Tyr47 engages in a H-bond network with residues Asp28 and Arg71 in HLA-DR1 (Panel C) or Asp28 and Lys71 in HLA-DR4 (Panel D). As a consequence there is reduced availability of electron donor groups capable to coordinate Be.
Conclusion
In conclusion, both the HLA typing and the in vitro T-cell data in this study indicate a role for HLA-DR genes in determining susceptibility to beryllium hypersensitivity among individuals not expressing the HLA-DPGlu69 variant. The typing data point to the HLA-DRPhe47 supratypic variant as the susceptibility gene in this sub-population, and analysis of the molecule's structure suggests that HLA-DRPhe47 could bind beryllium and present it to T-cells, using a mechanism different from what used by HLA-DPGlu69. Together, HLA-DPGlu69 and HLA-DRPhe47 could account for susceptibility in almost 100% of the affected population.
Competing interests
1. Massimo Amicosante: I declare that I have NOT financial and non-financial competing interests in relation to this manuscript.
2. Floriana Berretta: I declare that I have NOT financial and non-financial competing interests in relation to this manuscript.
3. Milton Rossman: I declare that I have NOT financial and non-financial competing interests in relation to this manuscript.
4. Richard H. Butler: I declare that I have NOT financial and non-financial competing interests in relation to this manuscript.
5. Paola Rogliani: I declare that I have NOT financial and non-financial competing interests in relation to this manuscript.
6. Ella van den Berg-Loonen: I declare that I have NOT financial and non-financial competing interests in relation to this manuscript.
7. Cesare Saltini: I declare that I have NOT financial and non-financial competing interests in relation to this manuscript.
Authors' contributions
MA carried out the analysis of the HLA class II polymorphisms, participated at the functional studies of HLA class II restriction of response to beryllium, participated at the study design and drafted the manuscript. FB carried out the functional studies of HLA class II restriction of response to beryllium and participated the analysis of the HLA class II polymorphisms. MR participated in the design of the study, contributed to the data analysis review and control and contributed to draft the manuscript. RHB performed the HLA class II modeling evaluation and beryllium-binding hypothesis to HLA-DR molecules carrying Phe47. PR performed the study population re-evaluation. EvdBL performed the HLA class II high resolution typing. CS performed the study design, supervised the study population clinical follow up and analysis work and drafted the manuscript. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
The additional file (supplemented material.pdf) includes 8 tables reporting the allelic frequency for HLA-DPB1, DQB1 and DRB1, 3, 4 and 5 both in general population (tables #1–4) and in the HLA-DPGlu69 negative subjects (tables #5–8).
Click here for file
Acknowledgements
We would like to thank Chiara Dotti BS (University of Tor Vergata, Roma, IT) and Luca Richeldi and Alberto Franchi (University of Modena, Italy) for their help at the beginning of the study, Enrico Girardi (INMI, Roma, IT) for his help with statistical analysis, David Deubner (Brush Wellman, Elmore OH, USA) for his assistance with the re-evaluation of the patient population and Roberto Tosi (CNR, Rome, Italy) for the critical reading of the manuscript.
This study was supported in part by the US Department of Energy (DoE) grant DE-FG02-93ER61714 and DE-FG02-ER63416 and by a grant from Guzzini foundation. MA and FB are supported by a Guzzini foundation post-doctoral fellowship.
==== Refs
Kreiss K Miller F Newman L Ojo-Amaize EA Rossman M Saltini C Chronic beryllium disease. From the workplace to cellular immunology, molecular immunogenetics, and back Clin Immunol Immunopath 1994 71 123 129 10.1006/clin.1994.1061
Rossman MD Kern JA Elias LA Cullen MR Epstein PE Preuss OP Markham TN Daniele RP Proliferative response of bronchoalveolar lymphocytes to beryllium Ann Int Med 1988 108 687 693 3282464
Fontenot AP Newman LS Kotzin BL Chronic beryllium disease: T cell recognition of a metal presented by HLA-DP Clin Immunol 2001 100 4 14 11414740 10.1006/clim.2001.5053
Saltini C Winestock K Kirby M Pinkston P Crystal RG Maintenance of alveolitis in patients with chronic beryllium disease by beryllium-specific helper T cells N Engl J Med 1989 320 1103 1109 2469014
Saltini C Kirby M Trapnell B Tamura N Crystal RG Biased accumulation of T-lymphocytes with "memory"-type CD45 leukocyte common antigen gene expression on the epithelial surface of the human hung J Exp Med 1990 171 1123 1140 2139099 10.1084/jem.171.4.1123
Fontenot AP Canavera SJ Gharavi L Newman LS Kotzin BL Target organ localization of memory CD4(+) T cells in patients with chronic beryllium disease J Clin Invest 2002 110 1473 1482 12438445 10.1172/JCI200215846
Richeldi L Sorrentino R Saltini C HLA-DPB1 glutamate 69: a genetic marker of beryllium disease Science 1993 262 242 244 8105536
Richeldi L Kreiss K Mroz MM Zhen B Tartoni P Saltini C Interaction of genetic and exposure factors in the prevalence of berylliosis Am J Ind Med 1997 32 337 340 9258386 10.1002/(SICI)1097-0274(199710)32:4<337::AID-AJIM3>3.0.CO;2-R
Wang Z White PS Petrovic M Tatum OL Newman LS Maier LA Marrone BL Differential susceptibilities to chronic beryllium disease contributed by different Glu69 HLA-DPB1 and -DPA1 alleles J Immunol 1999 163 1647 1653 10415070
Saltini C Richeldi L Losi M Amicosante M Voorter C van den Berg-Loonen E Dweik RA Wiedemann HP Deubner DC Tinelli C Major histocompatibility locus genetic markers of beryllium sensitization and disease Eur Respir J 2001 18 677 684 11716174 10.1183/09031936.01.00106201
Wang Z Farris GM Newman LS Shou Y Maier LA Smith HN Marrone BL Beryllium sensitivity is linked to HLA-DP genotype Toxicology 2001 165 27 38 11551429 10.1016/S0300-483X(01)00410-3
Rossman MD Stubbs J Lee CW Argyris E Magira E Monos D Human leukocyte antigen Class II amino acid epitopes: susceptibility and progression markers for beryllium hypersensitivity Am J Respir Crit Care Med 2002 165 788 794 11897645
Maier LA McGrath DS Sato H Lympany P Welsh K Du Bois R Silveira L Fontenot AP Sawyer RT Wilcox E Newman LS Influence of MHC class II in susceptibility to beryllium sensitization and chronic beryllium disease J Immunol 2003 171 6910 6918 14662898
McCanlies EC Ensey JS Schuler CR Kreiss K Weston A The association between HLA-DPB1Glu69 and chronic beryllium disease and beryllium sensitization Am J Ind Med 2004 46 95 103 15273960 10.1002/ajim.20045
Lombardi G Germain C Uren J Fiorillo MT du Bois RM Jones-Williams W Saltini C Sorrentino R Lechler R HLA-DP allele-specific T cell responses to beryllium account for DP-associated susceptibility to chronic beryllium disease J Immunol 2001 166 3549 3555 11207315
Fontenot AP Torres M Marshall WH Newman LS Kotzin BL Beryllium presentation to CD4+ T cells underlies disease-susceptibility HLA-DP alleles in chronic beryllium disease Proc Natl Acad Sci U S A 2000 97 12717 12722 11050177 10.1073/pnas.220430797
Amicosante M Sanarico N Berretta F Arroyo J Lombardi G Lechler R Colizzi V Saltini C Beryllium binding to HLA-DP molecule carrying the marker of susceptibility to berylliosis glutamate beta 69 Hum Immunol 2001 62 686 693 11423174 10.1016/S0198-8859(01)00261-0
Amicosante M Berretta F Franchi A Rogliani P Dotti C Losi M Dweik R Saltini C HLA-DP-unrestricted TNF-α release in beryllium-stimulated peripheral blood mononuclear cells Eur Respir J 2002 20 1174 1178 12449171 10.1183/09031936.02.02232001
Stange AW Furman FJ Hilmas DE The beryllium lymphocyte proliferation test: relevant issues in beryllium health surveillance Am J Ind Med 2004 46 453 462 15490468 10.1002/ajim.20082
Khanolkar-Young S Kolk AH Andersen AB Bennedsen J Brennan PJ Rivoire B Kuijper S McAdam KP Abe C Batra HV Results of the third immunology of leprosy/immunology of tuberculosis antimycobacterial monoclonal antibody workshop Infect Immun 1992 60 3925 3927 1500202
Rossman MD Thompson B Frederick M Maliarik M Iannuzzi MC Rybicki BA Pandey JP Newman LS Magira E Beznik-Cizman B Monos D ACCESS group HLA-DRB1*1101: a significant factor for sarcoidosis in blacks and whites Am J Hum Genet 2003 73 720 735 14508706 10.1086/378097
Rozas J Rozas R DnaSP, DNA sequence polymorphism: an interactive program for estimating population genetics parameters from DNA sequence data Comput Appl Biosci 1995 11 621 625 8808578
Newman LS Mroz MM Balkissoon R Maier LA Beryllium sensitization progresses to chronic beryllium disease: a longitudinal study of disease risk Am J Respir Crit Care Med 2005 171 54 60 15374840 10.1164/rccm.200402-190OC
Degli-Esposti MA Leaver AL Christiansen FT Witt CS Abraham LJ Dawkins RL Ancestral haplotypes: conserved population MHC haplotypes Hum Immunol 1992 34 242 252 1464552 10.1016/0198-8859(92)90023-G
Evavold BD Sloan-Lancaster J Allen PM Tickling the TCR: selective T-cell functions stimulated by altered peptide ligands Immunol Today 1993 14 602 609 8305133 10.1016/0167-5699(93)90200-5
Spain LM Jorgensen JL Davis MM Berg LJ A peptide antigen antagonist prevents the differentiation of T cell receptor transgenic thymocytes J Immunol 1994 152 1709 1717 8120380
Potolicchio I Festucci A Hausler P Sorrentino R HLA-DP molecules bind cobalt: a possible explanation for the genetic association with hard metal disease Eur J Immunol 1999 29 2140 2147 10427976 10.1002/(SICI)1521-4141(199907)29:07<2140::AID-IMMU2140>3.3.CO;2-H
Lu L Vollmer J Moulon C Weltzien HU Marrack P Kappler J Components of the ligand for a Ni++ reactive human T cell clone J Exp Med 2003 197 567 574 12615898 10.1084/jem.20021762
Brown JH Jardetzky TS Gorga JC Stern LJ Urban RG Strominger JL Wiley DC Three-dimensional structure of the human class II histocompatibility antigen HLA-DR1 Nature 1993 364 33 39 8316295 10.1038/364033a0
Ghosh P Amaya M Mellins E Wiley DC The structure of an intermediate in class II MHC maturation: CLIP bound to HLA-DR3 Nature 1995 378 457 464 7477400 10.1038/378457a0
Dessen A Lawrence CM Cupo S Zaller DM Wiley DC X-ray crystal structure of HLA-DR4 (DRA* DRB1*0401) complexed with a peptide from human collagen II Immunity 0101 7 473 481 9354468 10.1016/S1074-7613(00)80369-6
Smith KJ Pyrdol J Gauthier L Wiley DC Wucherpfennig KW Crystal structure of HLA-DR2 (DRA* DRB1*1501) complexed with a peptide from human myelin basic protein J Exp Med 0101 188 1511 1520 9782128 10.1084/jem.188.8.1511
Fremont DH Hendrickson WA Marrack P Kappler J Structures of an MHC class II molecule with covalently bound single peptides Science 1996 272 1001 1004 8638119
Sturniolo T Bono E Ding J Raddrizzani L Tuereci O Sahin U Braxenthaler M Gallazzi F Protti MP Sinigaglia F Hammer J Generation of tissue-specific and promiscuous HLA ligand databases using DNA microarrays and virtual HLA class II matrices Nat Biotechnol 1999 17 555 561 10385319 10.1038/9858
|
16098233
|
PMC1198259
|
CC BY
|
2021-01-04 16:23:24
|
no
|
Respir Res. 2005 Aug 14; 6(1):94
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-94
|
oa_comm
|
==== Front
Theor Biol Med ModelTheoretical Biology & Medical Modelling1742-4682BioMed Central London 1742-4682-2-281607899210.1186/1742-4682-2-28ResearchModelling of oedemous limbs and venous ulcers using partial differential equations Ugail Hassan [email protected] Michael J [email protected] School of Informatics, University of Bradford, Bradford BD7 1DP, UK2 Department of Applied Mathematics, University of Leeds, Leeds LS2 9JT, UK2005 3 8 2005 2 28 28 11 5 2005 3 8 2005 Copyright © 2005 Ugail and Wilson; licensee BioMed Central Ltd.2005Ugail and Wilson; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Oedema, commonly known as tissue swelling, occurs mainly on the leg and the arm. The condition may be associated with a range of causes such as venous diseases, trauma, infection, joint disease and orthopaedic surgery. Oedema is caused by both lymphatic and chronic venous insufficiency, which leads to pooling of blood and fluid in the extremities. This results in swelling, mild redness and scaling of the skin, all of which can culminate in ulceration.
Methods
We present a method to model a wide variety of geometries of limbs affected by oedema and venous ulcers. The shape modelling is based on the PDE method where a set of boundary curves are extracted from 3D scan data and are utilised as boundary conditions to solve a PDE, which provides the geometry of an affected limb. For this work we utilise a mixture of fourth order and sixth order PDEs, the solutions of which enable us to obtain a good representative shape of the limb and associated ulcers in question.
Results
A series of examples are discussed demonstrating the capability of the method to produce good representative shapes of limbs by utilising a series of curves extracted from the scan data. In particular we show how the method could be used to model the shape of an arm and a leg with an associated ulcer.
Conclusion
We show how PDE based shape modelling techniques can be utilised to generate a variety of limb shapes and associated ulcers by means of a series of curves extracted from scan data. We also discuss how the method could be used to manipulate a generic shape of a limb and an associated wound so that the model could be fine-tuned for a particular patient.
==== Body
1 Introduction
Oedema, commonly known as tissue swelling, is associated with a range of causes such as venous disease, trauma, infection, joint disease, orthopaedic surgery and removal of the lymph nodes. Oedema and associated venous ulcers occur on mainly on the leg and the arm. It can be a painful, embarrassing and costly disorder [1,2]. It occurs widely in the general population, especially from late middle age, in diabetics and in immobile patients [3-5]. Apart from the tissue swellings the ulcers themselves can typically range in size from around 0.5 cm to 10 cm across, and are of variable depth [6,7]. Fig. 1 shows an example of an oedemous leg infected with venous ulcers.
Figure 1 An example of an oedemous leg infected with oedema venous ulcers.
An important task, during the treatment of oedema and venous ulcers, is the measurement of the amount of oedema as well as the area and volume of the ulcer wounds. This is because without an accurate and objective means of measuring changes in the size or shape of ulcers, it is difficult or impossible to evaluate the efficiency of the available therapies properly. Therefore, a prerequisite for this development is a reliable method of measuring ulcers. There exist a variety of measurement methods none of which is ideal. At present direct contacting measurements are widely used but they are not accurate, carry a risk of infection and are, to say the least, uncomfortable for the patient. For example, conventional techniques for measuring the area and volume of wounds depend on making physical contact with the wound, for example by drawing around the periphery on an acetate sheet or by making an alginate cast of the wound [6]. There is currently significant interest in developing non-invasive measurement systems using optical methods such as 'structured light' (a technique that projects stripes on to a surface and infers the shape from changes in the linearity of the reflected stripe) [8] or stereo-photogrammetry. The availability of high-resolution 3D digital cameras, increasing computing power and the development of software techniques for manipulating three-dimensional information have benefited this area. However, equipment associated with these sorts of measurement methods is not often portable and is often costly, thus making it prohibitive for routine medical use.
The aim of this paper is to show how it is possible to develop a system for measuring the shape and size of limbs and venous ulcers by means of utilising an economical mathematical model. In particular, one of the outcomes we hope to achieve from this work is a technique with potential for clinical use. For this, a small number of key measurements of limbs (with minimal possible contact with the limb and the associated ulcer), made using readily available instruments such as callipers and tape measures, can be input to a computer program. The program will then be able to reconstruct a good estimate of the limb shape and dimensions. It is believed that such a technique will provide a cheap, efficient, non-invasive instrument for measuring the degree of oedema and consequently enabling various treatment plans to be evaluated.
At present there exists a wide variety of methods that can be utilised to generate the geometry of limbs affected by oedema and venous ulcers. These include boundary based methods such as polygon based design [9], extrusions and surface of revolution [10] and polynomial patches [11]; procedural modelling such as implicit surfaces [12] and fractals [13]; and volumetric models such as constructive solid geometry [14] and subdivision [15]. Many of these techniques, especially polygon based design and polynomial patches, would be very appropriate for limb shape reconstruction, although they may not be ideally suited for the problem we address here. For example, conventional spline patches would require a large array of control points and weights in order to represent a realistic shape of a limb and associated wounds.
In this initial stage of the work we are concerned with developing efficient techniques in order to perform two important tasks. They are: the generation of smooth surfaces resembling the surface data obtained from a 3D scanner; and, once a smooth surface is obtained, manipulation of the geometry so as to obtain a good representation of the limb shape for any given patient. To do this we utilise real data from a series of surface scans provided by a medical partner namely, the Department of Medical Physics and Vascular Surgery of Bradford Teaching Hospitals National Health Services Trust (BTHNHST), UK, with whom we work closely on these problems. The department of Medical Physics at BTHNHST acquired the surface data using multiple-camera photogrammetry with a DSP400 system from 3dMD Ltd. This commercial technology has been widely used for acquiring medical images, especially in the USA, and captures data in a few milliseconds. The surface resolution (i.e. the separation of data points) is approximately 2 mm with a positional accuracy of approximately 0.2 mm. When developing our PDE based techniques for modelling human limbs, which are affected by oedema and venous ulcers, our medical partner has two aims. Firstly they require a compact and smooth surface representation of their captured data. Secondly, and rather more importantly, they require a modelling tool that would enable them to manipulate the shape of a given limb so as to provide a good representative limb shape of any given patient.
In this paper we utilise the so called PDE method [16-18] to address the problem. A positive feature of the PDE method is that it can define surfaces in terms of a small set of design variables [19], instead of many hundreds of control points. In broad terms this is because its boundary-value approach means that PDE surfaces are defined by data distributed around just their boundaries, instead of data distributed over their surface area, e.g. control points. Thus, a PDE model, when changed by altering the values of its design parameters, remains continuous; there is no need for a designer to intervene in order to close up any holes that might appear at patch boundaries. In the present context, this means that PDE surfaces can be made to adapt to changes in the shape of the limb and the associated wounds.
2 PDE Surfaces
A PDE surface is a parametric surface patch , defined as a function of two parameters u and v on a finite domain Ω (⊂) R2 by regarding the function as a mapping of a point in Ω to a point in the physical space. The shape of the surface patch is usually determined by specifying a set of boundary data at the edge of (∂)Ω. Typically the boundary data are specified in the form of and a number of its derivatives on (∂)Ω. Hence, by casting the surface generation as a boundary value problem, the surface is regarded as a solution of an elliptic PDE.
Various elliptic PDEs could be used; the ones we utilise for this work are based on the biharmonic and triharmonic equations, namely,
and
Also, periodic boundary conditions are very often considered. Assuming we are working with the above two elliptic PDEs, we require them to satisfy a set of 2N conditions, where N is 2 in the case of Equation (1) and N 3 in the case of Equation (2). The general form of these conditions can then be written as,
X(0, v) = f1(v), (3)
X(ui, v) = gi(v), i = 2 ... 2N - 1 (4)
X(1, v) = f2N(v), (5)
where f1(v) in Equation (3) and f2N(v) in Equation (5) are function conditions specified at u = 0 and u = 1 respectively. The conditions X(ui, v) = gi(v) in Equation (4) can take the form either
X(ui, v) = fi for 0 <ui < 1, i = 2 ... 2N - 1, (6)
or
In simpler terms the above conditions imply that for a PDE surface patch of order 2N, we can specify two function conditions, as given in Equations (3) and (5), that should be satisfied at the edges (at u = 0 and u = 1) of the surface patch, and a number of function or derivative conditions, as given in Equation (4), amounting to 2N – 2 conditions that the PDE should also satisfy.
2.1 Solution of the PDEs
There exist many methods for solving Equations (1) and (2) ranging from analytic solutions to sophisticated numerical methods. The problems we address in this paper involve modelling of human limbs, which are essentially closed and cylindrical, and therefore the broad range of shapes encountered can be incorporated by solving the chosen PDEs with periodic conditions. Note here periodic conditions imply that for the v parameter the condition, , is satisfied. Thus, for the work described here, we restrict ourselves to periodic conditions and obtain a closed form analytic solution of Equations (1) and (2).
Choosing the parametric region to be 0 ≤ u ≤ 1 and 0 ≤ v ≤ 2π, and assuming that the conditions given in Equations (3), (4) and (5) are periodic functions, we can use the method of separation of variables and spectral approximation [20] to write down the analytic solution of Equations (1) and (2) as,
where is a polynomial function and and are exponential functions. The specific forms of and for the case of Equations (1) can be found in [17] and for the case of Equations (2) can be found in [21].
The main point to bear in mind regarding the above solution method is that it enables one to represent a set of general periodic conditions in terms of a finite M Fourier series, where M is typically taken to be ≤ 10, whilst the term , which acts as a correction term, enables the conditions to be satisfied exactly. Detailed discussions of this solution method can be found in [20].
2.2 Methods of Generating PDE Surfaces
In this section we discuss a series of examples, showing the various methods by which PDE surfaces can be generated where the PDEs are chosen to be Equations (1) and (2) and the conditions are taken in the format described in Equations (3), (4) and (5).
As a first example we show how a fourth order PDE surface is generated where all the conditions are taken to be function conditions. Fig. 2(b) shows the shape of a surface generated by the fourth order PDE where the conditions are specified in terms of the curves shown in Fig. 2(a). In particular, the conditions are such that: and . Since we are taking four function conditions to solve the fourth order PDE, all the curves in this case lie on the resulting surface. Thus, in this particular case the resulting PDE surface is a smooth interpolation between the given set of functional conditions.
Figure 2 The shape of a surface generated by the fourth order PDE where the conditions are all taken to be function conditions (a) The conditions defined in the form of curves in 3-space. (b) The resulting surface shape.
The next example shows how a fourth order PDE surface is generated when the conditions are taken to be a mixture of function conditions and derivative conditions. Fig. 4(b) shows the shape of a surface generated by the fourth order PDE where two function boundary conditions and two derivative boundary conditions are specified in terms of the curves shown in Fig. 4(a). In particular, the boundary conditions are chosen such that: and , where s is a scalar. In this case the surface patch generated as a solution to the fourth order PDE contains the boundary curves c1 and c4 whilst it does not necessarily contain the curves c2 and c3. A typical scenario where a surface of this nature is required would be a blend design where the derivative boundary curves can be adjusted to produce a smooth blend surface that bridges between two primary surfaces.
Figure 4 The shape of a surface generated by the fourth order PDE where the boundary conditions are taken to be both positions and derivatives (a) The boundary conditions defined in the form of curves in 3-space. (b) The resulting surface shape.
Fig. 3(b) shows the shape of a surface generated by the sixth order PDE where the conditions are all taken to be positions specified in terms of the curves shown in Fig. 3(a). In particular, the boundary conditions are such that and . As in the first example of the fourth order case, since we are taking six function conditions to solve the sixth order PDE, all the curves in this case lie on the resulting surface. Thus, the resulting PDE surface is a smooth interpolation between the six prescribed curves.
Figure 3 The shape of a surface generated by the sixth order PDE where the conditions are all taken to be positions (a) The conditions defined in the form of curves in 3-space. (b) The resulting surface shape.
As a final example we show how a sixth order PDE surface is generated where the boundary conditions are taken to be a mixture of function boundary conditions and derivative conditions (both first and second order). Fig. 5(b) shows the shape of a surface generated by the sixth order PDE where two function boundary conditions, two first order derivative boundary conditions and two second order derivative boundary conditions are specified in terms of the curves shown in Fig. 5(a). In particular, the boundary conditions are chosen such that and . where s and t are scalars. As in the example of fourth order case shown in Fig. 4 the surface generated in this case contains the curves c1 and c6 whilst it does not necessarily contain the rest of the curves. Again this type of surface shape can be utilised in blend design where higher order continuity is desired in producing a smooth blend surface that bridges between two primary surfaces. As one can see from the format of these derivative condition definitions, the derivative conditions are all defined using simple finite difference schemes. The curves defining the derivative conditions provide an intuitive shape manipulation tool in that the shape of the surface closely follows the shape of the boundary conditions.
Figure 5 The shape of a surface generated by the sixth order PDE where the boundary conditions are taken to be both positions and derivatives (a) The boundary conditions defined in the form of curves in 3-space. (b) The resulting surface shape.
The above examples demonstrate how PDE surfaces of order four and six can be utilised to generate surface shapes, which are applicable to a wide variety of design scenarios. Thus, the basic idea here is to generate a series of curves (both function and derivative) that can be utilised to define the boundary conditions for the chosen PDE. As seen in the examples, the resulting surface shape can always be intuitively predicted from the shapes of the chosen curves.
3 Modelling of Limbs and Ulcers
In this section we discuss the shape modelling of human limbs affected by oedema and venous ulcers. In what follows, we discuss two examples of shape modelling of human limbs namely modelling of an arm shape and modelling of a leg shape with an ulcer. We utilise a mixture of PDEs of order four and six in order to model the surface shapes in question. In order to generate a representative smooth PDE surface shape, we extract a series of curves along the profile of the geometric model.
Fig. 6 shows a typical surface scan data set provided by the medical partner where in this particular case the data set corresponds to an arm shape. Note that scan data are only available for half the surface. In order to generate a representative PDE surface shape, we extract a series of curves along the profile of the geometric model. To do this, first we import the geometric model into an interactive graphical environment through which we can examine and interact with the model. The geometric definitions of the scan data are provided in .obj file format where the 3D polygonal data with connectivity information are readily available. This enables us to display the model as well as compute the normal curvature distribution across the surface. A series of regions on the scan data model are then manually identified based on changes in the surface curvature. These regions are then utilised to determine the number of surface patches required to produce a good representative model of the limb in question. In determining the number of PDE surface patches required the aim is to reduce the number of patches that need to be utilised to produce a good representative geometric model with given accuracy. Once the number of surface patches required is decided the appropriate number of curves for each surface patch is extracted from the scan geometry data. To do this we create a series of free-form cubic spline curves within the interactive environment. The spline curves are then projected on to the scan geometry at the positions where the PDE curves are to be extracted. Note that the surface data obtained in this case do not naturally give us curves that are periodic. Thus, in this case, for each curve extracted, a series of fictitious points is added to each curve in order to make the curve periodic. The surfaces are then generated using the analytic solution described previously, where the surface is generated for the region 0 ≤ v ≤ π which forms the portion of the curves extracted from the scanned data.
Figure 6 Scanned surface data of an arm.
3.1 Example 1: Modelling of an Arm Shape
As a first example, we discuss the modelling of the shape of a human arm. Fig. 6 shows the scan surface data corresponding to an arm shape provided by the medical partner. Fig. 7(a) shows a series of curves extracted from the original scan surface data. Fig. 7(b) shows the arm shape generated using PDE surfaces. In particular the shape is generated as a combination of two fourth order patches and a single sixth order surface patch. i.e. the curves c5, c6, c7 and c8 and c8, c9, c10 and c11 form boundary conditions for two fourth order surface patches with the common boundary at c8 whereby all the conditions are taken to be position conditions. The curves c1, c2, c3 and c5 form a sixth order surface patch where c1, c2, c4, c5 are taken to be four position conditions and the differences between c2, c3 and c5, c4 are taken to be two first order derivative boundary conditions. The value of the parameter s is taken to be 0.34.
Figure 7 The arm shape generated using PDE surfaces by means of utilising curves extracted from scanned data (a) The extracted curves. (b) The resulting surface shape generated using two fourth order patches and a sixth order patch.
3.2 Example 2: Modelling of a Leg Shape with a Venous Ulcer
As a second example we discuss the modelling of the shape of a leg infected with a venous ulcer. Fig. 8 shows the scan surface data, corresponding to the infected leg with a venous ulcer. As in the previous example, in order to generate a representative smooth surface shape, we first extract a series of curves along the leg and the associated wound. Fig. 9(a) shows a series of curves extracted from the original scan data.
Figure 8 Scanned surface data of a leg infected with an ulcer.
In order to create a smooth shape that closely resembles the geometry of the leg, we utilise two sixth order patches to generate the main portion of the leg. Thus, the curves c1, c2, c3, c4, c5 and c6 form the position condition for a sixth order surface patch where the surface patch passes through these curves. The other surface patch is generated using the curves c6, c7, c8, c9 and c10 where the curve c6 is common to both surface patches. Moreover, for the later surface patch the curves c6, c7 and c9, c10 form four position boundary conditions and the differences between the curves c7, c8 and c9, c8 form two first order derivative boundary condition thus ensuring a smooth geometry transition between the foot and the leg. The parameter s is taken to be 0.12.
To generate the wound shape on the leg, we define a curve on the PDE leg surface that closely resembles the edge of the wound. This curve, marked as c11 as shown in Fig. 9(a), is generated using the (u, v) parameter space of the corresponding the PDE surface. Next the surface portion corresponding to the interior of the curve c11 is trimmed out. This trimming process is again carried out using the (u, v) parameter space as described in [17]. Fig. 10(a) shows the main leg surface with the trim.
Figure 9 The main leg shape generated using PDE surfaces by means of utilising curves extracted from scan data (a) The extracted curves (including the wound). (b) The resulting surface shape corresponding to the main shape of the leg, generated using two sixth order patches and a sixth order patches.
Figure 10 Leg and ulcer geometry. (a) A portion of the surface is trimmed out using a curve resembling the edge of the ulcer wound (b) The complete leg geometry with the ulcer wound.
Once the appropriate trimming is carried out, a separate fourth order patch resembling the shape of the wound is generated where the curve c11 which lies on the main leg surface is utilised as one of the four position boundary conditions. Fig. 10(b) shows the complete leg shape along with the ulcer wound.
Both the examples discussed above show how PDE surfaces of low order (i.e. order 4 and 6 in this case) can be utilised to generate good representative shapes using little information from the scan data. One could argue that a single PDE surface of higher order can be equally well suited to generating a single surface patch through a given number of curves. However, from the min-max principle for elliptic PDEs it is well known that PDEs of higher order (i.e. orders above 6) are difficult to control. Choosing lower order PDEs to generate the surface therefore makes sense. It is also noteworthy that the parameters s and t and the difference between the corresponding position and derivative curves enable both the size and the direction of the derivative boundary conditions at the edge of a given surface patch to be controlled. The derivative boundary conditions are used to control the smoothness of the blend between two surface patches. Such a tool cannot be deployed to reduce the number of curves used and hence the number of surface patches utilised to model the complete limb.
4 Conclusion
This paper describes how the PDE method can be utilised to model a wide variety of geometries of limbs affected by oedema and venous ulcers. The shape modelling is based on solving a PDE subject to a set of curves extracted from 3D scan data providing the shape of the affected limbs. For this work we utilise a mixture of fourth order and sixth order PDEs, depending on the accuracy and continuity requirements for obtaining a good representative shape of the limb and associated ulcers in question.
In this work we are concerned with developing efficient techniques in order to undertake two important tasks. They are: the generation of smooth surfaces closely resembling the surface data obtained from a 3D scanner; and once a smooth surface is obtained, manipulation of the geometry so as to provide a good representative limb geometry shape for any given patient. Thus, the prime aim of the technique we discuss here is to generate a good representative shape of the limb quickly from the scanned data and to be able to manipulate that shape efficiently. It is noteworthy that the process of PDE geometry generation from the scan data is currently carried out manually. We are currently working on developing a methodology for automating this process. We have shown examples that clearly demonstrate the ability of PDE shape modelling techniques to generate a variety of limb shapes and associated venous ulcers. The geometry models themselves are flexible in terms of their manipulation capabilities, i.e. the manipulation of geometry can be carried out via the manipulation of curves defining the surface.
Our future direction in this work is to define a shape parameterisation tool for limbs where a set of shape parameters can be associated with the curves. Such shape parameterisation can then be utilised to fine tune a given generic limb model to suit to a handful of data measured from a given patient's limb. This will enable one to develop efficient non-invasive techniques for measuring various properties (such as surface area and volume) of oedema and venous ulcers.
Acknowledgements
The authors wish to thank Dr. R.G. Cameron and Dr. W. Gardner of the Department of Medical Physics and Vascular Surgery at Bradford Teaching Hospitals National Health Services Trust of UK for fruitful discussions and supplying 3D scan data of limbs.
==== Refs
Bosanquet N Costs of Venous Ulcers: From Maintenance Therapy to Investment Programmes Phlebology 1992 7 44 46
Doherty D Ross F Yeo L Uttley J Leg Ulcer Management in an Integrated Service The South Thames Evidence Based Practice (STEP) Project Report (6) 2000 Kingston University
Prasad A Ali-Khan A Mortimer PS Leg Ulcers and Oedema: A Study Exploring the Prevalence, Aetiology and Possible Significance of Oedema in Leg Ulcers Phlebology 1990 5 181 187
Fletcher A Harding KG, Leaper DL, Turner TD Common Problems of Wound Management in the Elderly First European Conference on the Advances in wound management 1992 London: Macmillan 25 29
Lymphoedema Framework, Towards a National Framework for Lymphoedema Management
Plassmann P Melhuish JM Harding KG Methods of Measuring Wound Size: A Comparative Study WOUNDS: A Compendium of Clinical Research and Practice 1994 6 54 61
Elder D K G Venous Disease: How to Heal and Prevent Chronic Leg Ulcers Geriatrics 1995 50 30 36 7635325
Krouskop TA Baker R Wilson MS A Noncontact Wound Measurement System Journal of Rehabilitation Research and Development 2002 39 337 346 12173754
Hoschek J Lasser D Computer Aided Geometric Design 1993 A K Peters
Mortenson ME Geometric Modelling 1985 New York: Wiley-Interscience
Farin G Curves and Surfaces for Computer Aided Geometric Design, A Practical Guide 2001 Morgan-Kaufmann
Bajaj C Blinn J Bloomenthal J Cani-Gascuel M Rock-wood A Wyvill B Wyvill G Introduction to Implicit Surfaces 1997 Morgan-Kaufmann
Szeliski R Terzopoulos D From Splines to Fractals Proceedings of the 16th annual conference on Computer Graphics and Interactive Techniques 1989 ACM New York 51 60
Rappoport A Spitz S Interactive Boolean Operations for Conceptual Design of 3-d Solids Proceedings of the 24th Annual Conference on Computer Graphics and Interactive Techniques 1997 ACM New York 269 278
DeRose T Kass M Truong T Subdivision Surfaces in Character Animation Proceedings of SIGGRAPH 98 1998 Addison-Wesley 85 94
Bloor MIG Wilson MJ Generating Blend Surfaces Using Partial Differential Equations Computer-Aided Design 1989 21 165 171 10.1016/0010-4485(89)90071-7
Ugail H Bloor MIG Wilson MJ Techniques for Interactive Design Using the PDE Method ACM Transactions on Graphics 1999 18 195 212 10.1145/318009.318078
Du H Qin H Direct Manipulation and Interactive Sculpting of PDE Surfaces Computer Graphics Forum (Proceedings of Eurographics 2000) 2000 19 261 270 10.1111/1467-8659.00418
Ugail H Wilson MJ Efficient Shape Parameterisa-tion for Automatic Design Optimisation using a Partial Differential Equation Formulation Computers and Structures 2003 81 2601 2609 10.1016/S0045-7949(03)00321-3
Bloor MIG Wilson MJ Spectral Approximations to PDE Surfaces Computer-Aided Design 1996 28 145 152 10.1016/0010-4485(95)00060-7
Kubeisa S Ugail H Wilson M Interactive Design Using Higher Order PDE's The Visual Computer 2004 20 682 693 10.1007/s00371-004-0261-3
|
16078992
|
PMC1198260
|
CC BY
|
2021-01-04 16:39:25
|
no
|
Theor Biol Med Model. 2005 Aug 3; 2:28
|
utf-8
|
Theor Biol Med Model
| 2,005 |
10.1186/1742-4682-2-28
|
oa_comm
|
==== Front
J Transl MedJournal of Translational Medicine1479-5876BioMed Central London 1479-5876-3-301604865110.1186/1479-5876-3-30ResearchArtificial neural networks allow the use of simultaneous measurements of Alzheimer Disease markers for early detection of the disease Di Luca Monica [email protected] Enzo [email protected] Barbara [email protected] Martina [email protected] Elena [email protected] Francesca [email protected] Fabrizio [email protected] Marco [email protected] Alessandro [email protected] Massimo [email protected] Centre of Excellence for Neurodegenerative Disorders and Department of Pharmacological Sciences, University of Milan, Italy2 Medical Department, Bracco Spa, Milan, Italy3 Department of Neurological Sciences, University of Brescia, Italy4 Centro Ricerche Semeion, Rome, Italy2005 27 7 2005 3 30 30 12 5 2005 27 7 2005 Copyright © 2005 Di Luca et al; licensee BioMed Central Ltd.2005Di Luca et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Previous studies have shown that in platelets of mild Alzheimer Disease (AD) patients there are alterations of specific APP forms, paralleled by alteration in expression level of both ADAM 10 and BACE when compared to control subjects. Due to the poor linear relation among each key-element of beta-amyloid cascade and the target diagnosis, the use of systems able to afford non linear tasks, like artificial neural networks (ANNs), should allow a better discriminating capacity in comparison with classical statistics.
Objective
To evaluate the accuracy of ANNs in AD diagnosis.
Methods
37 mild-AD patients and 25 control subjects were enrolled, and APP, ADM10 and BACE measures were performed. Fifteen different models of feed-forward and complex-recurrent ANNs (provided by Semeion Research Centre), based on different learning laws (back propagation, sine-net, bi-modal) were compared with the linear discriminant analysis (LDA).
Results
The best ANN model correctly identified mild AD patients in the 94% of cases and the control subjects in the 92%. The corresponding diagnostic performance obtained with LDA was 90% and 73%.
Conclusion
This preliminary study suggests that the processing of biochemical tests related to beta-amyloid cascade with ANNs allows a very good discrimination of AD in early stages, higher than that obtainable with classical statistics methods.
==== Body
Introduction
Neurodegenerative diseases as Alzheimer Disease (AD) are posing a tremendous impact on our society. There is no remission in the progression, and pharmacological interventions available at present require an early and accurate identification of the disease [1].
Up to now, the diagnosis of probable and possible AD is based on neuropsychological evaluation by multidimensional assessments. Differential diagnosis between AD and other types of dementia is often based on exclusion [2].
Thus, the possibility to use a biomarker supporting clinical diagnosis would be of great relevance in the early detection of the disease.
For early AD diagnosis, an ideal diagnostic tool must be sensitive to earlier cognitive and biological changes, but is should be able to differentiate among AD, normal aging and other forms of dementia or pseudo dementia. It should be reliable, readily applicable and simple [3]. In the last years, we have identified in easily accessible circulating cells, i.e. platelets, a combination of biological measurements that possess all characteristic to be considered as highly accurate biomarkers for AD [4-6]. In platelets it is possible to measure the levels of three molecular identities key-elements in the amyloid-cascade, namely Amyloid Precursor Protein (APP) forms as well as beta-secretase (beta-site- APP cleaving enzyme, BACE1) enzyme, responsible for amyloidogenic pathway, and alpha-secretase (ADAM10) responsible for non-amyloidogenic metabolism [7,8]. Further, we have demonstrated a concomitant and congruent modification of these biochemical parameters in platelets of AD patients when compared to control subjects [6]. Indeed, in platelets of mild AD patients we were able to show an alteration of specific APP forms, paralleled by a decreased expression level and activity of ADAM 10 as well as an increased BACE activity, when compared to control subjects [8].
The simultaneous measurement of these biochemical parameters can be considered as a useful "combining strategy" to enhance the accuracy of the biological testing. However, this approach has intrinsic constrains, related to the statistical analysis used since classical statistics approaches suffer from underlying non linearity among variables.
Artificial Neural Networks (ANNs) are adaptive models for the analysis of data which are inspired by the functioning processes of the human brain. They are systems which are able to modify their internal structure in relation to a function objective. They are particularly suited for solving problems of the non linear type, being able to reconstruct the approximate rules that put a certain set of data – which describes the problem being considered – with a set of data which provides the solution [9].
Thus, the aim of this study was to assess the efficacy of neural network in correctly classifying control subjects and mild AD patients only on the basis of peripheral beta-amyloid cascade biomarkers.
Methods
Subjects
The study was carried out on 37 probable mild AD patients, and 25 healthy age-matched controls (CON), in accordance with local clinical research regulations and an informed consent was obtained. Each subject underwent a clinical and a standardized neuropsychological assessment for evaluation of cognitive functions, activities of daily living and behavioral and psychological disturbances.
Probable AD diagnosis was based on National Institute of Neurological and Communicative Disorders and Stroke-Alzheimer's Disease and Related Disorders Association criteria (NINCDS-ADRDA) [10]. Severity of dementia was rated according to Clinical Dementia Rating (CDR) scale, and only patients with CDR≤1 entered the study.
The following exclusion criteria were applied: a) major depressive disorder, bipolar disorder, schizophrenia, substance use disorder, or mental retardation according to DSM-IV criteria; b) cerebro-vascular disorder, hydrocephalus, and intra-cranial mass, documented by CT or MRI within the last 12 months; c) abnormalities in serum folate and vitamin B12, syphilis serology, or thyroid hormones' levels; d) a history of traumatic brain injury or other neurological disease; e) significant medical problems (e.g. diabetes or hypertension; cancer; hepatic, renal, cardiac or pulmonary disorders).
Further, subjects on psychotropic agents, nootropic drugs, cholinergic or anticholinergic agents, antiplatelets agents, anticoagulants, steroids, and serotoninergic drugs, were excluded unless they entered a wash-out phase lasting at least 14 days before blood collection.
Platelets collection and Western blot analysis
Blood drawing occurred while fasting at 9.00–10.00 AM. A blood sample (27 ml) was taken, releasing the tourniquet, using a 19-gauge needle, and processed as previously described (6). Platelets were processed for Western blot analysis (WB) with monoclonal antibody (mAb) 22C11 dilution 1:2000 (Chemicon, Tamecula) raised against the N-terminal domain of APP, therefore recognizing the three APP forms with apparent molecular weight of 130, 110, and 106 kDa. The results were expressed as the ratio (APP ratio) of the optical density of the upper (130 kDa) to the lower (106–110 kDa) immunoreactive bands. WB was also performed with polyclonal antibody (pAb) to ADAM10 dilution 1:1000 (Proscience Inc., Poway, CA USA); pAb to BACE 1:500 (Affinity Bioreagents Inc., Golden, CO USA), clone AC40 for β-ACTIN 1:3000 (Sigma-Aldrich Steinheim, Germany) were used. Immunostaining β-ACTIN, a constitutive protein, was used as internal standard. The OD between ADAM 10 and β-ACTIN was measured. Similarly, BACE analysis was performed, and the OD of the two forms (BACE 36 kDa/BACE 57 kDa) was evaluated.
Quantitative analysis of WB was performed by means of computer assisted imaging (Quantity-OneR System; Biorad, CA, USA).
Artificial Neural Networks
In this study, we applied supervised ANNs, networks in which the result of the processing (the output desired) is already defined. Supervised ANNs calculate an error function that measures the distance between the desired fixed output (target) and their own output, and adjust the connection strengths during the training process to minimize the result of the error function. The learning constraint of the supervised ANNs is having their own output coincide with the determined target. The general form of these ANNs is: y = f(x,w*), where w* constitutes the set of parameters which best approximate the function. The ANNs used in the study are characterized by the law of learning and topology. The laws of learning identify equations which translate the ANNs inputs into outputs, and rules by which the weights are modified to minimize the error or the internal energy of the ANNs.
The topology identifies the structure of the nodes of ANNs connections and the signal's flow within it. ANNs can be further distinguished by two broad categories: feed-forward ANNs (FF), in which the signal proceeds from the input to the output of the ANNs, crossing all of the nodes only once; recurring ANNs, in which the signal is subject to specific feedback, determined beforehand, or tied to the occurrence of particular conditions.
The experiments carried out anticipate the use of both ANNs and Artificial Organisms, i.e. complex combinations of more networks. Supervised software (Semeion ©), which allows the combination of each law of learning with each type of topology, was used for all the testing.
Four models of ANNs and one statistical model used in the study, as follows.
1) The Back Propagation standard (BP-FF) is defined by different interconnected layers of nodes characterized by a non linear function, generally of the sigmoidal type [11].
2) The Sine Net (SN) (Semeion ©) is characterized by the organization of nodes in layers [12]. The fundamental difference with respect to the better known BP-FF is due to the characteristics of the function realized by each single node. In fact, the node activation function, which has similar characteristics to those of the BP-FF, operates on an input that is made up of a sum of sines, each characterized by its own frequency (the weight of the link/connection). This modification on the base structure of the node has profound consequences on the behaviors of the SN both for the function's calculated properties and for the learning process modalities.
3) The Bi-Modal networks (BM) (Semeion ©) are multi-layered networks like Sn and Bp, but differ from these because of the equations which characterize them [13]. Each hidden node is realized by 2 sub-nodes, each equipped with its own input connections: the first sub-node operates according to the descending gradient technique, the second through vectorial quantification. The outputs of the two nodes are then composed in a single output value. These networks therefore originate from the hybridization of the two most relevant learning mechanisms in the sphere of artificial intelligence: the descending gradient and the vectorial quantification. The Bi-Modal demonstrates an excellent convergence capacity on complex problems; furthermore it does not suffer from problems on local minimums shown by other algorithms, which use the principal of the descending, gradient exclusively.
4) The Soft Max Discriminant Analysis (SMDA) is an ANNs model composed by a single layer of units, without the layer of hidden units, where the Soft-Max function is applied to the output layer.
Statistical models
Results obtained with the neural networks cited above have been compared with a model of linear statistic, i.e. the Linear Discriminant Analysis (LDA) (Software SPSS®).
The validation protocol is a fundamental procedure to verify the models' ability to generalize the results reached in the testing phase. Among the different protocols reported in literature, the selected model is the protocol with the greatest generalization ability on data unknown to the model itself. The dataset was randomly subdivided into two sub-samples: the first called Training Set, and the second, called Testing Set; a fixed ANN (and/or Organism) which is trained on the Training Set was chosen. In this phase, the ANN learns to associate the input variables with those that are indicated as targets; the weight matrix produced by the ANNs at the end of the training phase was saved, and freezed with all parameters used for the training; finally, each case belonging to the Testing Set was shown to the ANN, in order to allow the ANN to express an evaluation based on the training just performed. This procedure takes place for each input vector but every result (output vector) is not communicated to the ANN; in this way, the ANN is evaluated only in reference to the generalization ability that it has acquired during the Training phase.
Then, a new ANN with identical architecture to the previous one was constructed and the procedure from point 1 was repeated.
This general training plan has been further developed to increase the level of reliability of the generalization of the processing models. The experiments have been done using a random criterion of distribution of the samples. We have employed the 5 × 2 cross-validation protocol, which produces 10 elaborations for every sample. It consists in dividing the sample five times in two sub samples, containing each similar distribution of cases and controls [14].
Results
Descriptive Analysis
Control subjects and mild AD patients were comparable as far as age and gender distribution (see Table 1).
Table 1 Demographic, clinical and biological variables of the subjects.
Variables CONTROLS Mild AD p
Number 25 37 -
Age, years 66.5 ± 3.8 67.3 ± 6.8 n.s.
Gender, M/F 11/14 17/20 n.s.
MMSE score 29.6 ± 1.0 24.9 ± 4.4 .0001
CDR score - 0.7 ± 0.28 .0001
APPr 0.61 ± 0.22 0.44 ± 0.18 .003
BACE 1.52 ± 1.27 0.84 ± 0.73 .008
ADAM 10 0.79 ± 0.49 0.43 ± 0.23 .0004
M: male; F: female; MMSE: Mini-Mental State Examination; CDR: Clinical Dementia Rating; APPr: Amyloid Precursor Protein ratio.
Figure 1 reports a representative Western Blot of APP forms, ADAM10, and BACE levels in 2 control subjects and in 2 mild AD patients, the latter showing an alteration of these three measures. As shown, the optical density of the upper (130 kDa) APP form was reduced in AD patients compared to controls, being APP ratio decreased in the former group (panel A). Moreover, a reduction in the optical density ratio between ADAM10 and internal standard control (actin) as well as between the 36 kDa and the 57 kDa BACE forms was demonstrated in AD (see panel B,C). A significant statistical difference was present in APP ratio, BACE and ADM10 values between the two groups (see Table 1).
Figure 1 Representative Western Blot analysis of APP forms (panel A), ADAM10 (panel B), and BACE (panel C) in AD patients and in healthy controls.
The statistical significance of the differences reached an higher value for ADAM 10 in comparison with BACE and APP respectively, essentially due to a minor coefficient of variation index and a higher correlation index value (R2 = 0.012) with the target variable (diagnostic class).
Despite these average differences there was a certain degree of overlap of beta amyloid markers values across the two study groups, which made difficult the precise distinction among individual subjects belonging to one of the two classes.
ANNs Analysis
In this study two kinds of protocols with neural networks models have been carried out in order to distinguish individual control subjects from mild AD patients only on the basis of APP, BACE1 and ADAM 10 values. In the first protocol, which included 170 analyses, APP, BACE and ADAM10 constituted the input variables. In the second protocol, which included 150 analyses, the three possible dual combinations of parameters constituted the input variables (APP – BACE; APP – ADAM10; BACE – ADAM)
Table 2 summarizes the average results related to 10 independent analyses with each different ANN model using three input variables and 5 × 2 cross-validation protocol.
The best result was obtained with Self Recurrent Network Dynamic Sine Net (SelfDASn) model, which reached an overall arithmetic mean accuracy rate of 93.08%. The corresponding result obtained with Linear Discriminant Analysis (LDA) was equal to 81.6%.
In Table 3 the average results related to 10 independent analyses with each different ANN model using two input variables and 5 × 2 cross-validation protocol are reported. In these analyses, Feed-Forward ANN models based on three learning laws (Back-Propagation, SineNet e BiModal) have been employed in order to define more precisely the contribution of each marker in discriminating AD.
Table 2 Three input variables analyses and 5 × 2 cross-validation protocol (average results obtained in 10 analyses with each model). LDA: Linear Discriminant Analysis, No: Number; SD: Standard Deviation.
MODELS No. Senitivity Specificity Mean Accuracy No. Errors SD Mean Accuracy p
SelfDASn 10 94,03% 92,12% 93,08% 2,1 2,98 0,000036
TasmDASn 10 95,12% 90,51% 92,82% 2,1 3,26 0,000016
TasmSASn 10 92,51% 91,99% 92,25% 2,4 2,67 0,000022
SelfSASn 10 94,50% 89,87% 92,19% 2,3 2,58 0,000020
TasmDABm 10 93,54% 88,91% 91,23% 2,6 2,84 0,000007
TasmDABp 10 93,51% 88,91% 91,21% 2,6 3,20 0,000001
SelfSABm 10 93,57% 88,14% 90,86% 2,7 3,76 0,000144
SelfDABp 10 92,43% 88,14% 90,29% 2,9 3,35 0,000031
TasmSABp 10 90,79% 88,98% 89,88% 3,1 3,14 0,000302
SelfDABm 10 91,34% 88,21% 89,78% 3,1 4,02 0,000398
TasmSABm 10 91,90% 87,37% 89,64% 3,1 4,07 0,000504
SelfSABp 10 90,29% 88,91% 89,60% 3,2 2,96 0,000311
FF_Sn 10 91,90% 84,17% 88,03% 3,5 4,11 0,000734
FF_Bm 10 88,63% 86,54% 87,58% 3,8 3,61 0,000860
FF_Bp 10 88,63% 85,77% 87,20% 3,9 3,72 0,001092
SMDA 10 89,71% 78,65% 84,18% 4,6 6,31 0,021734
LDA 10 90,26% 72,95% 81,61% 5,2 5,47 -
Table 3 Two variables analyses and 5 × 2 cross-validation protocol (average results obtained in 10 analyses with each model).
INPUT MODELS No. Sensitivity Specificity Mean Accuracy No. Errors SD Mean Accuracy
APP ADAM FF_Sn 10 92,98% 76,99% 84,99% 4,2 4,23
FF_Bm 10 93,51% 73,08% 83,29% 4,6 6,06
FF_Bp 10 89,12% 74,74% 81,93% 5,2 5,27
SMDA 10 87,98% 69,17% 78,58% 6,1 6,12
LDA 10 82,52% 63,46% 72,99% 7,8 8,74
APP BACE FF_Sn 10 84,33% 80,06% 82,20% 5,4 3,53
FF_Bm 10 81,69% 79,94% 80,82% 5,9 4,24
FF_Bp 10 83,27% 79,17% 81,22% 5,7 3,92
SMDA 10 82,81% 76,60% 79,70% 6,1 4,21
LDA 10 79,53% 64,68% 72,11% 8,2 4,54
ADAM BACE FF_Sn 10 86,52% 80,19% 83,36% 5 5,91
FF_Bm 10 86,02% 76,92% 81,47% 5,5 5,60
FF_Bp 10 84,88% 79,42% 82,15% 5,4 6,61
SMDA 10 83,86% 74,61% 79,24% 6,2 7,05
LDA 10 85,97% 63,53% 74,75% 7,2 6,76
LDA: Linear Discriminant Analysis, No: Number; SD: Standard Deviation.
The comparison with LDA was performed as well, being the accuracy of ANN higher. The differential performance rate between ANNs and LDA was the following 12% for APP+ADAM (FF-Sn 84.99% – LDA 72.99%), 10.09% for APP+BACE (FF-Sn 82.20% – LDA 72.11%), and 8.61% for ADAM+BACE (FF-Sn 83.36% – LDA 74.75%).
In the three variables protocol, the corresponding differential value obtained with the best feed forward ANNs was equal to 6.42% (FF-Sn 88.03% – LDA 81.61%)
Input relevance analysis
In addition to evaluate the overall performance rate in discriminating between target classes, the AAN models allow the definition of the net relevance of each input in the occurring model. This evaluation has been performed using the seven best performers ANN models (overall accuracy higher than 97% in blind testing). As reported in Table 4, the input relevance value referred to each independent marker was reported. APP ratio resulted to be the most relevant variable in one model, thus reaching the 41%, while ADAM10 and BACE showed the highest input relevance three times each.
Table 4 The input relevance value referred to each independent marker (in bold).
Input relevance values
ANN model Sensitivity Specificity Mean Accuracy No. errors APP BACE ADAM
SelfDASn 94,74% 100,00% 97,37% 1/32 0,4155 0,3001 0,2845
TasmDASn 94,74% 100,00% 97,37% 1/32 0,2528 0,2445 0,5026
SelfDABm 94,44% 100,00% 97,22% 1/30 0,2802 0,4074 0,3124
SelfDABp 94,44% 100,00% 97,22% 1/30 0,2084 0,4285 0,3631
SelfDASn 94,44% 100,00% 97,22% 1/30 0,3653 0,2598 0,3749
SelfSABm 94,44% 100,00% 97,22% 1/30 0,2769 0,4273 0,2958
SelfSASn 94,44% 100,00% 97,22% 1/30 0,3577 0,2486 0,3937
Discussion
The main finding of this study is that the application of artificial neural networks on a set of biomarkers for AD, designed on the amyloid cascade, improved by more than 10% the accuracy of the early diagnosis of the disease.
The potential medical application of diagnostic biomarkers has generated much excitement within the biomedical community for early disease diagnosis in AD. Several biomarkers have been proposed in the last years as promising candidates to assess AD susceptibility among individual at risk for the disease.
In recent years, we have assisted to the transition from a single biomarker to a multiple biomarkers approach [15]. The simultaneous use of different biomarkers, each of them exploring a component of a complex patho-physiological pathway, has the intrinsic advantage to capture more information linked to the disease under study compared to a single biomarker approach. The concomitant use of different factors may decrease the risk of random variation of single factor obscuring the true signal.
Many medical decisions are made in situations in which multiple factors must be weighted. Because multiple predictive features may interact and correlate with outcomes in complex ways, effective statistical and modeling tools are needed to integrate these data and to determine their implications. This is also true for biomarkers use. When single factor approach is applied to the analysis of multi-markers data, only one factor (marker) is varied at time, with the others factors held constant. This is the case of classical multivariable statistical techniques. With these techniques the combined interpretation of a given set of potential predictors with respect to individual patients may be difficult. This is mainly due to the limitations imposed by the underlying non-linearities and to complex interactions between the factors under study.
Neural networks can input multiple factors simultaneously, combining and recombining them in different ways according to specific (generally non linear) equations. Thus, the higher predictive values obtained by AAN could be explained because of the consideration of parameters which might not reach significance for the entire population, but are highly significant within subgroups. On the other hand, the conventional statistics reveal only parameters which are significant for the entire population.
The adaptive systems had superior accuracy when compared to models of classic linear statistics. In fact, assessing the presence of AD, ANNs were more efficient than conventional statistical analysis, e.g. discriminant analysis, achieving a correct performance (diagnosis) in extremely high percentage of subjects on average and reaching a predictive accuracy of 93% when the best net was used.
Recently, the hypothesis to use a combination of biochemical tests in order to improve the discrimination of AD in early stages has been put forward [8].
The use of ANNs for detecting association between biological markers and disease susceptibility has recently attracted the attention of the scientific and clinical community [16,17]. In fact, to enhance diagnostic accuracy combining tests should be considered to increase the discriminative power of the analysis. In this regard, two different strategies might be followed: either combining tests related to different patho-physiological pathways or associating biomarkers linked to the same biological cascade.
Here we showed that AAN models allow to evaluate three key-elements of amyloid cascade for diagnostic purpose, reaching sensitivity and specificity values higher than those obtained in clinical practice. Studies performed on a large sample with very mild AD patients, pre-symptomatic subjects or patients with other kind of dementia, will be useful to confirm the diagnostic value of this approach.
==== Refs
Ritchie K Lovestone S The dementias Lancet 2002 360 1759 66 12480441 10.1016/S0140-6736(02)11667-9
Nestor PJ Scheltens P Hodges JR Advances in the early detection of Alzheimer's disease Nat Med 2004 10 S34 41 15298007 10.1038/nrn1433
No authors listed Consensus report of the Working Group on: "Molecular and Biochemical Markers of Alzheimer's Disease". The Ronald and Nancy Reagan Research Institute of the Alzheimer's Association and the National Institute on Aging Working Group Neurobiol Aging 1998 19 109 16 9558143 10.1016/S0197-4580(98)00022-0
Di Luca M Pastorino L Bianchetti A Perez J Vignolo LA Lenzi GL Differential level of platelet amyloid beta precursor protein isoforms: an early marker for Alzheimer disease Arch Neurol 1998 55 1195 200 9740113 10.1001/archneur.55.9.1195
Baskin F Rosenberg RN Iyer L Hynan L Cullum CM Platelet APP isoform ratios correlate with declining cognition in AD Neurology 2000 54 1907 9 10822427
Padovani A Pastorino L Borroni B Colciaghi F Rozzini L Monastero R Amyloid precursor protein in platelets: a peripheral marker for the diagnosis of sporadic AD Neurology 2001 57 2243 8 11756604
Colciaghi F Borroni B Pastorino L Marcello E Zimmermann M Cattabeni F [alpha]-Secretase ADAM10 as well as [alpha]APPs is reduced in platelets and CSF of Alzheimer disease patients Mol Med 2002 8 67 74 12080182
Colciaghi F Marcello E Borroni B Zimmermann M Caltagirone C Cattabeni F Platelet APP, ADAM 10 and BACE alterations in the early stages of Alzheimer disease Neurology 2004 62 498 501 14872043
Lisboa PJ A review of evidence of health benefit from artificial neural networks in medical intervention Neural Netw 2002 15 11 39 11958484 10.1016/S0893-6080(01)00111-3
McKhann G Drachman D Folstein M Katzman R Price D Stadlan EM Clinical diagnosis of Alzheimer's disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer's Disease Neurology 1984 34 939 44 6610841
Rumelhart DE McClelland JL PDP Group (eds) Parallel Distributed Processing vol 1: Foundation, Vol 2: Psychological and Biological Models 1986 Cambridge, MA: The MIT Press
Buscema M Breda M Terzi S Sine Net: In Semeion Technical Paper n 21, Roma 2002
Buscema M BackPropagation Neural Networks: In Buscema & Semeion Group, Reti Neurali Artificiali e Sistemi Sociali Complessi, Teoria e Modelli Milano: F Angeli 1999 1 189 222
Dietterich TG Approximate statistical tests for comparing supervised classification learning algorithms 1998 10 Neural Computations MIT Press 1825 1923 9885903
Okamura N Arai H Maruyama M Higuchi M Matsui T Tanji H Combined Analysis of CSF Tau Levels and [(123)I]Iodoamphetamine SPECT in Mild Cognitive Impairment: Implications for a Novel Predictor of Alzheimer's Disease Am J Psychiatry 2002 159 474 6 11870015 10.1176/appi.ajp.159.3.474
Curtis D North BV Sham PC Use of an artificial neural network to detect association between a disease and multiple marker genotypes Ann Hum Genet 2001 65 95 107 11415525 10.1046/j.1469-1809.2001.6510095.x
North BV Curtis D Cassell PG Hitman GA Sham PC Assessing optimal neural network architecture for identifying disease-associated multi-marker genotypes using a permutation test, and application to calpain 10 polymorphisms associated with diabetes Ann Hum Genet 2003 67 348 56 12914569 10.1046/j.1469-1809.2003.00030.x
|
16048651
|
PMC1198261
|
CC BY
|
2021-01-04 16:39:28
|
no
|
J Transl Med. 2005 Jul 27; 3:30
|
utf-8
|
J Transl Med
| 2,005 |
10.1186/1479-5876-3-30
|
oa_comm
|
==== Front
Virol JVirology Journal1743-422XBioMed Central London 1743-422X-2-631610517510.1186/1743-422X-2-63ResearchDevelopment of an in vitro cleavage assay system to examine vaccinia virus I7L cysteine proteinase activity Byrd Chelsea M [email protected] Dennis E [email protected] Molecular and Cellular Biology Program, Oregon State University, 220 Nash Hall, Corvallis, Oregon, 97331, USA2 Siga Technologies, 4575 SW Research Way, Suite 230, Corvallis, Oregon, 97333, USA2005 16 8 2005 2 63 63 21 4 2005 16 8 2005 Copyright © 2005 Byrd and Hruby; licensee BioMed Central Ltd.2005Byrd and Hruby; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Through the use of transient expression assays and directed genetics, the vaccinia virus (VV) I7L gene product has been implicated as the major maturational proteinase required for viral core protein cleavage to occur during virion assembly. To confirm this hypothesis and to enable a biochemical examination of the I7L cysteine proteinase, an in vitro cleavage assay was developed. Using extracts of VV infected cells as the source of enzyme, reaction conditions were developed which allowed accurate and efficient cleavage of exogenously added core protein precursors (P4a, P4b and P25K). The cleavage reaction proceeded in a time-dependent manner and was optimal when incubated at 25°C. I7L-mediated cleavage was not affected by selected inhibitors of metalloproteinases, aspartic acid proteinases or serine proteinases (EDTA, pepstatin, and PMSF, respectively), but was sensitive to several general cysteine proteinase inhibitors (E-64, EST, Iodoacetic acid, and NEM) as well as the I7L active site inhibitor TTP-6171 [C. Byrd et al., J. Virol. 78:12147–12156 (2004)]. Finally, in antibody pull down experiments, it could be demonstrated that monospecific αI7L serum depleted the enzyme activity whereas control sera including αG1L, directed against the VV metalloproteinase, did not. Taken together, these data provide biochemical evidence that I7L is a cysteine proteinase which is directly involved in VV core protein cleavage. Furthermore, establishment of this I7L-mediated in vitro cleavage assay should enable future studies into the enzymology and co-factor requirements of the proteolysis reaction, and facilitate antiviral drug development against this essential target.
==== Body
Background
The Orthopoxviridae include vaccinia virus, camelpox, cowpox, ectromelia, monkeypox, raccoonpox, skunkpox, taterapox, volepox, and variola. Viruses in this family are the cause of numerous diseases including smallpox (variola), and recent human outbreaks of monkeypox. Orthopoxviruses are large double-stranded DNA viruses that are unique amongst DNA viruses in that they replicate exclusively within the cytoplasm of infected cells. Vaccinia virus (VV) is the most extensively studied virus in this group and is the prototypic member. The genome of VV is predicted to encode over 200 open reading frames. VV expresses its genetic information in three stages, as early, intermediate, and late genes. The early genes, which account for approximately half of the genome and are transcribed prior to DNA replication, encode many of the proteins involved in viral DNA replication and intermediate gene expression. The intermediate genes, of which only a handful have been identified, are expressed after the onset of DNA replication, and encode proteins that are activators of late gene expression. The late genes encode many proteins required for the transcription of early genes, the viral structural proteins and the enzymes necessary to process these proteins into their mature form.
Many viruses use proteolytic processing as a key step in their developmental cycle. RNA viruses and retroviruses commonly undergo formative proteolysis in which large polyproteins are cleaved by viral encoded proteinases to produce the structural and nonstructural proteins required for morphogenesis. DNA viruses such as poxviruses and adenoviruses commonly use another type of proteolysis, called morphogenic proteolysis where precursor proteins are first synthesized and then cleaved by viral proteinases to produce the mature form of the protein. The mature protein then plays an essential role in virion formation. During VV assembly, as the spherical immature virions (IVs) are maturing into the first infectious form of vaccinia virus, intracellular mature virus (IMV), a series of events takes place including proteolytic processing of viral core proteins [1-4].
Our laboratory has worked to identify and characterize the proteinases of VV in order to understand their regulation, function, and biochemistry, with a long term goal of developing inhibitors of these enzymes as antiviral drugs. The gene product of the I7L open reading frame recently has been suggested to be the core protein proteinase of VV through the use of an in vivo trans processing assay [5,6]. I7L is an essential late gene, as shown through temperature sensitive mutant viruses [7,8] and conditional lethal mutant viruses [9,10] where under non-permissive conditions, viral morphogenesis is blocked prior to the formation of IMV. I7L is predicted to be a 47 kDa cysteine proteinase that cleaves the major core protein precursors P4a, P4b, and P25K, products of the A10L, A3L, and L4R open reading frames respectively, at a novel Ala-Gly-Xaa cleavage site with cleavage occurring after the glycine residue [5,6]. I7L also is likely to be responsible for cleavage of the A17 membrane protein, at an Ala-Gly-Ala site [9]. This consensus Ala-Gly-Xaa cleavage site of vaccinia is similar to that used for both the adenovirus and African swine fever virus proteinases which cleave after the second glycine in a Gly-Gly-Xaa motif [11,12].
Comparative sequence analysis has suggested that the VV I7L proteinase is related to the ASFV and adenovirus cysteine proteinases and may form a new family of SUMO-1 related enzymes [13,12]. The nucleophilic cysteine is responsible for cleavage and is activated by the imidazol group of the catalytic histidine residue. Substrate specificity is determined by the substrate binding pocket and is unique for each proteinase. Several critical residues have been identified as being necessary for enzymatic activity of I7L including the catalytic triad residues [6]. Based on the identification of the catalytic residues and the predicted structure of the I7L proteinase, a new class of small molecule inhibitors was developed that are capable of inhibiting the replication of VV, and were found to specifically target I7L through the generation of drug resistant mutant viruses with the mutations mapping to I7L [14].
To date, direct studies on the enzymology of I7L-mediated proteolysis have not been possible due to the absence of a suitable biochemical assay. In the experiments reported here, we describe the development of an in vitro I7L-mediated cleavage assay. We have used this system to obtain both biochemical and immunological data to prove that I7L is directly involved in cleavage of the major VV core protein precursors. Having this assay available will now facilitate biochemistry of the I7L enzyme and identification of all the required reaction components to be undertaken.
Results
To date, all studies of VV I7L activity have been carried out indirectly in transfected/infected tissue culture cells. Although this approach has provided some important insights into I7L biology, it is limited with respect to the study of I7L enzymology and identification of all the cis and trans factors required for substrate identification and catalysis. In order to approach these questions, we have sought to develop an in vitro cleavage assay for I7L. Thus far, the obvious approaches of expressing and purifying I7L from prokaryotic and eukaryotic expression vectors and combining with peptides or proteins containing a canonical A-G-X cleavage site have not been successful (data not shown), perhaps due to either the lack of essential co-factors or inappropriate assay conditions. As an alternative approach, we sought to develop a cleavage assay using infected cell extracts as the source of I7L activity and labeled core protein precursors made in vitro as the substrate. If successful, this system would provide the starting point for a dissection of the essential reaction components.
In vitro Processing of Core Protein Precursors
The three major core protein precursors P4a (A10L), P4b (A3L), and P25K (L4R) which are known to be cleaved to a mature form (Figure 1) were cloned into plasmid vectors driven off of a T7 promoter to be used as a source of substrate for the assay. To investigate the ability of I7L to cleave the P4a, P4b, and P25K substrates in vitro, we have used a system where the substrates are produced from an in vitro transcription and translation assay using rabbit reticulocyte lysates and then mixed with I7L expressed from virus infected cells. BSC40 cells are infected with ts16, a temperature sensitive mutant virus in which the responsible mutation maps to I7L. The virus infected cells are incubated at the non-permissive temperature and transfected with plasmids expressing either wild-type I7L (pI7L) or I7L with the catalytic histidine residue mutated to an alanine (pI7LH241A). The extracts are prepared as described in the Materials and Methods. The extracts are mixed and incubated with the substrates for 3 hrs and then analyzed through SDS-PAGE and chemiluminescent detection. As shown in Figure 2, a specific band corresponding to unprocessed P4a (top panel), P4b (middle panel), or P25K (bottom panel) is produced when the substrate is run alone. When mixed with cellular extracts, or extracts from cells infected with ts16 at the non-permissive temperature and transfected with mutant I7L, no cleavage products are observed. However, when mixed with extracts from either cells infected with ts16 at the permissive temperature or cells infected with ts16 at the non-permissive temperature transfected with wild-type I7L, the cleaved products 4a, 4b, and 25K are observed. Substrates with mutated A-G-X sites were not cleaved indicating that cleavage was occurring at the correct sites (data not shown). For the rest of the reported studies, P25K was used as the source of substrate since it gave the best cleavage profile.
Figure 1 Schematic representation of the major core protein precursor cleavage products. The vaccinia virus genome is represented depicting three of the major core protein precursors, the gene products of the L4R, A10L, and A3L open reading frames, P25K, P4a, and P4b respectively. The precursors are shown being cleaved into their mature form. Molecular mass is indicated.
Figure 2 In vitro proteolytic processing of P4a, P4b, and P25K. 1 μl of TNT produced substrate either P4a (A), P4b (B), or P25K (C) was mixed with 5 μl of Hepes buffer and 14 μl of enzyme extracts, either from uninfected cells, or cells infected with ts16 at the permissive or non-permissive temperature. At the non-permissive temperature, plasmid borne I7L, either wild-type (pI7L) or mutant I7L (pI7LH241A) was transfected in as the source of enzyme. The reaction was incubated at 29°C for 3 hrs before being stopped by the addition of SDS sample buffer. Molecular weight is indicated on the left and the core protein precursor and product on the right. Lane 1 is substrate alone, lane 2 is substrate mixed with cellular extracts and lanes 3–5 are substrate mixed with the enzyme extract indicated.
Processing Kinetics of Core Protein Precursors
To determine the optimal temperature and kinetics of processing of the core protein precursors in the in vitro cleavage assay, a time course of I7L-mediated processing at various temperatures was performed. As shown in Figure 3A, at 0°C, no processing was observed during the 20 hr time period. At 25°C, a gradual increase in the amount of P25K cleavage product was observed starting at 15 min and increasing throughout the 20 hr incubation period (Fig. 3B). Compared with the rate of cleavage at 25°C, cleavage was slower at 30°C (Fig. 3C), starting around 30 min and increasing through the 20 hr period, but never to the same level as at 25°C. Processing is greatly reduced at 37°C with only a faint processed band ever appearing (Fig 3D).
Figure 3 Processing kinetics of P25K. Samples were incubated at either 0°C (A), 25°C (B), 30°C (C), or 37°C (D) for up to 20 hrs, harvested at the indicated times and the reaction stopped by the addition of SDS sample buffer. Incubation temperature is indicated on the left and P25K precursor and 25K mature product are indicated on the right.
Influence of Thiol Reagents on the Protease Activity
Based on its sequence similarity to the adenovirus protease, the African swine fever virus protease, and an ubiquitin-degrading enzyme in yeast, as well as the identity of a catalytic triad composed of histidine, cysteine, and aspartic acid, I7L has been classified as a cysteine proteinase. The thiol reagents dithiothreitol (DTT) and cysteine have been shown to enhance the cleavage activity of the adenovirus protease in an in vitro peptide cleavage assay [15]. To determine whether these agents have a similar effect on the activity of I7L, they were added to the in vitro assay in a final concentration from 0–10 mM. However, no increase in cleavage activity was observed with the addition of either DTT or cysteine (data not shown). It is possible that once purified recombinant enzyme is produced these thiol reagents may increase its activity.
Effect of Inhibitors on Protease Activity and Characterization as a Cysteine Proteinase
The in vitro assay allowed us to test the effects of various protease inhibitors, as well as specific small molecule inhibitors on the activity of I7L. As shown in Figure 4 and Table 1, the metalloproteinase inhibitor ethylenediaminetetraacetic acid (EDTA), the aspartic proteinase inhibitor pepstatin, and the serine proteinase inhibitor phenylmethanesulfonyl (PMSF) had no detectable effect on cleavage activity. The cysteine proteinase inhibitors iodoacetic acid (IA) and N-ethylmaleimide (NEM) efficiently blocked I7L mediated proteolysis of P25K. The cysteine proteinase inhibitors E-64 and EST were shown to inhibit protease activity at a relatively high concentration, but not at the lower concentration tested. This is consistent with what has been observed for both the adenovirus protease [16], and the African swine fever virus protease [17]. The failure of E-64 to inhibit protease activity at the lower concentration tested, and the location of the active site residues may suggest that each of these enzymes are not conventional papain-like enzymes, but may be a new family of cysteine proteinases. The cysteine protease inhibitor leupeptin also failed to inhibit protease activity, although this lack of inhibition was also observed with the adenovirus proteinase [16].
Figure 4 Effect of inhibitors on in vitro processing. Various concentrations of protease inhibitors were added to the in vitro processing assay for 6 hr at 29°C. The first lane is P25K expressed alone with no extract added. The second lane is P25K mixed with cellular extracts and the third lane is P25K mixed with I7L enzyme extracts. Each of the remaining lanes has P25K mixed with I7L enzyme extracts plus indicated inhibitor. Ethylenediaminetetraacetic acid (EDTA) was used at 1 mM. Pepstatin A, Pep, was used at 10 μM. Phenlymethanesulfonyl fluoride (PMSF) was used at 1 mM. N-(trans-Epoxysuccinyl)-L-leucine 4-guanidinobutylamide trans-Epoxysuccinyl-L-leucylamido(4-guanidino)butane (E-64) and a related product EST, were both used at 10 μM and 100 μM concentrations. Iodoacetic acid (IA) was used at 1 mM. Leupeptin (Leu) was used at 1 mM, and N-ethlymaleimide (NEM) was used at 2.5 mM. The concentrations of TTP-6171, TTP-1021, and TTP-0961 are indicated. The table indicates the concentration of inhibitor used and whether cleavage activity was observed.
Table 1 Effect of inhibitors on in vitro processing.
Inhibitor Name Concentration Inhibit Cleavage
Metalloproteinase EDTA 1 mM No
Aspartic acid proteinase Pepstatin 10 μM No
Serine proteinase PMSF 1 mM No
Cysteine proteinase E-64 10 μM No
E-64 100 μM Yes
EST 10 μM No
EST 100 μM Yes
IA 1 mM Yes
Leupeptin 1 mM No
NEM 2.5 mM Yes
TTP inhibitors TTP-6171 50 μM Yes
TTP-6171 20 μM Yes
TTP-1021 50 μM Yes
TTP-1021 20 μM Yes
TTP-0961 50 μM No
TTP-0961 20 μM No
Next we wanted to determine if the small molecule I7L inhibitors previously developed as antiviral drug candidates [14] could be shown to specifically inhibit the activity of I7L in the in vitro assay. The compound TTP-6171 has been shown to inhibit viral replication in tissue culture, with drug resistant virus mutations mapping to I7L [14]. Here we see that this compound along with TTP-1021, which was also found to inhibit I7L in tissue culture, inhibits the processing of P25K in vitro. However the compound TTP-0961, which was not found to generate resistant mutants in the I7L gene (data not shown), does not inhibit cleavage. These results demonstrate that this assay can be used for the screening of specific I7L inhibitors and confirms that this class of molecules targets I7L.
Effects of I7L antibody competition on cleavage
To directly demonstrate that the cleavage observed in the in vitro assay requires the presence of I7L, increasing concentrations of I7L specific antiserum were added to the enzyme extracts overnight, and then the complex was precipitated with Protein A sepharose beads to deplete the extract of I7L and any associated co-factors. As shown in Figure 5, both of the I7L antisera tested inhibited cleavage of P25K while an antiserum targeting a different VV gene product, G1L, did not inhibit cleavage.
Figure 5 Effect of antibody competition on in vitro processing. Lane 1 is P25K expressed alone. Lane 2 is P25K mixed with I7L enzyme extracts. Lane 3 is P25K mixed with I7L extracts that have been diluted with Hepes buffer and treated with Sepharose beads. Lanes 4, 5, and 6 are P25K mixed with I7L extracts that have been incubated overnight with different I7L antiserum (indicated on each lane), treated with Sepharose beads and the antibody complex removed by centrifugation. Lane 7 is P25K mixed with I7L extracts incubated with G1L antiserum as above.
Discussion
In this report, a cell-free transcription and translation system was used to develop an in vitro cleavage assay for the VV cysteine proteinase I7L. Proteolytic activity was obtained by co-expression of I7L in ts16 infected cells at the non-permissive temperature. Each of the major core protein precursors, P4a, P4b, and P25K, were shown to be cleaved to their mature products by I7L using the in vitro assay. Evidence that this cleavage is specific to I7L was shown through the fact that expressing a mutant form of I7L resulted in the inability to cleave the core protein precursors. Antibody pull down experiments with αI7L supported the conclusion that I7L plays a direct role in the proteolytic reaction.
A time course of processing at various temperatures indicated that for this particular assay, the optimal temperature for the reaction to be carried out at is 25°C with processing beginning as soon as 15 minutes after addition of enzyme and increasing as time progresses. The cleavage reaction was never driven to completion and this may be due to a lack of replenishing co-factors or the enzyme may have been used up in the reaction. It was surprising that the optimal reaction temperature was 25°C instead of 37°C which is the optimal growth temperature for VV in cell culture. One possible explanation is that I7L is present at high concentrations in the extract and one can measure marginal activity at low temperature, whereas at higher temperatures other proteinases are activated which degrade the I7L enzyme.
Known cysteine protease inhibitors such as E-64, iodoacetic acid, and NEM were shown to inhibit the in-vitro cleavage reaction while the metalloproteinase inhibitor EDTA, the aspartic acid protease inhibitor pepstatin, and the serine protease inhibitor PMSF all failed to inhibit the cleavage reaction indicating that the enzyme responsible for cleavage is a cysteine protease. Interestingly the cysteine protease inhibitors leupeptin, and low concentrations of E-64 did not inhibit the reaction. These cysteine protease inhibitors were also not shown to be effective against either the African Swine Fever Virus protease [17] or the adenovirus protease [16], further providing support for the theory that these enzymes may form a new family of cysteine proteases that differ from papain-like cysteine proteases.
Of particular interest, the small molecule inhibitors designed to fit into the active site pocket of I7L and previously shown to inhibit viral replication [14], were found to be active in inhibiting the in vitro cleavage reaction described here. A related compound (TTP-0961) that was not found to map to I7L was not able to abolish cleavage. This indicated that this assay may be useful for high-throughput screening of compounds to identify those that have specific activity for I7L.
Conclusion
Until this point, all work demonstrating that I7L is the core protein proteinase has been done through transient-expression assays and the use of conditional lethal viruses in tissue culture [9,5,6,10]. The data obtained has indicated that I7L is essential for these processing activities, it did not rule out the possibility that some other factor or enzyme was also required for this activity to occur. Through the use of an in vitro assay we have shown that I7L is capable of cleaving the core protein precursors but that an additional co-factor is required for this activity to occur since expression of the enzyme through cell-free translation produced inactive enzyme. The co-factor(s) necessary for cleavage have yet to be determined. However, having the assay described in this report available will now enable a reductive analysis to be conducted to identify all the essential components of the reaction and to study their individual biochemical characteristics.
Methods
Cells and Viruses
BSC40 cells [18] were grown in Eagle's minimal essential medium containing 5% fetal calf serum (FCS) (Sigma, St. Louis, MO), 2 mM glutamine (Invitrogen, Carlsbad, CA), and 15 μg/ml gentamicin sulfate (Invitrogen) in a 37°C incubator with 5% CO2. Purified ts16 Vaccinia virus was prepared as described [19]. Escherichia coli strains were grown in Luria-Bertani broth or on Luria-Bertani medium containing 1.5% agar and ampicillin at 50 μg/ml.
Plasmids
The A10L (P4a) gene was amplified by polymerase chain reaction using oligonucleotides KH10 (5' CATGCCATGGATGATGCCTATTAAGTCAATAGTTACT CTT-3') and KH11 (5'-CCGCTCGAGTTATTCATCATCAAAAGAGACAGAGTC-3'), digested with NcoI and XhoI, and cloned into the pTM1 vector, yielding pTM-P4a which utilizes a T7 promoter for expression. The A3L (P4b) gene was amplified using oligonucleotides KH08 (5'-CATGCCATGGATGGAAGCCGTGGTCAATAG-3') and KH09 (5'-TCCCCCGGGCTAAAAATAGTTCTGTAATATGTCTAGCGCT-3'), digested with NcoI and SmaI, and cloned into the pTM1 vector to yield pTM-P4b. The L4R (P25K) gene was amplified using oligonucleotides DN51 (5'-CATGCCATG GATGAGTCTACTGCTAGAAAAC-3') and KH07 (5'-CCGCTCGAGTCAATCCTTT GTCG-3'), digested with NcoI and XhoI, and cloned into the pTM1 vector to yield pTM-P25K. The pI7L and pI7LH241A plasmids were described in Byrd et al., 2002 [5].
Preparation of polyprotein or proteinase-containing extracts
Confluent monolayers of BSC40 cells in 6-well plates were infected with ts16 VV at a multiplicity of infection of 2 plaque-forming units per cell and transfected with 2 μg of plasmid DNA (either pI7L, or pI7LH241A) using DMRIE-C (Invitrogen) following the manufacturer's indications. Infected cells were incubated either at the permissive temperature of 31.5°C or the non-permissive temperature of 39°C. Cells were harvested at 24 h post-infection by pipetting up and down to lift the cells from the surface. The infected cells were centrifuged at 10,000 × g for 10 min, the supernatant was aspirated off, and the pellet was resuspended in 500 μL homogenization buffer containing 20 mM HEPES (pH 7.4), 0.28 M sucrose, 2 mM EDTA. This was passed through a 25-gauge syringe 15 times. The homogenate was centrifuged at 700 × g for 5 min to separate the nuclei and unbroken cells from the supernatant. The supernatant was centrifuged at 100,000 × g for 30 min at 4°C to separate the membrane/particulate material from the supernatant. The supernatant was used as the source of enzyme.
Coupled TNT reactions with T7 RNA polymerase were performed according to the manufacturer's instructions (Promega Corporation, Madison, Wisconsin) as a source of substrate. Briefly, the TNT reactions were performed at 30°C in a final volume of 25 μL with 1 μg of plasmid DNA, using the non-radioactive Transcend label (biotinylated lysine residues are incorporated in the protein) provided with the kit for detection of protein.
In vitro cleavage assay
Reactions were performed at the indicated temperature in a final volume of 20 μL containing 1 μL of substrate, 13 μL of enzyme extract, and 6 μL of 20 mM HEPES (pH 7.4) buffer, pH 7.4. After the indicated times, the reaction was stopped by the addition of SDS sample buffer, and the samples were subjected to SDS-polyacrylamide gel electrophoresis. The results were analyzed by immunoblotting following the instructions provided by the TNT kit.
Inhibitor studies
For inhibitor studies, the reactions described above were incubated for 6 hr in the presence or absence of the following protease inhibitors: 1 mM phenylmethanesulfonyl fluoride (PMSF) (Sigma), 10 μM Pepstatin A (Sigma), 1 mM ethylenediaminetetraacetic acid (EDTA) (Sigma), 10 μM or 100 μM N-(trans-Epoxysuccinyl)-L-leucine 4-guanidinobutylamide trans-Epoxysuccinyl-L-leucylamido(4-guanidino)butane (E-64) (Sigma), 1 mM iodoacetic acid (Sigma), 10 μM or 100 μM Leupeptin (Roche, Indianapolis, IN), 2.5 mM N-ethylmaleimide (NEM) (Sigma). For I7L specific inhibition studies, the reactions described above were incubated for 6 hr in the presence or absence of TTP-6171, TTP-1021, or TTP-0961 [14] at 5 μM or 20 μM final concentrations.
Antibody competition studies
For the antibody competition studies, 25 μl of I7L or G1L specific antiserum was added to 25 μL of enzyme extract on a rotating shaker overnight at 4°C. ProteinA: Sepharose beads (Amersham Biosciences, Uppsla, Sweden) were added for 3 hrs and the antibody complex was centrifuged to pull down the I7L enzyme. The supernatant was used as the source of extract in the in vitro assay described above. As a control, enzyme extract was mixed with buffer instead of antibody and treated with beads in a similar manner.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
CMB conceived the study, conducted all the experiments and wrote the manuscript. DEH coordinated the research efforts and edited the paper. Both authors read and approved the final manuscript.
Acknowledgements
We would like to thank Kady Honeychurch for constructing pTM:L4R, pTM:A3L, and pTM:A10L, Rich Condit for providing ts16, and TransTech Pharma for supplying TTP-6171, TTP-1021, and TTP-0961. This work was funded by NIH grant AI-060160.
==== Refs
Moss B Rosenblum EN Protein cleavage and poxvirus morphogenesis: tryptic peptide analysis of core precursors accumulated by blocking assembly with rifampicin J Mol Biol 1973 81 267 269 4777308 10.1016/0022-2836(73)90195-2
Silver M Dales S Biogenesis of vaccinia: interrelationship between post-translational cleavage, virus assembly, and maturation Virology 1982 117 341 356 7039091 10.1016/0042-6822(82)90474-3
VanSlyke JK Franke CA Hruby DE Proteolytic maturation of vaccinia virus core proteins: identification of a conserved motif at the N termini of the 4b and 25K virion proteins J Gen Virol 1991 72 411 416 1993877
VanSlyke JK Whitehead SS Wilson EM Hruby DE The multistep proteolytic maturation pathway utilized by vaccinia virus P4a protein: a degenerate conserved cleavage motif within core proteins Virology 1991 183 467 478 1853556 10.1016/0042-6822(91)90976-I
Byrd CM Bolken TC Hruby DE The vaccinia virus I7L gene product is the core protein proteinase J Virol 2002 76 8973 6 12163618 10.1128/JVI.76.17.8973-8976.2002
Byrd CM Bolken TC Hruby DE Molecular dissection of the vaccinia virus I7L core protein proteinase J Virol 2003 77 11279 11283 14512576 10.1128/JVI.77.20.11279-11283.2003
Ericsson M Cudmore S Shuman S Condit RC Griffiths G Locker JK Characterization of ts16, a temperature-sensitive mutant of vaccinia virus J Virol 1995 69 7072 7086 7474127
Kane EM Shuman S Vaccinia virus morphogenesis is blocked by a temperature-sensitive mutation in the I7 gene that encodes a virion component J Virol 1993 67 2689 2698 8386272
Ansarah-Sobrinho C Moss B Role of the I7 protein in proteolytic processing of vaccinia virus membrane and core components J Virol 2004 78 6335 6343 15163727 10.1128/JVI.78.12.6335-6343.2004
Byrd CM Hruby DE A conditional-lethal vaccinia virus mutant demonstrates that the I7L gene product is required for virion morphogenesis Virol J 2005 2 4 15701171 10.1186/1743-422X-2-4
Webster A Russell S Talbot P Russell WC Kemp GD Characterization of the adenovirus proteinase: substrate specificity J Gen Virol 1989 70 3225 3234 2691633
Andres G Alejo A Simon-Mateo C Salas ML African swine fever virus protease, a new viral member of the SUMO-1-specific protease family J Biol Chem 2001 276 780 787 11031264 10.1074/jbc.M006844200
Li SJ Hochstrasser M A new protease required for cell-cycle progression in yeast Nature 1999 398 246 51 10094048 10.1038/18457
Byrd CM Bolken TC Mjalli AM Arimilli MN Andrews RC Rothlein R Andrea T Rao M Owens KL Hruby DE New class of orthopoxvirus antiviral drugs that block viral maturation J Virol 2004 78 12147 12156 15507601 10.1128/JVI.78.22.12147-12156.2004
Webster A Hay RT Kemp GD The adenovirus protease is activated by a virus-coded disulphide-linked peptide Cell 1993 72 97 104 8422686 10.1016/0092-8674(93)90053-S
Webster A Russell WC Kemp GD Characterization of the adenovirus proteinase: development and use of a specific peptide assay J Gen Virol 1989 70 3215 3223 2691632
Rubio D Alejo A Rodriguez I Salas ML Polyprotein processing protease of African swine fever virus: purification and biochemical characterization J Virol 2003 77 4444 4448 12634404 10.1128/JVI.77.7.4444-4448.2003
Raczynski P Condit RC Specific inhibition of vaccinia virus growth by 2'-O-methyladenosine: isolation of a drug-resistant virus mutant Virology 1983 128 458 62 6412452 10.1016/0042-6822(83)90270-2
Hruby DE Guarino LA Kates JR Vaccinia virus replication. I. Requirement for the host-cell nucleus J Virol 1979 29 705 15 107327
|
16105175
|
PMC1198262
|
CC BY
|
2021-01-04 16:38:58
|
no
|
Virol J. 2005 Aug 16; 2:63
|
utf-8
|
Virol J
| 2,005 |
10.1186/1743-422X-2-63
|
oa_comm
|
==== Front
Virol JVirology Journal1743-422XBioMed Central London 1743-422X-2-681611149210.1186/1743-422X-2-68ResearchMatrix attachment regions as targets for retroviral integration Johnson Chassidy N [email protected] Laura S [email protected] Department of Microbiology & Immunology and Tulane Cancer Center, Tulane University School of Medicine, New Orleans, Louisiana, 70112, USA2005 19 8 2005 2 68 68 13 6 2005 19 8 2005 Copyright © 2005 Johnson and Levy; licensee BioMed Central Ltd.2005Johnson and Levy; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The randomness of retroviral integration has been debated for many years. Recent evidence indicates that integration site selection is not random, and that it is influenced by both viral and cellular factors. To study the role of DNA structure in site selection, retroviral integration near matrix attachment regions (MARs) was analyzed for three different groups of retroviruses. The objective was to assess whether integration near MARs may be a factor for integration site selection.
Results
Results indicated that MLV, SL3-3 MuLV, HIV-1 and HTLV-1 integrate preferentially near MARs, specifically within 2-kilobases (kb). In addition, a preferential position and orientation relative to the adjacent MAR was observed for each virus. Further analysis of SL3-3 MuLV insertions in common integration sites (CISs) demonstrated a higher frequency of integration near MARs and an orientation preference that was not observed for integrations outside CISs.
Conclusion
These findings contribute to a growing body of evidence indicating that retroviral integration is not random, that MARs influence integration site selection for some retroviruses, and that integration near MARs may have a role in the insertional activation of oncogenes by gammaretroviruses.
==== Body
Background
An essential step in the replication cycle of all retroviruses is integration of the double-stranded DNA proviral form of the genome into host DNA. The degree of randomness of proviral integration has been debated for many years [1,2]. Studies have suggested that DNaseI hypersensitive sites [3-7], AT-rich regions [8], transcriptionally active regions [2,9-12], repeat elements including Alu and LINE elements [13] and regions of DNA bending, specifically regions with the most DNA distortion [14-18], are preferred sites of proviral integration. Alternatively, studies have shown that high levels of transcription disfavor integration of avian leukosis virus (ALV) [2]. The conflicting results that have been reported may be explained by the small sample sizes examined or by potential biases introduced from the cloning strategies used to identify insertion sites. In addition, many of the studies were performed in vitro, and thus did not take into account the native conformation of chromatin. Before the completion and publication of the human and mouse genome databases, theories for randomness of retroviral integration were difficult to prove or disprove because of the technical challenge of analyzing a large sample size of integrations from infected cells. Since publication of the genome databases, several studies have isolated and mapped hundreds of proviral insertion sites for murine leukemia virus (MLV), human immunodeficiency virus type-1 (HIV-1), avian sarcoma virus (ASV) and human T-cell leukemia virus type-1 (HTLV-1) [11,12,19,20]. For those viruses, the results showed preferential integration into transcriptionally active NCBI Reference Sequences (RefSeqs), but distinct patterns of integration were evident as well. These studies provided strong evidence that distinct viruses differ in proviral integration patterns, but that integration is clearly non-random. The specific pressures that influence site selection for retroviral integration remain incompletely understood.
Accumulating evidence indicates that retroviral integration site selection is influenced by properties of cellular DNA structure [11,21-24]. A recent large-scale study found that DNA structural features such as bendability and A-philicity served as preferred integration sites [22]. The present study was performed to assess the role of matrix attachment regions (MARs) in retroviral integration site selection. MARs are DNA sequences located at the bases of DNA loops that attach to the nuclear matrix, and are thus positioned near the machinery for DNA replication, transcription, RNA processing and transport (reviewed in [25]). There is no consensus sequence that defines a MAR; however, MARs are commonly found to have intrinsic DNA bending properties, to contain transcription factor binding sites, AT-rich stretches, sites for topoisomerase I and II binding and cleavage, and high unwinding potential [26,27]. MARs function as structural regulatory elements by organizing the DNA into loop domains. Studies have shown that MARs influence the expression of cellular genes, and can enhance viral gene expression when in the vicinity of viral promoters and enhancers [28-30]. This property has made the inclusion of MARs in gene therapy vectors attractive for enhanced and prolonged expression of the transgene in a specific cell-type or developmental stage [31-33]. MARs have been implicated in virus-mediated malignancies, particularly as targets of integration by small DNA tumor viruses. Specifically, integrated SV40, HBV, HPV16 and HPV18 have been found within or in close proximity to MARs in tumors or transformed cell lines [34]. Other reports indicate that HTLV-1 and HIV-1 may integrate preferentially near MARS [34,35].
The gammaretroviruses represent a group of mammalian oncogenic retroviruses typically associated with the induction of long-latency leukemia and lymphoma in the natural host. Gammaretroviruses do not encode an oncogene or any other gene to which their malignant potential can be directly attributed. Rather, their ability to induce tumors has been linked to a process termed insertional activation, in which integration of the proviral genome into host DNA is associated with activated expression of an adjacent oncogene. When the same genetic locus is observed to be interrupted by proviral integration in multiple independent tumors, it is inferred that the commonly interrupted locus encodes an oncogene whose activation is relevant to tumor induction [36-38]. Such a locus is referred to as a common insertion site (CIS). We recently described CISs utilized by a recombinant gammaretrovirus, MoFe2-MuLV (MoFe2), in T-cell lymphomas in the NIH/Swiss mouse. To construct MoFe2, the U3 region of the Moloney murine leukemia virus (M-MuLV) long terminal repeat (LTR) was substituted with homologous sequences from a natural isolate of feline leukemia virus termed FeLV-945 [39]. FeLV-945 is characterized by a unique motif in the U3 region of the LTR, which contains a single copy of the transcriptional enhancer followed downstream by the tandem triplication of a 21-bp sequence. Substitution of FeLV-945 LTR sequences into M-MuLV was shown to alter the pattern of insertional activation and to identify new CISs [40]. As described below, the identification of two potential MARs near a CIS in MoFe2-induced lymphomas suggested that MARs may represent a determinant of integration site selection. That hypothesis was addressed in the present study by analyzing the proximity of proviral integrations to MARs in lymphomas and in unselected cultured cells. The patterns of integration with respect to MARs were compared for three groups of retroviruses, including several murine gammaretroviruses, human deltaretrovirus (HTLV-1) and lentivirus (HIV-1).
Results
Previous studies showed that inoculation of neonatal mice with MoFe2 resulted in the development of T-cell lymphoma. Analysis of patterns of common proviral insertion in lymphomas revealed that MoFe2 utilized a set of CISs distinct from either parent virus from which it was constructed [39,40]. Sequence surrounding one of the previously described CISs in MoFe2-induced lymphomas, termed MF8T (Rasgrp1), was analyzed for the presence of MARs using a MAR-prediction program termed MAR-Finder . MAR-Finder is a statistical algorithm that analyzes the pattern density for characteristic DNA sequence motifs that predict the occurrence of MARs, including replication origins, TG-richness, curved DNA, kinked DNA, topoisomerase II recognition and cleavage sites and AT-richness. MAR-Finder has been previously validated for predicting the presence of MARs [34,41,42]. An alternative method to predict MARS is based on detecting the location and extent of stress-induced duplex destabilization (SIDD) through the use of a statistical algorithm termed WebSIDD [43-45]. Although this method has been validated to predict the presence of MARs accurately, recent evidence indicates that stress-induced destabilization of duplex DNA is not sufficient for a sequence to bind to the nuclear matrix; thus, the use of SIDD for the prediction of MARs may lead to false positives [46]. Using MAR-Finder, the results indicated the presence of two MARs in the 60-kb sequence surrounding MF8T, located 5.1-kb and 3.6-kb from the domain of common insertion (Figure 1). The predicted elements were observed to be enriched in motifs characteristic of MARS, including kinked DNA, curved DNA, AT-rich regions, origin of replication patterns and vertebrate and Drosophila topoisomerase II consensus sequences [26,27]. The close proximity of two MARS to the MF8T CIS suggested that integration near MARS may represent a mechanism for retroviral target site selection. To evaluate this possibility, the distance from proviral integration to predicted MARs was analyzed for three different groups of retroviruses, specifically murine gammaretroviruses (MoFe2, SL3-3 MuLV, MLV) human deltaretrovirus (HTLV-1) and lentivirus (HIV-1). Sequence information on MoFe2 integrations was obtained from the CISs and other insertion sites identified previously from a large collection of MoFe2-induced tumors [40]. MoFe2 integration sites were also analyzed from acutely infected SC-1 cells. In total, 42 MoFe2 integration sites were identified and analyzed in the present study. SL3-3 MuLV (SL3-3) integration sites had been previously identified from T-cell lymphomas in NIH-Swiss mice by inverse PCR [47]. In total, 86 SL3-3 integration sites were examined in the present study [47]. MLV and HIV-1 integration sites had been previously identified from HeLa cells infected with pseudotyped retroviral genomes [19]. From the 903 MLV and 379 HIV-1 insertions identified in that study, 49 (MLV) or 41 (HIV-1) integration sites for each virus were chosen at random for the present analysis. HTLV-1 integration sites from tumor-derived cells lines or from ATLL patients had been previously identified [8,12,34], 26 of which were examined in the present study. For each integration site examined in the present study, host-virus junction fragment sequences were obtained from GenBank or the Mouse Retroviral Tagged Cancer Gene Database (RTCGD; ) and the integration sites were thereby positioned in the respective mouse or human genome using the NCBI mouse or human genome database or .
Figure 1 Physical map of the MF8T locus. Depicted is the 3.9-kb domain of common proviral insertion designated MF8T. Vertical lines represent the positions of the proviral integrations with the transcriptional orientation of provirus depicted by the direction of the arrow. Depicted is Rasgrp1, the predicted oncogene in the MF8T locus. Two predicted MARs of 0.9-kb and 0.8-kb in size are located 5.1-kb and 3.6-kb from the domain of common insertion. Also depicted are structural motifs typical of MARs, including kinked DNA, curved DNA, AT-rich regions, ORI patterns and Topoisomerase II cleavage site patterns.
Initial analysis of insertion sites and their proximity to MARs revealed that some integrations were located more than 20-kb from a predicted MAR; therefore, to ensure a thorough identification of MARS in the vicinity of proviral integrations, 60-kb of sequence information surrounding each insertion site was obtained from the respective genome for analysis. Using 60-kb of sequence information surrounding each integration event, the distance from the proviral insertion site to the closest predicted MAR was plotted as the percentage of integration events analyzed (Figure 2). For the murine gammaretroviruses, the results indicated a preference to integrate within 2-kb of a predicted MAR. For example, 46% of SL3-3 integrations and 50% of MLV integrations occurred within 2-kb of a predicted MAR (Figure 2A). It has been reported that MARs occur every 10-kb in the mammalian genome [34,41]. Based on this report, a Monte Carlo simulation was performed where the mean distance to the closest MAR was computed under the assumption that viral integration occurs randomly with respect to regions that are predicted MARs and that MARs occur every 10-kb. The results indicated that, under these assumptions, the mean distance to the closest MAR during a random integration event would be 4-kb [34]. Thus, preferential integration near MARs is indicated for SL3-3 and MLV. By comparison, MoFe2 integration did not show the same preference (Figure 2A); rather, the distribution of MoFe2 integration sites in relation to MARs was significantly different from the distribution observed for SL3-3 and MLV (p < 0.01). In fact, the distribution of MoFe2 insertions in relation to MARs was consistent with the expectation for random integration. The same analysis was then performed on HTLV-1 and HIV-1 to determine if integration near MARs is also common for retroviruses that do not act in disease induction by insertional activation. The results indicated a preference for integration near MARs, since 43.9% of HIV-1 integrations and 42.3% of HTLV-1 integrations occurred within 2-kb of a predicted MAR (Figure 2B). As expected, a small percentage of integration events occurred more than 10-kb from a predicted MAR (Figure 2). In fact, for some integrations sites, the closest MAR in one direction was more than 60-kb away (data not shown). These results illustrate that, although MARs are predicted to be positioned at 10-kb intervals, there are regions of DNA that are either enriched or deficient in MARs as well.
Figure 2 Distance of closest predicted MAR to proviral insertion site. Results are plotted as the percentage of integration events that occurred within 25-kb from a MAR using MAR-Finder for (A) gammaretroviruses (SL3-3, MoFe2 and MLV) and (B) HIV-1 and HTLV-1. SL3-3 and MLV integration distribution was significantly different than MoFe2 as determined by a one-way ANOVA followed by Tukey's multiple comparison test.
Previous reports have indicated that MAR-mediated enhancement of viral gene expression is directional [32,34]. Other reports, in contrast, have indicated that MARs function to enhance gene expression in an orientation- and position- independent manner when located near the promoter [48]. To examine whether the preferred gammaretroviral integration near MARs is directional, it was next determined whether the closest predicted MAR was located upstream or downstream of the proviral integration site with respect to the transcriptional direction of the genetic locus. Results of the analysis, plotted as a percentage of integration events, indicated that the majority of MLV integrations occurred 1- to 2-kb from a predicted MAR on the downstream side (Figure 3A). For SL3-3, it was useful to consider independently the integrations previously identified as CISs in tumor DNA, since those integrations presumably function to activate nearby oncogenes [47]. Interestingly, SL3-3 insertions identified as CISs were found to integrate commonly within 2-kb from a predicted MAR and to be positioned on the upstream side. Of 31 such insertions examined, 29% were integrated within 2-kb upstream as compared to 5.8% integrated within 2-kb downstream of a predicted MAR (Figure 3A). By comparison, 44 SL3-3 integrations identified as only single insertion sites (ISs) did not show the same directional preference for integration near MARs (Figure 3A). These findings imply that SL3-3 integration immediately upstream of MARs within CISs may be related to insertional activation of the adjacent oncogene.
Figure 3 Position of MAR closest to the proviral integration site. The closest predicted MAR to the site of proviral insertion was determined to be located upstream or downstream from the site of insertion with respect to the transcriptional direction of the genetic locus. The results are plotted as the percentage of integrations that occurred up to 10-kb from a MAR for (A) SL3-3 insertions at single insertion site (SL3-3 IS), SL3-3 insertions at common insertion sites (SL3-3 CIS), MLV and (B) HTLV-1 and HIV-1.
When examined by the same approach, analysis of HIV-1 and HTLV-1 integrations indicated that the majority of proviral insertions occurred near MARs, and 80% of the HIV-1 proviral integrations that occurred within 1- to 2-kb of a MAR were positioned downstream (Figure 3B). HTLV-1, while integrated preferentially within 2-kb of a MAR, did not show a position preference. A recent study also analyzed HIV-1 integration sites for their proximity to MARs. Consistent with our findings, that study indicated HIV-1 integration near MARs, specifically in the downstream position [35]. Another study, however, reported that MARs are commonly found downstream from the sites of HTLV-1 integrations [34]. As noted, we did not observe a position preference for HTLV-1 integrations relative to MARs (Figure 3B). The conflicting results may be due to the small sample size (n = 3) examined in the previous study.
Several recent studies have reported that HIV-1, MLV and HTLV-1 integrate preferentially into genes [12,19,20]. With these findings in mind, SL3-3 and MoFe2 insertion sites were analyzed to determine whether a preference is evident for integration into RefSeqs. The analysis revealed that 17.6% SL3-3 integrations at CISs, 40.3% of SL3-3 integrations at single insertion sites, and 33.3% MoFe2 insertions occurred within RefSeqs (data not shown). By comparison, the frequency of integration into genes by random chance has been estimated at 22% [12,19,20]. Thus, preferential integration into genes was identified for MoFe2 and SL3-3 at single insertion sites, although not for SL3-3 integrated at CISs. Analysis was then performed to determine if preferred integration into genes was associated with integration near MARs. Using the NCBI mouse or human genome database, integration events were first grouped as to whether they occurred within or between genes. For each of the groups, the percentage of integrations that occurred within 2-kb of a predicted MAR was then determined (Figure 4). The results indicated no relationship to the nearest MAR when integration occurred within genes for SL3-3 at single insertion sites, MLV, MoFe2 or HTLV-1. In contrast, 71.4% of HIV-1 integrations that occurred within genes were observed to occur within 2-kb of a MAR. A strong relationship to MARs was also observed for SL3-3 integrations at CISs that occurred between genes. Of these integrations, 68.8% were observed to occur within 2-kb of a predicted MAR.
Figure 4 Analysis of the relationship between integration near a MAR and integration within or between genes. The percentage of integrations that occurred within 2-kb of a MAR is reported for those that occurred within a gene or between genes. Data are reported for SL3-3 insertions at single insertion site (SL3-3 IS), SL3-3 insertions at common insertion sites (SL3-3 CIS), MLV, MoFe2, HTLV-1 and HIV-1.
Conclusion
Evidence is accumulating to indicate that proviral integration is not random, and that the secondary structure of DNA plays a major role in integration site selection [2-18]. In the present study, the integration patterns of three different groups of retroviruses with distinct mechanisms of disease induction were analyzed to determine if integration near MARs is a common mechanism of retroviral integration site selection. The results indicated that gammaretroviruses (MLV and SL3-3), lentivirus (HIV-1) and deltaretrovirus (HTLV-1) integrate preferentially near MARs, specifically within 2-kb (Figure 2). These results suggest that integration near MARs is a common mechanism of retroviral integration site selection. The findings are consistent with the previous identification of preferred integration sites that contained sequence motifs such as DNaseI hypersensitive sites [3-7], AT-rich regions [8], transcriptionally active regions [2,9-12], and regions of DNA bending, specifically regions with the most DNA distortion [14-18], all of which are motifs shared by MARs. A recent study analyzed the proximity of retroviral integration to MARs when the virus was delivered to the cell by infection or by electroporation of naked DNA [49]. The results showed a strong correlation for integration near MARs during infection, but not when transfected as naked DNA. These results further support a role for MARs in integration site selection during retroviral infection. There are several possible explanations for preferential integration near MARs. One possibility is that MARs, due to their position at the bases of chromatin loops, are likely to be the first region of the DNA encountered by the provirus when entering the nucleus. A second possibility is that MARs may represent the most accessible regions for integration in the DNA due to the open confirmation and high propensity for base-unpairing associated with the AT-richness. A third possibility relates to the observation that retroviruses may contain their own MARs. In fact, the mouse mammary tumor virus (MMTV) has been shown to contain a MAR in the LTR that binds a well characterized MAR-binding protein, SATB1 [50]. As the proviral pre-integration complex enters the nucleus, MAR binding proteins may bind and direct integration due to their affinity for binding to cellular MARs. It is known that sequence insertion within or near a MAR results in greatly reduced binding to the nuclear matrix [45]. In contrast, it has been shown that when retroviral integration occurs near MARs, contact with the nuclear matrix is maintained, suggesting that the presence of a MAR in the viral genome may stabilize the contact between the chromosomal MAR and the nuclear matrix [49].
The selective advantage of integration near MARs may be that it positions the provirus in close proximity to transcription, RNA processing and transport machinery that is localized at the nuclear matrix (reviewed in [25]), thus activating expression from the viral promoter. In addition, our findings suggest the possibility that integration near MARs may have a role in malignant induction, specifically by gammaretroviruses. SL3-3 proviruses integrated at CISs in tumor DNA were shown to position preferentially within 2-kb upstream from a MAR, whereas SL3-3 proviruses integrated at single insertion sites in the same tumors did not show the same preference (Figure 3). Considering that gammaretroviruses like SL3-3 induce malignancy through insertional activation of oncogenes at CISs, this observation suggests that SL3-3 integration immediately upstream of MARs may be associated with activation of adjacent cellular gene expression. Such an effect might occur by disruption of the normal function of the MAR, thus altering local chromatin conformation. Changes in chromatin conformation, leading to changes in gene expression, are known to contribute to malignancy (reviewed in [51]). Alternatively, integration at a specific distance and orientation with respect to a MAR may result in stimulation of expression from the viral promoter, thus enhancing virus-mediated activation of an adjacent cellular oncogene. Integration near MARs has also been implicated in malignant induction by small DNA tumor viruses [34]. These viruses do not induce disease by insertional activation; thus, the advantage of integration near MARs may relate to increased expression from the viral promoter.
Previous studies have reported that HIV-1, MLV, ASV and HTLV-1 prefer to integrate into genes [11,12,19,20]. In the present study, integration patterns of SL3-3 and MoFe2 were examined to determine if they also preferentially integrate into RefSeqs. Consistent with previous reports, our results indicated that SL3-3 proviruses at single insertion sites (40.3%) and MoFe2 proviruses (33.3%) integrate preferentially within RefSeqs as compared to the predicted frequency for random integrations (22%). SL3-3 proviruses integrated at CISs did not demonstrate the same preference, an observation consistent with the role of these integrants in enhancer-mediated activation of an adjacent oncogene. Of SL3-3 integrations at CISs that occurred between genes, 68.8% were observed within 2-kb of a predicted MAR (Figure 4). Taken together, these studies provide additional evidence that proviral integration is not random, that MARs influence retroviral integration site selection, and that integration near MARs may have a role in the insertional activation of oncogenes by gammaretroviruses. Understanding the pressures that influence retroviral integration site selection is critical for further knowledge of the mechanisms of retroviral pathogenesis and for the development of retroviral vectors for gene-therapy.
Methods
Isolation of MoFe2-MuLV host-virus junction fragments
MoFe2 proviral integrations were analyzed from lymphomas induced in a previous study [40] and from acutely infected tissue culture cells. For that purpose, 5 × 105 SC-1 murine fibroblasts at 25% confluence were infected with 105 infectious units (TCID50) of MoFe2 in the presence of 8 μg/ml of polybrene for 5 hours. Medium was removed, replaced with fresh EMEM with 10% FBS, and cells were harvested three days later. Genomic DNA was digested with DraI (TTT/AAA) or StuI (AGG/CCT), and libraries were constructed using Universal Genome Walker Kit (BD Biosciences) as described by the manufacturer. Libraries were constructed from both restriction enzyme digests to avoid introducing a bias for AT- or GC-rich sequences. Host-virus junction sequences were amplified by PCR using oligonucleotide primers and Universal Genome Walker Kit reagents as previously described [40]. Amplification products were cloned into TOPO-TA vector (Invitrogen Corp.) and submitted for automated sequence analysis. The resulting sequences were considered to represent valid MoFe2 integrations if they contained the viral 3' LTR and if the immediately flanking host sequence had a ≥95% identity to a single genomic locus.
MAR analysis
A MAR prediction program termed MAR-Finder was used to predict MARs on 60-kb intervals surrounding the insertion site using default detection and clipping parameters for SL3-3 (n = 86), MoFe2 (n = 42), MLV (n = 49), HIV-1 (n = 41) and HTLV-1 (n = 26) [34,41,42]. High scoring regions were considered valid if the average strength of a single peak representing a predicted MAR was >0.65 [34].
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
CNJ performed all experimental and computer-based analyses. LSL directed the experimental design, implementation and interpretation of data. Both authors read and approved the final manuscript.
Acknowledgements
This work was supported by PHS grant CA83823, by Development Funds of the Tulane Cancer Center and by a grant from the Ladies Leukemia League. CNJ was supported in part by a grant from the Cancer Association of Greater New Orleans.
==== Refs
Neil JC Cameron ER Retroviral insertion sites and cancer: fountain of all knowledge? Cancer Cell 2002 2 253 255 12398888 10.1016/S1535-6108(02)00158-7
Weidhaas JB Angelichio EL Fenner S Coffin JM Relationship between retroviral DNA integration and gene expression J Virol 2000 74 8382 8389 10954538 10.1128/JVI.74.18.8382-8389.2000
Robinson HL Gagnon GC Patterns of proviral insertion and deletion in avian leukosis virus-induced lymphomas J Virol 1986 57 28 36 3001351
Rohdewohld H Weiher H Reik W Jaenisch R Breindl M Retrovirus integration and chromatin structure: Moloney murine leukemia proviral integration sites map near DNase I-hypersensitive sites J Virol 1987 61 336 343 3027365
Vijaya S Steffen DL Robinson HL Acceptor sites for retroviral integrations map near DNase I-hypersensitive sites in chromatin J Virol 1986 60 683 692 3490582
Goodenow MM Hayward WS 5' long terminal repeats of myc-associated proviruses appear structurally intact but are functionally impaired in tumors induced by avian leukosis viruses J Virol 1987 61 2489 2498 3037111
Panet A Cedar H Selective degradation of integrated murine leukemia proviral DNA by deoxyribonucleases Cell 1977 11 933 940 560916 10.1016/0092-8674(77)90304-X
Leclercq I Mortreux F Gabet AS Jonsson CB Wattel E Basis of HTLV type 1 target site selection AIDS Res Hum Retroviruses 2000 16 1653 1659 11080806 10.1089/08892220050193100
Scherdin U Rhodes K Breindl M Transcriptionally active genome regions are preferred targets for retrovirus integration J Virol 1990 64 907 912 2296087
Mooslehner K Karls U Harbers K Retroviral integration sites in transgenic Mov mice frequently map in the vicinity of transcribed DNA regions J Virol 1990 64 3056 3058 2335826
Narezkina A Taganov KD Litwin S Stoyanova R Hayashi J Seeger C Skalka AM Katz RA Genome-wide analyses of avian sarcoma virus integration sites J Virol 2004 78 11656 11663 15479807 10.1128/JVI.78.21.11656-11663.2004
Ozawa T Itoyama T Sadamori N Yamada Y Hata T Tomonaga M Isobe M Rapid isolation of viral integration site reveals frequent integration of HTLV-1 into expressed loci J Hum Genet 2004 49 154 165 14991527 10.1007/s10038-004-0126-7
Stevens SW Griffith JD Human immunodeficiency virus type 1 may preferentially integrate into chromatin occupied by L1Hs repetitive elements Proc Natl Acad Sci U S A 1994 91 5557 5561 8202527
Pruss D Reeves R Bushman FD Wolffe AP The influence of DNA and nucleosome structure on integration events directed by HIV integrase J Biol Chem 1994 269 25031 25041 7929189
Pruss D Bushman FD Wolffe AP Human immunodeficiency virus integrase directs integration to sites of severe DNA distortion within the nucleosome core Proc Natl Acad Sci U S A 1994 91 5913 5917 8016088
Pryciak PM Varmus HE Nucleosomes, DNA-binding proteins, and DNA sequence modulate retroviral integration target site selection Cell 1992 69 769 780 1317268 10.1016/0092-8674(92)90289-O
Pryciak PM Sil A Varmus HE Retroviral integration into minichromosomes in vitro Embo J 1992 11 291 303 1310932
Muller HP Varmus HE DNA bending creates favored sites for retroviral integration: an explanation for preferred insertion sites in nucleosomes Embo J 1994 13 4704 4714 7925312
Wu X Li Y Crise B Burgess SM Transcription start regions in the human genome are favored targets for MLV integration Science 2003 300 1749 1751 12805549 10.1126/science.1083413
Schroder AR Shinn P Chen H Berry C Ecker JR Bushman F HIV-1 integration in the human genome favors active genes and local hotspots Cell 2002 110 521 529 12202041 10.1016/S0092-8674(02)00864-4
Cereseto A Giacca M Integration site selection by retroviruses AIDS Rev 2004 6 13 21 15168737
Wu X Li Y Crise B Burgess SM Munroe DJ Weak palindromic consensus sequences are a common feature found at the integration target sites of many retroviruses J Virol 2005 79 5211 5214 15795304 10.1128/JVI.79.8.5211-5214.2005
Wu X Burgess SM Integration target site selection for retroviruses and transposable elements Cell Mol Life Sci 2004 61 2588 2596 15526164 10.1007/s00018-004-4206-9
Holman AG Coffin JM Symmetrical base preferences surrounding HIV-1, avian sarcoma/leukosis virus, and murine leukemia virus integration sites Proc Natl Acad Sci U S A 2005 102 6103 6107 15802467 10.1073/pnas.0501646102
Holth LT Chadee DN Spencer VA Samuel SK Safneck JR Davie JR Chromatin, nuclear matrix and the cytoskeleton: role of cell structure in neoplastic transformation (review) Int J Oncol 1998 13 827 837 9735414
Bode J Kohwi Y Dickinson L Joh T Klehr D Mielke C Kohwi-Shigematsu T Biological significance of unwinding capability of nuclear matrix-associating DNAs Science 1992 255 195 197 1553545
Cockerill PN Garrard WT Chromosomal loop anchorage sites appear to be evolutionarily conserved FEBS Lett 1986 204 5 7 3091398 10.1016/0014-5793(86)81377-1
Yanagisawa J Ando J Nakayama J Kohwi Y Kohwi-Shigematsu T A matrix attachment region (MAR)-binding activity due to a p114 kilodalton protein is found only in human breast carcinomas and not in normal and benign breast disease tissues Cancer Res 1996 56 457 462 8564952
Leman ES Getzenberg RH Nuclear matrix proteins as biomarkers in prostate cancer J Cell Biochem 2002 86 213 223 12111991 10.1002/jcb.10218
Coffey DS Nuclear matrix proteins as proteomic markers of preneoplastic and cancer lesions : commentary re: G. Brunagel et al., nuclear matrix protein alterations associated with colon cancer metastasis to the liver. Clin. Cancer Res., 8: 3039-3045, 2002 Clin Cancer Res 2002 8 3031 3033 12374668
Kurre P Morris J Thomasson B Kohn DB Kiem HP Scaffold attachment region-containing retrovirus vectors improve long-term proviral expression after transplantation of GFP-modified CD34+ baboon repopulating cells Blood 2003 102 3117 3119 12869503 10.1182/blood-2003-03-0962
Agarwal M Austin TW Morel F Chen J Bohnlein E Plavec I Scaffold attachment region-mediated enhancement of retroviral vector expression in primary T cells J Virol 1998 72 3720 3728 9557654
Lutzko C Senadheera D Skelton D Petersen D Kohn DB Lentivirus vectors incorporating the immunoglobulin heavy chain enhancer and matrix attachment regions provide position-independent expression in B lymphocytes J Virol 2003 77 7341 7351 12805432 10.1128/JVI.77.13.7341-7351.2003
Shera KA Shera CA McDougall JK Small tumor virus genomes are integrated near nuclear matrix attachment regions in transformed cells J Virol 2001 75 12339 12346 11711624 10.1128/JVI.75.24.12339-12346.2001
Kulkarni A Pavithra L Rampalli S Mogare D Babu K Shiekh G Ghosh S Chattopadhyay S HIV-1 integration sites are flanked by potential MARs that alone can act as promoters Biochem Biophys Res Commun 2004 322 672 677 15325282 10.1016/j.bbrc.2004.07.170
Hansen GM Skapura D Justice MJ Genetic profile of insertion mutations in mouse leukemias and lymphomas Genome Res 2000 10 237 243 10673281 10.1101/gr.10.2.237
Li J Shen H Himmel KL Dupuy AJ Largaespada DA Nakamura T Shaughnessy JDJ Jenkins NA Copeland NG Leukaemia disease genes: large-scale cloning and pathway predictions Nat Genet 1999 23 348 353 10610183 10.1038/14349
Mikkers H Allen J Knipscheer P Romeijn L Hart A Vink E Berns A Romeyn L High-throughput retroviral tagging to identify components of specific signaling pathways in cancer Nat Genet 2002 32 153 159 12185366 10.1038/ng950
Starkey CR Lobelle-Rich PA Granger S Brightman BK Fan H Levy LS Tumorigenic potential of a recombinant retrovirus containing sequences from Moloney murine leukemia virus and feline leukemia virus J Virol 1998 72 1078 1084 9445002
Johnson C Lobelle-Rich PA Puetter A Levy LS Substitution of feline leukemia virus long terminal repeat sequences into murine leukemia virus alters the pattern of insertional activation and identifies new common insertion sites J Virol 2005 79 57 66 15596801 10.1128/JVI.79.1.57-66.2005
Glazko GV Rogozin IB Glazkov MV Comparative study and prediction of DNA fragments associated with various elements of the nuclear matrix Biochim Biophys Acta 2001 1517 351 364 11342213
Rogozin IB Glazko GV Glazkov MV Computer prediction of sites associated with various elements of the nuclear matrix Brief Bioinform 2000 1 33 44 11466972
Benham C Kohwi-Shigematsu T Bode J Stress-induced duplex DNA destabilization in scaffold/matrix attachment regions J Mol Biol 1997 274 181 196 9398526 10.1006/jmbi.1997.1385
Albrecht P Bode J Buiting K Prashanth AK Lohmann DR Recurrent deletion of a region containing exon 24 of the RB1 gene caused by non-homologous recombination between a LINE-1HS and MER21B element J Med Genet 2004 41 e122 15591264 10.1136/jmg.2004.021923
Goetze S Gluch A Benham C Bode J Computational and in vitro analysis of destabilized DNA regions in the interferon gene cluster: potential of predicting functional gene domains Biochemistry 2003 42 154 166 12515550 10.1021/bi026496+
Goetze S Baer A Winkelmann S Nehlsen K Seibler J Maass K Bode J Performance of genomic bordering elements at predefined genomic loci Mol Cell Biol 2005 25 2260 2272 15743822 10.1128/MCB.25.6.2260-2272.2005
Kim R Trubetskoy A Suzuki T Jenkins NA Copeland NG Lenz J Genome-based identification of cancer genes by proviral tagging in mouse retrovirus-induced T-cell lymphomas J Virol 2003 77 2056 2062 12525640 10.1128/JVI.77.3.2056-2062.2003
Klehr D Maass K Bode J Scaffold-attached regions from the human interferon beta domain can be used to enhance the stable expression of genes under the control of various promoters Biochemistry 1991 30 1264 1270 1991107 10.1021/bi00219a015
Goetze S Huesemann Y Baer A Bode J Functional characterization of transgene integration patterns by halo fluorescence in situ hybridization: electroporation versus retroviral infection Biochemistry 2003 42 7035 7043 12795598 10.1021/bi0340907
Liu J Bramblett D Zhu Q Lozano M Kobayashi R Ross SR Dudley JP The matrix attachment region-binding protein SATB1 participates in negative regulation of tissue-specific gene expression Mol Cell Biol 1997 17 5275 5287 9271405
Zink D Fischer AH Nickerson JA Nuclear structure in cancer cells Nat Rev Cancer 2004 4 677 687 15343274 10.1038/nrc1430
|
16111492
|
PMC1198263
|
CC BY
|
2021-01-04 16:39:00
|
no
|
Virol J. 2005 Aug 19; 2:68
|
utf-8
|
Virol J
| 2,005 |
10.1186/1743-422X-2-68
|
oa_comm
|
==== Front
BMC Nucl MedBMC Nuclear Medicine1471-2385BioMed Central London 1471-2385-5-41604864810.1186/1471-2385-5-4Research ArticlePlanar Tc99m – sestamibi scintimammography should be considered cautiously in the axillary evaluation of breast cancer protocols: Results of an international multicenter trial Massardo Teresa [email protected] Omar [email protected] Augusto [email protected] Levin [email protected] Uma [email protected] Rossana [email protected] Lucía [email protected] Ajit K [email protected] Nuclear Medicine, University of Chile Clinical Hospital, Santiago, Chile2 Nuclear Medicine Centre and Medical Oncology Department, Hospital de Clínicas, University of La República, Montevideo, Uruguay3 Nuclear Medicine Department, National Cancer Institute, Bogotá, Colombia4 Nuclear Medicine Department, Cerrahpasa Medical Faculty, Istanbul University, Turkey5 Nuclear Medicine Department, Indraprastha Apollo Hospitals, New Delhi, India6 Department of Nuclear Medicine, Neoplastic Disease Institute and Peruvian Institute of Nuclear Energy, Lima, Peru7 Medicine Section, Department of Human Health, International Atomic Energy Agency, Vienna, Austria2005 27 7 2005 5 4 4 29 12 2004 27 7 2005 Copyright © 2005 Massardo et al; licensee BioMed Central Ltd.2005Massardo et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Lymph node status is the most important prognostic indicator in breast cancer in recently diagnosed primary lesion. As a part of an interregional protocol using scintimammography with Tc99m compounds, the value of planar Tc99m sestamibi scanning for axillary lymph node evaluation is presented. Since there is a wide range of reported values, a standardized protocol of planar imaging was performed.
Methods
One hundred and forty-nine female patients were included prospectively from different regions. Their mean age was 55.1 ± 11.9 years. Histological report was obtained from 2.987 excised lymph nodes from 150 axillas. An early planar chest image was obtained at 10 min in all patients and a delayed one in 95 patients, all images performed with 740–925 MBq dose of Tc99m sestamibi. Blind lecture of all axillary regions was interpreted by 2 independent observers considering any well defined focal area of increased uptake as an involved axilla. Diagnostic values, 95% confidence intervals [CI] and also likelihood ratios (LR) were calculated.
Results
Node histology demonstrated tumor involvement in 546 out of 2987 lymph nodes. Sestamibi was positive in 30 axillas (25 true-positive) and negative in 120 (only 55 true-negative). The sensitivity corresponded to 27.8% [CI = 18.9–38.2] and specificity to 91.7% [81.6–97.2]. The positive and negative LR were 3.33 and 0.79, respectively. There was no difference between early and delayed images. Sensitivity was higher in patients with palpable lesions.
Conclusion
This work confirmed that non tomographic Tc99m sestamibi scintimammography had a very low detection rate for axillary lymph node involvement and it should not be applied for clinical assessment of breast cancer.
==== Body
Background
Lymph node status is the most important prognostic indicator in breast cancer in recently diagnosed primary lesion. The evidence of metastatic involvement in the axilla requires the indication of adjuvant therapy posterior to surgical tumor resection. There is not an accurate anatomical test for this purpose and clinical examination has inappropriate diagnostic values. Routine lymph node dissection is the only accepted method for therapeutic decisions but it is invasive and produces significant associated morbidity such as lymphedema and, eventually, infections. On the other hand, an important proportion of breast cancer patients are node-negative. Ultrasonography has also been reported as helpful, especially adding fine needle aspiration biopsy [1,2].
The role of nuclear techniques is controversial in the area related with breast cancer [3,4]. Positron emission tomography (PET) with fluorine deoxyglucose (FDG) is an excellent method for breast cancer evaluation even though is not easily available; it is used for diagnosis and surgical planning, staging and restaging of local regional recurrence or distant metastases and also for monitoring therapy response. Its value for detecting axillary involvement is somehow debated and it has not been used in routine practice in all centers, due to its current resolution for that purpose. However, it appears to be very helpful in internal mammary node evaluation [5-8].
Sentinel node detection with radioguided biopsy has a well defined role in early staging of breast cancer and small tumors. This technique allows the recognition of lymphatic spreading. It requires nodal histology to decide complete posterior lymphadenectomy. The strategy involves diverse methodologies, is technically challenging, and requires a learning curve [9-12].
Scintimammography is widely available and its diagnostic value in axillary detection is not optimal when using planar images with 99mTc-sestamibi or phosphonates. However, reports using single photon emission tomography (SPECT) images with sestamibi and tetrofosmine labeled with Tc99m have better figures and even pinhole SPECT appears promising.
The aim of the present report was to evaluate through an unbiased standardized method the diagnostic value of planar sestamibi images for axillary involvement in breast cancer patients. This was accomplished in the scope of a multicenter interregional trial evaluating Tc99m compounds for scintimammography in breast cancer evaluation [13,14].
Methods
Population
This prospective study included 149 female patients ranging from 29 to 82 years (mean ± SD: 55,1 ± 11.9), from a multicenter trial on scintimammography Tc99m radiopharmaceuticals co-ordinated by the International Atomic Energy Agency (IAEA). Sixty per cent were postmenopausal. All patients had confirmed breast carcinoma (one patient had bilateral lesions). Only 50 patients (33.3%) presented also with axillary palpable nodes.
Primary breast tumour histology is documented in Table 1.
Table 1 Breast tumor histology
Biopsy diagnosis Number of cases
Ductal invasive carcinoma 123
Lobular invasive carcinoma 14
Colloidal carcinoma 5
Tubular carcinoma 3
Carcinoma in situ 3
Medullary carcinoma 1
Sarcoma 1
Total 150
The median size breast lesion was 25 mm ranging from 7 – 80 mm (mean ± SD = 28.8 ± 13.9 mm).
Scintimammography was performed before the histopathological confirmation of the cancer. Cases with fine-needle aspiration as the only confirmatory procedure were excluded. Axillary lymph node dissection in 150 axillary beds was performed as a part of the standard staging.
All patients included in this group provided written informed consent according to their local institutions at participating centres (Chile, China, Colombia, Cuba, Greece, India, Peru, Turkey and Uruguay).
Tc-99m scintimammography protocol
The same protocol was used in all centres. The radiochemical purity of Tc-99m-MIBI was ≥95%. Patients were injected with a bolus of 740 -925 MBq of sestamibi into an antecubital vein in the contra-lateral arm to the breast lesion or in a pedal vein in the patient with bilateral lesions. A plastic cannula was used to avoid interstitial infiltration and the injection was followed by a saline flush.
The acquisition began 10 min post injection with the patient supine. Imaging parameters were: matrix 256 × 256, peak energy of 140 ± 10% KeV, high-resolution low-energy collimator. The breast-collimator distance was kept to a minimum and a static 10 min image was always acquired. Anterior thoracic images included the neck, both axillas and breasts (with arms up). Lateral views were obtained with the patients in prone position using a commercially available breast pad set, (Pinestar Technology, Inc. Greenville, PA, USA), allowing the organ to hang freely, compressing the contra lateral breast. Delayed images were also obtained 90 min post injection in 95 patients using the same protocol. The gamma cameras were standard for clinical practice, including GE Starcam o, Elscint Apex, Siemens Diacam, and Sopha Sophy. Standardized contrasted images in gray scale were recorded.
Data analysis
All scintimammograms were interpreted by two experienced nuclear medicine physicians, blinded to clinical status of the patients as well as to all other tests results. The readers decided if the scan was positive or negative for lymph node involvement in both axillas. One or more focal areas of increased sestamibi uptake was considered positive. Their number was also consigned. The injection site was available for the observers only when a false positive interpretation was suspected due to radiopharmaceutical retention in a lymph or venous vessel.
Lymph node histology was considered as the gold standard. Results were incorporated to Arcus Quickstat and Instat data set for analysis.
Diagnostic values with a 95% confidence interval [CI] and Likelihood Ratios (LR) were calculated. Student t test was applied.
Results
One-hundred and fifty axillary lymph node dissections were performed in the 149 patients. Malignant involvement was reported in 89 out of 149 patients, (90 axillas). A total of 2987 lymph nodes were removed with a range of 4–47 nodes per patient (mean ± SD: 19.9 ± 9.7). Of these 2987 nodes, 546 presented histological tumoral status.
Sestamibi scintimammography was positive in 30 axillas (25 of them true-positive) and negative in 120 (55 true-negative). Thus, the sensitivity corresponded to 27.8% [CI = 18.9–38.2] and specificity to 91.7% [CI = 81.6–97.2]. The positive and negative LR were 3.33 and 0.79, respectively.
Two thirds of the axillas with single node involvement were false-negative (12 cases). When multi-nodal involvement was present, 31 cases with 2–5 nodes were false negative as well as 14 cases with 6–10 nodes, and in cases with more than 11 nodes involved, 8 cases were false-negative. There was a trend to lower sensitivity in the axillas with less than 5 nodes involved: 13.8 % versus 32.4% (p:0.47). With the currently used cut-off of 3 nodes involved, 53% of the false-negatives axillas were equal or under that number.
The only five false-positives corresponded to reactive lymphadenitis, follicular hyperplasia or were just specified as non-malignant.
The sensitivity of scintimammography in the group with palpable axillary nodes was significantly higher than in the non palpable group (p:0.036). They corresponded to 39.0% [CI = 8.8–32] versus 18.4% [CI = 24.2–55.5]. Specificities were 100% [CI = 66.4–100] versus 90,2% [CI = 78.6–92.7]; positive LR was 3.9 versus 1.87 and negative LR 0.61 versus 0.91, respectively. See Figure 1.
Figure 1 Tc99m sestamibi performance according to axillary status.
There was no difference between early and delayed diagnostic values in the 95 patients with both exams performed in identical conditions (p:0.65). See Figure 2.
Figure 2 Comparison between axillary results in early and delayed Tc99m sestamibi in 95 patients.
Discussion
These results support that planar imaging with scintimammography and Tc99m- sestamibi should be definitively excluded or considered cautiously for axillary evaluation protocols in breast cancer.
Different techniques for axillary evaluation
Yutani et al. [15] in their comparative study between FDG PET and sestamibi-SPECT reported sensitivities of 50.0 and 37.5%, respectively, for axillary detection in 40 consecutive patients with head to head comparison. Their results with tomographic images are relatively concordant with ours. However, in this setting, theirs and our sensitivity values were disappointingly low and are clearly opposed to several prior reports with either planar or SPECT techniques (See Table 2; [15-28]). This could be explained by the size and depth of the lesions, their relative low uptake and especially by the equipment resolution. Our lower detection rate compared with other reports may be explained, in part, by the method of robust blind reading with no interpretation bias.
Table 2 Scintimammography results according to number of axillary nodes involved
N° involved nodes/axilla N° axillas False-Negative N° axillas True-Positive
1 12 6
2–5 31 *
6–10 14 *
11–20 4 *
>20 4 *
Total 65 25
* Individual group data is not available (2 nodes or more subgroups = 19 nodes)
It is interesting to mention that sestamibi is helpful for the diagnosis of melanoma lymph node assessment [29], contrary to the observed situation discussed in breast cancer. The reason for this fact could be the most superficial and somehow easier to locate melanomatous involved nodes. The nodes in axillas are deeply positioned which can probably contribute to the lower sestamibi uptake in breast cancer.
PET FDG has been proposed in order to reduce the proportion of patients requiring axillary dissection with variable results, but until now the technique cannot adequately assess the number of nodes involved. However, it could be very helpful in the evaluation of internal mammary chain in upper medial quadrant primary tumours, as well as in patients with large lesions. According to Danforth et al. [30] in 495 patients its global sensitivity for axillary involvement was 89% [95%CI = 86–92], with a specificity of 87% [95%CI = 84–90]. Yutani et al. [15] reported that FDG is sufficiently sensitive to rule out lymph node metastasis. Greco et al. [5] reported in 167 patients FDG sensitivity of 94%, specificity 86% and accuracy of 90% for axillary evaluation.
We agree with other authors [15,23] who have published that planar scintimammography is not recommended for axillary evaluation. Tolmos et al. [20] do not consider the test as reliable (they observed a kappa value of 0.49 for interobserver agreement). Even though, there are posterior and recent publications with new results still reporting relatively good values [17,25-28]. Limachi et al. [27] reported lower sensitivity if fewer nodes were affected, similar to our findings (in patients with <3 metastases, sensitivity was 69.7%, and only one out of six patients with a single lesion had a positive scan). See Table 3.
Table 3 Diagnostic value of the published literature (PUBMED) in breast cancer axillary lymph node evaluation using Tc99m sestamibi.
Author Sensitivity (%) Specificity (%) N° of patients Ref.N°
Lam et al. Eur J Nucl Med, 1996 64 90 31 16
Cistaro et al. Minerva Chir, 1997 75 90 45 17
Schillaci et al. Anticancer Res, 1997 61.9 81 * 96.4 92.9 * 49 18
Akcay et al. Clin Nucl Med, 1997 66 100 30 19
Tolmos et al. Am Surg, 1997 75 82 31 20
Perre et al. Eur J Surg Oncol, 1997 91 64 36 21
Taillefer et al. J Nucl Med, 1998 79.2 84.6 100 22
Danielsson et al. Acta Radiol, 1999 67 80 58 23
Arslan et al. Nucl Med Commun, 1999 68 93 77 24
Mulero et al. Rev Esp Med Nucl, 2000 36 100 84 25
Yutani et al. J Comput Assist Tomography, 2000 38* NA 40 15
Nishiyama et al. Eur J Nucl Med, 2001 73 NA 50 26
Lumachi et al. Eur J Surg Oncol, 2001 82.3 94.1 239 27
Chen et al. Chin Med J, 2003 83.3 86.1 60 28
IAEA group 28 92 149
NA : Not available
* : SPECT
Other compounds labeled with Tc99m
Regarding data with other compounds labeled with Tc99m, commonly used, especially tetrofosmin also a cationic lipophilic molecule, the values are similar to sestamibi in breast cancer evaluation [19,31]. Akcay [19] found comparable diagnostic value for both in a small number of patients with involved axillary nodes. The experience with SPECT is significantly better including small primary breast tumours [32]. Tc99m diphosphonates (MDP) proposed as an interesting alternative as well as pentavalent DMSA, have less diagnostic value than sestamibi for breast primary lesions and also for axillary node evaluation, according to our group results and others [13,26].
The addition of P-SPECT
Madeddu and Spanu, using tetrofosmin, proposed recently SPECT with pinhole (P-SPECT) as the best technique to evaluate the axilla. Their group demonstrated that P-SPECT has better sensitivity compared to SPECT and they, individually, were superior to planar imaging, even for non palpable axillary lesions [33-35]. Their group previously reported also that tetrofosmin SPECT has better sensitivity than planar scintimammography for palpable and non palpable axillary lesions [36]. When P-SPECT was performed with sentinel node detection both techniques combined gave 100% accuracy and P-SPECT was able to identify 81.2% of cases with a single node, and correctly classified 93.7% of the patients with ≤ or > 3 metastatic nodes [37].
Other interesting points
It has been reported that sestamibi and FDG are related with low radiopharmaceutical uptake in early forms of breast carcinoma that make tumoral detection more difficult in certain cancer subtypes, such as invasive lobular carcinoma and low-grade tumors, even with locally advanced disease [38-40]. It appears that favorable response to neoadjuvant therapy, in locally advanced disease is complex due to tumoral flow and metabolic changes [41].
Finally, it should be considered that in women with a clinically negative axilla the information obtained from surgical dissection in order to decide adjuvant therapy is related to age and other factors, such as tumor characteristics [42]. SPECT equipment capacity should be ameliorated in order to improve the detection of smaller lesions in breast carcinoma, as was published with phantom models [43]. The recent and excellent review by Taillefer (44) regarding scintimammography suggested that it is necessary to define the clinical niches of the test. In axilla, the diagnostic accuracy of sestamibi varied between 80–85% (with an overall accuracy of 81% (411/509) for 12 reports including two with SPECT); for him, this value is still too low to advocate its use to avoid axillary node dissection in patients with proven invasive primary breast cancer.
Conclusion
There is strong information supporting that planar sestamibi data is not an adequate alternative for axillary evaluation in breast cancer. We believe that countries with limited resources regarding radiopharmaceuticals and equipment availability, should avoid the non-tomographic protocol.
List of abbreviations
CI: Confidence Interval
LR: Likelihood Ratio
PET: Positron Emission Tomography
FDG: Fluorine deoxyglucose-F18
SPECT: Single Photon Emission Tomography
P-SPECT: SPECT with pinhole
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
TM carried out nuclear studies and participated in the design and discussion, read blindly all the studies and redacted the final manuscript
OA carried out nuclear studies and participated in the design and discussion, read blindly all the studies performed the statistical analysis and reviewed the final manuscript
AL-0 carried out part of the studies and participate in the former redaction of the manuscript
LK carried out nuclear studies and participated in the design and discussion
UR carried out nuclear studies and participated in the design and discussion
RM carried out nuclear studies and participated in the design and discussion
LD analyzed the oncological data
AKP, as the chief of the IAEA research group, conceived the study, participated in the design and global coordination.
All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
To all physicians and nuclear medicine personnel involved in this IAEA project.
The study was supported by a grant form the International Atomic Energy Agency (IAEA) Coordinated Research Project E1.30.17
==== Refs
Bedrosian I Bedi D Kuerer HM Fornage BD Harker L Ross MI Ames FC Krishnamurthy S Edeiken-Monroe BS Meric F Feig BW Akins J Singletary SE Mirza NQ Hunt KK Impact of clinicopathological factors on sensitivity of axillary ultrasonography in the detection of axillary nodal metastases in patients with breast cancer Ann Surg Oncol 2003 10 1025 30 14597440 10.1245/ASO.2003.12.017
Sapino A Cassoni P Zanon E Fraire F Croce S Coluccia C Donadio M Bussolati G Ultrasonographically-guided fine-needle aspiration of axillary lymph nodes: role in breast cancer management Br J Cancer 2003 88 702 706 12618878 10.1038/sj.bjc.6600744
Schillaci O Buscombe JR Breast scintigraphy today: indications and limitations Eur J Nucl Med Mol Imaging 2004 31 S35 45 15103505 10.1007/s00259-004-1525-x
Buscombe JR Holloway B Roche N Bombardieri E Position of nuclear medicine modalities in the diagnostic work-up of breast cancer Q J Nucl Med Mol Imaging 2004 48 109 18 15243412
Greco M Crippa F Agresti R Seregni E Gerali A Giovanazzi R Micheli A Asero S Ferraris C Gennaro M Bombardieri E Cascinelli N Axillary lymph node staging in breast cancer by 2-fluoro-2-deoxy-D-glucose-positron emission tomography: clinical evaluation and alternative management J Natl Cancer Inst 2001 93 630 5 11309439 10.1093/jnci/93.8.630
Crippa F Gerali A Alessi A Agresti R Bombardieri E FDG-PET for axillary lymph node staging in primary breast cancer Eur J Nucl Med Mol Imaging 2004 31 S97 102 15133635 10.1007/s00259-004-1531-z
Eubank WB Mankoff DA Takasugi J Vesselle H Eary JF Shanley TJ Gralow JR Charlop A Ellis GK Lindsley KL Austin-Seymour MM Funkhouser CP Livingston RB 18fluorodeoxyglucose positron emission tomography to detect mediastinal or internal mammary metastases in breast cancer J Clin Oncol 2001 19 3516 23 11481358
Bellon JR Livingston RB Eubank WB Gralow JR Ellis GK Dunnwald LK Mankoff DA Evaluation of the internal mammary lymph nodes by FDG-PET in locally advanced breast cancer (LABC) Am J Clin Oncol 2004 27 407 10 15289736 10.1097/01.coc.0000128869.19357.9b
Fehr MK Hornung R Varga Z Burger D Hess T Haller U Fink D von Schulthess GK Steinert HC Axillary staging using positron emission tomography in breast cancer patients qualifying for sentinel lymph node biopsy Breast J 2004 10 89 93 15009033 10.1111/j.1075-122X.2004.21455.x
Leidenius M Krogerus L Toivonen T Leppanen E von Smitten K The sensitivity of axillary staging when using sentinel node biopsy in breast cancer Eur J Surg Oncol 2003 29 849 53 14624776 10.1016/j.ejso.2003.09.014
Goyal A Douglas-Jones A Newcombe RG Mansel RE ALMANAC Trialists Group Predictors of non-sentinel lymph node metastasis in breast cancer patients Eur J Cancer 2004 40 1731 7 15251163 10.1016/j.ejca.2004.04.006
Van Der Vegt B Doting MH Jager PL Wesseling J De Vries J Axillary recurrence after sentinel lymph node biopsy Eur J Surg Oncol 2004 30 715 20 15296984 10.1016/j.ejso.2004.05.015
Massardo T Alonso O Kabasakal L Llamas-Olier A Shankar UR Zhu H Delgado L Gonzalez P Mut F Padhy AK Diagnostic value of 99mTc-methylene diphosphonate and 99mTc-pentavalent DMSA compared with 99mTc-sestamibi for palpable breast lesions J Nucl Med 2002 43 882 888 12097457
Alonso O Massardo T Delgado LB Horvath J Kabasakal L Llamas-Olier A Maunda KK Morales R Padhy AK Shankar UR Is (99m)Tc-sestamibi scintimammography complementary to conventional mammography for detecting breast cancer in patients with palpable masses? J Nucl Med 2001 42 1614 21 11696629
Yutani K Shiba E Kusuoka H Tatsumi M Uehara T Taguchi T Takai SI Nishimura T Comparison of FDG-PET with MIBI-SPECT in the detection of breast cancer and axillary lymph node metastasis J Comput Assist Tomogr 2000 24 274 80 10752892 10.1097/00004728-200003000-00017
Lam WW Yang WT Chan YL Stewart IE Metreweli C King W Detection of axillary lymph node metastases in breast carcinoma by technetium-99m sestamibi breast scintigraphy, ultrasound and conventional mammography Eur J Nucl Med 1996 23 498 503 8698052 10.1007/BF01249619
Cistaro A Bonazzi G Bello M Bessone M Tetti M Villata E Coluccia C Bisi G Massaioli N Breast carcinoma and the axillary lymph node status: the use of 99mTc-sestamibi scintimammography for the diagnosis and preoperative staging. Preliminary results Minerva Chir 1997 52 1177 82 9471568
Schillaci O Scopinaro F Danieli R Tavolaro R Cannas P Picardi V Colella AC Technetium-99m sestamibi imaging in the detection of axillary lymph node involvement in patients with breast cancer Anticancer Res 1997 17 1607 1610 9179202
Akcay MN Akin Y Karabag B Ozcan O Oren D Tc-99m tetrofosmin in breast carcinoma and axillary lymph node metastases: a comparative study with Tc-99m MIBI Clin Nucl Med 1997 22 832 4 9408644 10.1097/00003072-199712000-00005
Tolmos J Khalkhali I Vargas H Stuntz M Cutrone J Mishkin F Diggles L Venegas R Klein S Detection of axillary lymph node metastasis of breast carcinoma with technetium-99m sestamibi scintimammography Am Surg 1997 63 850 3 9322655
Perre CI Rutter JE Vos PA de Hooge P Technetium-99m-sestamibi uptake in axillary lymph node metastases in breast cancer patients Eur J Surg Oncol 1997 23 142 4 9158189 10.1016/S0748-7983(97)80008-X
Taillefer R Robidoux A Turpin S Lambert R Cantin J Leveille J Metastatic axillary lymph node technetium-99m-MIBI imaging in primary breast cancer J Nucl Med 1998 39 459 64 9529292
Danielsson R Bone B Perbeck L Aspelin P Evaluation of planar scintimammography with 99mTc-MIBI in the detection of axillary lymph node metastases of breast carcinoma Acta Radiol 1999 40 491 5 10485237
Arslan N Ozturk E Ilgan S Urhan M Karacalioglu O Pekcan M Tufan T Bayhan H 99mTc-MIBI scintimammography in the evaluation of breast lesions and axillary involvement: a comparison with mammography and histopathological diagnosis Nucl Med Commun 1999 20 317 25 10319351
Mulero F Nicolas F Castellon MI Claver MA Abad L Nuno de la Rosa JA Scintigraphy with 99mTc-MIBI in the diagnosis of axillary lymph node invasion of breast cancer Rev Esp Med Nucl 2000 19 416 22 11060271
Nishiyama Y Yamamoto Y Ono Y Irie A Yamauchi A Satoh K Ohkawa M Comparative evaluation of 99mTc-MIBI and 99mTc-HMDP scintimammography for the diagnosis of breast cancer and its axillary metastases Eur J Nucl Med 2001 28 522 8 11357504 10.1007/s002590000473
Lumachi F Ferretti G Povolato M Marzola MC Zucchetta P Geatti O Bui F Brandes AA Usefulness of 99m-Tc-sestamibi scintimammography in suspected breast cancer and in axillary lymph node metastases detection Eur J Surg Oncol 2001 27 256 9 11373101 10.1053/ejso.2000.1096
Chen J Wu H Zhou J Hu J Using Tc-99m MIBI scintimammography to differentiate nodular lesions in breast and detect axillary lymph node metastases from breast cancer Chin Med J (Engl) 2003 116 620 4 12875735
Alonso O Martinez M Delgado L Lago G Juri C Borges AL Lopes de Amorim MC de Boni D Espasandin J Priario J Comparison of 99mTc-MIBI scintigraphy and sentinel node biopsy in the detection of occult lymph node metastases from cutaneous melanoma Eur J Dermatol 2003 13 449 54 14693488
Danforth DN JrAloj L Carrasquillo JA Bacharach SL Chow C Zujewski J Whatley M Galen B Merino M Neumann RD The role of 18F-FDG-PET in the local/regional evaluation of women with breast cancer Breast Cancer Res Treat 2002 75 135 46 12243506 10.1023/A:1019664126220
Horne T Pappo I Cohen-Pour M Baumer M Orda R 99Tc(m)-tetrofosmin scintimammography for detecting breast cancer: a comparative study with 99Tc(m)-MIBI Nucl Med Commun 2001 22 807 11 11453054 10.1097/00006231-200107000-00012
Spanu A Schillaci O Meloni GB Porcu A Cottu P Nuvoli S Falchi A Chessa F Solinas ME Madeddu G The usefulness of 99mTc-tetrofosmin SPECT scintimammography in the detection of small size primary breast carcinomas Int J Oncol 2002 21 831 40 12239623
Madeddu G Spanu A Use of tomographic nuclear medicine procedures, SPECT and pinhole SPECT, with cationic lipophilic radiotracers for the evaluation of axillary lymph node status in breast cancer patients Eur J Nucl Med Mol Imaging 2004 31 S23 34 15085293 10.1007/s00259-004-1524-y
Spanu A Dettori G Chiaramida P Cottu P Falchi A Porcu A Solinas ME Nuvoli S Madeddu G The role of 99mTc-tetrofosmin Pinhole-SPECT in breast cancer axillary lymph node staging Cancer Biother Radiopharm 2000 15 81 91 10740656
Spanu A Tanda F Dettori G Manca A Chessa F Porcu A Falchi A Nuvoli S Madeddu G The role of (99m)Tc-tetrofosmin pinhole-SPECT in breast cancer non palpable axillary lymph node metastases detection Q J Nucl Med 2003 47 116 28 12865872
Spanu A Dettori G Nuvoli S Porcu A Falchi A Cottu P Solinas ME Scanu AM Chessa F Madeddu G (99)mTc-tetrofosmin SPET in the detection of both primary breast cancer and axillary lymph node metastasis Eur J Nucl Med 2001 28 1781 94 11734916 10.1007/s00259-001-0657-5
Spanu A Dettori G Chessa F Porcu A Cottu P Solinas P Falchi A Solinas ME Scanu AM Nuvoli S Madeddu G 99mTc-Tetrofosmin pinhole-SPECT (P-SPECT) and radioguided sentinel node (SN) biopsy and in breast cancer axillary lymph node staging Cancer Biother Radiopharm 2001 16 501 13 11789027 10.1089/10849780152752100
Mankoff D Dunnwald L Kinahan P Are We Ready for Dedicated Breast Imaging Approaches? J Nucl Med 2003 44 594 595 12679404
Avril N Dose J Janicke F Metabolic characterization of breast tumors with positron emission tomography using F-18 fluorodeoxyglucose J Clin Oncol 1996 14 1848 57 8656253
Mankoff DA Dunnwald LK Gralow JR Blood flow and metabolism in locally advanced breast cancer: relationship to response to therapy J Nucl Med 2002 43 500 9 11937594
Tseng J Dunnwald LK Schubert EK Link JM Minoshima S Muzi M Mankoff DA 18F-FDG Kinetics in Locally Advanced Breast Cancer: Correlation with Tumor Blood Flow and Changes in Response to Neoadjuvant Chemotherapy J Nucl Med 2004 45 1829 37 15534051
Chua B Ung O Taylor R Boyages J Is Information from Axillary Dissection Relevant to Patients with Clinically Node-Negative Breast Cancer? Breast J 2003 9 478 84 14616942 10.1046/j.1524-4741.2003.09607.x
Tornai MP Bowsher JE Jaszczak RJ Pieper BC Greer KL Hardenbergh PH Coleman RE Mammotomography with pinhole incomplete circular orbit SPECT J Nucl Med 2003 44 583 93 12679403
Taillefer R Clinical applications of 99mTc-sestamibi scintimammography Semin Nucl Med 2005 35 100 115 15765373 10.1053/j.semnuclmed.2004.11.002
|
16048648
|
PMC1199526
|
CC BY
|
2021-01-04 16:30:50
|
no
|
BMC Nucl Med. 2005 Jul 27; 5:4
|
utf-8
|
BMC Nucl Med
| 2,005 |
10.1186/1471-2385-5-4
|
oa_comm
|
==== Front
BMC PsychiatryBMC Psychiatry1471-244XBioMed Central London 1471-244X-5-301604580510.1186/1471-244X-5-30CorrespondenceInternet-based search of randomised trials relevant to mental health originating in the Arab world Takriti Yahya [email protected] Hany G [email protected] Clive E [email protected] Aire Court Community Unit, Lingwell Grove, Leeds, LS10 4BS, UK2 Academic Unit of Psychiatry and Behavioural Sciences, University of Leeds, 15 Hyde Terrace, Leeds, LS2 9LT, UK3 Cochrane Schizophrenia Group, Academic Unit of Psychiatry and Behavioural Sciences, University of Leeds, 15 Hyde Terrace, Leeds, LS2 9LT, UK2005 26 7 2005 5 30 30 22 3 2005 26 7 2005 Copyright © 2005 Takriti et al; licensee BioMed Central Ltd.2005Takriti et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The internet is becoming a widely used source of accessing medical research through various on-line databases. This instant access to information is of benefit to busy clinicians and service users around the world. The population of the Arab World is comparable to that of the United States, yet it is widely believed to have a greatly contrasting output of randomised controlled trials related to mental health. This study was designed to investigate the existence of such research in the Arab World and also to investigate the availability of this research on-line.
Methods
Survey of findings from three internet-based potential sources of randomised trials originating from the Arab world and relevant to mental health care.
Results
A manual search of an Arabic online current contents service identified 3 studies, MEDLINE, EMBASE, and PsycINFO searches identified only 1 study, and a manual search of a specifically indexed, study-based mental health database, PsiTri, revealed 27 trials.
Conclusion
There genuinely seem to be few trials from the Arab world and accessing these on-line was problematic. Replication of some studies that guide psychiatric/psychological practice in the Arab world would seem prudent.
==== Body
Background
The well-conducted randomised trial is the gold standard for the evaluation of medical interventions including mental health treatments [1]. This methodology was first formally used in the late 1940's [2] although much earlier examples do exist including one controlled experiment from the Middle East recorded in the Bible about 2nd-1st century BC (Daniel 1:1–16). At first randomised trials were few in number and easy to summarise but the exponential rise of these studies made it increasingly difficult to produce clear unbiased summaries of the best evidence on the effects of health care interventions [3]. Currently there are thousands of journals worldwide which publish the results of trials and many thousands of randomised trials reported each year. Clinicians, trialists, policy makers and patients cannot hope to keep up with the annual volume of literature so reviews are useful.
Frequently, reviewing is subjective with little or no attempt to ensure that the results are reproducible and as free from bias as possible [3]. More systematic reviews, however, endeavour to minimise bias and combine the results of all relevant trials in objective, explicit and reproducible ways [4]. This is exemplified by the work of The Cochrane Collaboration, an international, independent and not-for-profit organisation which disseminates systematic reviews of healthcare interventions worldwide [5]. This Collaboration recognises that including only easily-identified trials in reviews leaves work vulnerable to the inclusion of bias, as highly accessible trials tend to be more positive than those that are more difficult to find [6-8]. Cochrane reviewers therefore make strenuous efforts to find all relevant studies-a model that all researchers would do well to emulate when searching for relevant randomised controlled trials.
The Arab world is a culturally, religiously and ethnically diverse region consisting of 22 countries from Mauritania in the west, to Iraq in the east. Its rapidly expanding population is estimated at over 280 million. The media in this region reports little, except in the context of oil reserves, religious tensions and conflict. Despite the Arab world having a population comparable in size to that of the USA, its contribution to recent global medical literature is thought to be relatively small [9]. Reasons for this are probably many and complex, but may include military conflict and arms expenditure, the 'brain drain', research culture, trade embargoes and humanitarian crises [9,10].
Although a comparatively small proportion of randomised controlled trials are conducted outside of the relatively affluent West, there is evidence that other parts of the developing world are beginning to make progress. One particular study looked at randomised controlled trials originating from Sub-Saharan Africa, and showed that there may have been an increase in publication of these trials over time. The same article however, identified that there was a relative lag between the overall burden of mental illnesses on the continent and trials focussing on these conditions [11].
This paper describes a search for, and survey of, randomised controlled trials relevant to mental health published in the Arab literature or having clearly originated from one of the countries of the Arab world.
Methods
First we manually searched the ArabPsyNet database The function of this internet-based database, in part, is as a current awareness service for the contents of Arab psychiatric and psychological periodicals. The list of journals and the years for which contents pages are available are presented in Table 1. An initial search was conducted in December 2003 and repeated in May 2004. Articles published in Arabic, English and French could have been included. We recognise that publication in Arab World journals did not necessarily imply the Arabic descent of authors and, similarly, publication in an Arab world journal did not necessarily mean that the research was undertaken in the Arab world.
Table 1 Number of issues available to search electronically in Arab psychiatric journals and number of randomised controlled trials identified. Table modified from
Journal Country of publication Language Dates available Total number of. issues in available dates No. issues with contents pages available on this site No. RCTs identified on available content pages
Addiction Bulletin Egypt Arabic 1999–2000 8 4 0
Arabpsynet journal Tunisia Arabic & English 2004–2004 2 2 0
Arab Journal of Psychiatry Jordan English 1989–2003 28 18 2
Arab Psychologist Egypt English 2000–2001 2 2 0
Assaha Al Aklia Yemen Arabic 1999–2001 12 9 0
Bulletin of Egyptian Psychiatric Association Egypt Arabic 1999–2001 12 5 0
Current Psychiatry Egypt English 1994–1996 6 3 1
Egyptian Journal of Mental Health Egypt Arabic & English None 0 0 0
Egyptian Journal of Psychiatry Egypt English None 0 0 0
Egyptian Journal of Psychological Studies Egypt Arabic 1992–2000 36 1 0
Interdisciplinary Psychology Lebanon Arabic 1990–2004 58 54 0
Journal on Arab Children Kuwait Arabic 1998–2004 24 13 0
Man and Evolution Egypt Arabic 1984–2000 76 41 0
Mental Health Yemen Arabic 1992–2000 18 7 0
Mental Peace Journal of WIAMH Saudi Arabia Arabic 1994–2002 36 12 0
News Letter of the AFNGO for Drug Abuse Prevention Egypt Arabic 1999–2001 6 2 0
Psychological Quarterly Egypt Arabic 2000–2002 12 4 0
Psychology Egypt Arabic 1987–2002 64 22 0
Tunisian Annals of Psychiatry Tunisia French 1996–1997 4 2 0
Tunisian Journal of Psychiatry Tunisia French 1998–2001 8 1 0
WIAMH Newsletter Egypt English 1998–2003 38 10 0
Next, we manually searched EMBASE, MEDLINE and PsycINFO with " (Randomi* and (arab* or [names of individual countries]) and ([broad terms for mental disorder/illness]))".
Thirdly, we manually searched PsiTri for the name of every Arab country in the 'country of origin' field. PsiTri is a freely available based electronic database on published and unpublished controlled clinical trials, reporting on treatments and interventions for a wide range of conditions within the field of mental health. Uniquely, it is study-based, rather than simply consisting of lists of citations that could relate to the same study. It is also reliably indexed with country of origin.
Results
The search of ArabPsyNet identified only three randomised controlled trials which had been published in the 212 available issues from the 21 journals between1984 and 2004. If all issues had been available from all 21 journals we predict that three more studies would have been identified. All three trials were published in English (see Table 1). The first, in the Arab Journal of Psychiatry in 1989, reports a small randomised controlled trial of alprazolam versus cognitive therapy for outpatients with panic disorders [12]. The second, published in 1996, also in the Arab Journal of Psychiatry [13] reports another small randomised controlled trial, this time comparing zuclopenthixol acetate with haloperidol for people with schizophrenic psychoses, affective psychoses and paranoid states. Finally, the third trial was published in Current Psychiatry in 1996 and reports a randomised controlled trial of sertaline versus placebo for over 300 people with obsessive-compulsive disorder [14]. This final trial was not conducted in the Middle East, but was published in an Egyptian journal.
MEDLINE and EMBASE searches did not identify any of the three studies. However, one of the studies [13] was identified through PsycINFO. The majority of studies included on these three popular databases come from journals with a high citation index, all of which tend to be published in the Western world. In this way, the low citation rate of Arab journals is likely to be perpetuated as these databases are commonly used in searches.
PsiTri contained most randomised controlled trials relevant to this survey. Searching on country of origin several relevant articles published outside of Arab journals, some of which were not in psychiatric or psychological periodicals. In all, we identified 27 randomised controlled trials from eight Arab countries (see Table 2).
Table 2 Results from PsyTri searching by country of origin
Country in which trial was thought to be undertaken No. of randomised trials Journal of publication Subject Year
Algeria 1 European Psychiatry Depression [18] 1995
Bahrain 1 Arab Journal of Psychiatry Psychoses [13] 1996
Egypt 12 Journal of International Medical Depression [19] 1980
Research Depression [20] 1976
Journal of International Medical Impotence [21] 1992
Research Journal of Urology Affective disorders [22] 1983
International Journal of Nursing Depression [23] 1976
Studies Enuresis [24] 1999
Journal of the Egyptian Medical Enuresis [25] 1990
Association Pubertal
Urology development [26] 1969
Urology Neurotic Disorders;
American Journal of Clinical Schizophrenia;
Nutrition Depressive
Acta Psychiatrica Scandinavica Disorder; Mental Disorders [27] 1970
Journal of the European College of Neuropsychopharmacology Schizophrenia [28] 1999
British Journal of Psychiatry Schizophrenia [29] 1970
Iraq 2 Journal of Urology Enuresis [30] 1989
Clinical and Experimental Pharmacology & Physiology Enuresis [31] 1986
Lebanon 2 Pediatric Nursing Anxiety [32] 1998
International Journal of Geriatric Psychiatry Alzheimers [33] 2004
Libya 1 Gerontologia Clinica Senile restlessness [34] 1970
Kuwait 3 British Journal of Psychiatry ECT-induced cognitive
Journal of the Kuwait Medical impairement [35] 1985
Association Schizophrenia [36] 1981
Biological Psychiatry Chronic schizophrenia [37] 1999
Saudi Arabia 5 Journal of Clinical Depression [38] 1995
Pharmacology Anxiety [39] 1999
British Journal of Anaesthesia Dental anxiety [40] 1999
Journal of Dental Research Acta Psyhiatrica Scandinivica Psychiatric illness [41] 1997
Behavioural and Cognitive Psychotherapy Schizophrenia [42] 1997
Discussion
ArabPsyNet, the only mental-health specific database detailing 'table of contents' pages of journals rarely seen outside of Arabic-speaking world, is making a concerted effort to disseminate research from the region. As yet the coverage of journals is limited and varied- as is the information produced on each article. As this site develops, its contribution should become greater. Current searches suggest that trials from the Arab world are rarely published in these journals. The main-stream databases are also not good sources of mental health trials from this region and may make the possibility of 'index-bias' more likely when searching for trial of Arab World origin. It is possible that we failed to identify studies, as reporting of country of origin is variable. Further complexity is added to these searches of general databases by having to use a phrase covering the many ways mental health problems are indexed, as well as one for the many country's names in title, abstract, and address fields. The mental health specific database, PsiTri, avoided the need for this. Being study-based and specifically indexed for country of origin confers great advantage over the other two sources. This database, as a compilation of all registers of mental health Cochrane groups (which are themselves created with extensive searches of many diverse databases) is likely to be the best source in existence of clearly indexed mental health studies from the Arab world.
It is possible that we failed to identify relevant trials published in Arab journals. Many countries, including The Czech Republic, Hungary, India, Korea, Latin America, Russia, and the Ukraine (see Table 3), recognising that coverage of their home literature in databases such as MEDLINE will always be limited, produce their own medical bibliographic databases in which searching for randomised trials is relatively simple. We know of no such database in the Arab world, but such a project would make a most valuable contribution to dissemination of medical research from the area.
Table 3 Bibliographic and full text sources by country/region of origin.*
AFRICA [43]
CHINA [44]
CZECH REPUBLIC [45]
EASTERN MEDITERRANEAN [46]
EGYPT [47]
FRANCE [48]
HUNGARY [49]
INDIA [50]
KOREA [51]
LATIN AMERICA [52]
POLAND [53]
RUSSIA [54] [55] [56]
SAUDI ARABIA [57]
THAILAND [58] [59] [60] [61]
UKRAINE [62]
* This list is not meant to be comprehensive – but is presented at request of peer review to illustrate the wealth of sources for bibliographic registers of biomedical literature
Lower profile Arab journals may not be able to attract submission of randomised trials from home or overseas. The results from PsiTri did suggest that studies from the Arab world are most often seen in journals outside of the region. There is evidence from other studies that trials from non-English speaking countries are more likely to be published in Anglophone journals if statistically significant and those same authors are more likely to publish their other, less 'significant' studies in their home language [15]. We cannot tell how many trials would have been sent to 'mainstream' journals and rejected and we would have expected to see them published in the home literature. We did not. With raised awareness amongst editors of publication bias due to study origins, we hope this phenomenon, if contributing to the dearth of accessible trials from the Arab world becomes less prevalent.
Perhaps there are many fewer studies published in the Arab world than would be expected from the population. Previous work suggests that population is not a good predictor of productivity of trials relevant to schizophrenia [16]. Gross Domestic Product (GDP) proved a much more potent predictor. The greater the GDP, the greater the productivity of schizophrenia trials. Assuming this holds for all mental health trials, one potent factor could be that many of the Arab countries are in the low income bracket where output of [at least] schizophrenia trials would be expected to be low. Even the few richer Arab countries have a GDP which is focused on oil revenue. This uneven poverty may well be a factor in the Arab world's lack of productivity of mental health trials. Although poverty may be predictive, it is not necessarily causative and this fact may not fully explain why trial numbers in other more economically-disadvantaged parts of the developing world appear to be increasing [11]. Not all trials are highly expensive. Infrastructure for trials, however, may be suffering from under investment, and the culture or political climate may be hostile. One cannot underestimate the potential effects that culture may play in clinical trial conduction in the region. These potential effects may be expected to be seen much more acutely in the much-stigmatised area of mental health. Many factors could mediate against evaluative research.
There were a few further areas related to the study on which the authors would like to make further comment on.
Although not specifically reported in this study mainly owing to poor data availability, it may have been useful to comment on the number of studies which reported statistically significant results- which in turn may give an indication of potential publication bias. Most of the included trials did not appear to focus on topics that had a particular Arab-focus and therefore, notwithstanding issues around inter-regional validity, results from large, well designed Western trials (albeit covering similar subject matters) may provide equally-accurate results on which to base local service provision. It was difficult to confirm these observations as we were only able to gain access to the table-of-contents pages for most of the electronic databases and hence had limited ability to comment on the regional nature of the research. Although our study looked only at randomised controlled trials on mental health, we acknowledge the possibility that valuable research in other methodological formats may still be published with an Arab-focus and may provide useful answers to health questions in the area. The author's also acknowledge that the overall quality of the reporting of studies may have been a useful additional measure, however it was seen as being outside of the remit of this article. The ethical nature of included studies again could have been reported on if further data have been available. There is some evidence that trials conducted in 'resource-poor settings' may be more likely to include comparator arms that would be deemed sub-optimal, or even unethical in the West [17]. Conversely, what is deemed ethical in the West could be unethical in situations where resources are not so abundant. Judging the ethics of others by standards that are not based in local knowledge and culture may be problematic.
There is also the issue of researchers migrating from the region. Researchers of Arab origin may leave to work in more wealthy countries where funding for randomised controlled trials is more readily available.
Conclusion
Even if we have failed to identify many relevant studies, there is a suggestion that this large part of the world is not significantly contributing to evaluative mental health research. This perceived paucity of randomised controlled trials on mental health originating from the Middle East may in part be compounded by the lack of availability and dissemination of trial results via the internet. The apparent lack of randomised controlled trials in the Arab world could, in part, be due to a lack of funding. Despite the lack of randomised controlled trials, researchers in the Arab world are publishing studies on mental health issues. Although our research has highlighted the limited numbers of relevant trials accessible through the internet, it should be noted that there may be a substantial body of work being conducted and published in other formats. Creating a regional database may provide researchers with greater trial accessibility and hence locate a hitherto-'untapped well' of information.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
YT participated in the design of the study, carried out the literature search and joint drafted the manuscript. HES contributed to the analysis and interpretation of data and joint drafted the manuscript. CA participated in the design of the study, contributed to the analysis and interpretation of the data and revised the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors would also like to thank Gill Rizzello for her help in manuscript editing and revision. This work was supported by the intramural support of the University of Leeds and the Leeds Community Teaching NHS Trust.
==== Refs
WHO Scientific Group on Treatment of Psychiatric Disorders Evaluation of methods for the treatment of mental disorders 1991 Geneva: WHO
Medical Research Council Streptomycin in Tuberculosis Trials Committee Streptomycin treatment for pulmonary tuberculosis BMJ 1948 2 769 79
Chalmers I Hedges L Cooper H A brief history of research synthesis Eval Health Prof 2002 25 12 37 11868442 10.1177/0163278702025001003
Sackett D Richardson W Rosenberg W Evidence-based medicine: how to practice and teach EBM 1997 New York; Edinburgh: Churchill Livingstone
Chalmers I Dickersin K Chalmers T Getting to grips with Archie Cochrane's agenda BMJ 1992 305 786 8 1422354
Easterbrook P Berlin J Gopalan R Matthews D Publication bias in clinical research Lancet 1991 337 867 72 1672966 10.1016/0140-6736(91)90201-Y
Gilbody S Song F Publication bias and the integrity of psychiatry research Psychol Med 2000 30 253 8 10824646 10.1017/S0033291700001732
Stern J Simes R Publication bias: evidence of delayed publication in a cohort study of clinical research projects BMJ 1997 315 640 5 9310565
Tadmouri GO Bissar-Tadmouri N Biomedical publications in an unstable region: the Arab world, 1988–2002 Lancet 2003 362 1766 14643139 10.1016/S0140-6736(03)14868-4
El Ansari W Afifi Soweid RA Jabbour S Geography of biomedical publications Lancet 2004 363 489 author reply 489-90 14962532 10.1016/S0140-6736(04)15498-6
Isaakidis P Swingler GH Pienaar E Volmink J Ioannidis JP Relation between burden of disease and randomised evidence in sub-Saharan Africa: survey of research BMJ 2002 324 702 11909786 10.1136/bmj.324.7339.702
Takriti A Zakaria Z Panic disorder: Alprazolam versus cognitive therapy Arab Journal of Psychiatry 1989 1 12 21
Al-Haddad M Kamel C Sequeria RP Mawgood MA Zuclopenthixol versus haloperidol in the initial treatment of schizophrenic psychoses, affective psychoses and paranoid states: A controlled clinical trial Arab J Psychiatr 1996
Greist JH Jefferson JW Kobak KA Chouinard G DuBoff E Halaris A Kim SW Koran L Liebowtiz MR Lydiard B McElroy S Mendels J Rasmussen S White K Flicker C A 1 year double-blind placebo controlled fixed dose study of sertraline in the treatment of obsessive-compulsive disorder Inc Clin Psychopharmacol 1995 10 57 65
Egger M Zellweger-Zahner T Schneider M Junker C Lengeler C Antes G Language bias in randomized controlled trials published in English and German Lancet 1997 350 326 9 9251637 10.1016/S0140-6736(97)02419-7
Moll C Gessler U Bartsch S El-Sayeh HG Fenton M Adams CE Gross Domestic Product (GDP) and productivity of schizophrenia trials: an ecological study BMC Psychiatry 2003 3 18 14656379 10.1186/1471-244X-3-18
Kent DM Mwamburi M Bennish ML Kupelnick B Ioannidis JPA Clinical trials in Sub-Saharan Africa and established standards of care: A systematic review of HIV, tuberculosis, and malaria trials JAMA 2004 292 237 242 15249573 10.1001/jama.292.2.237
Darcourt G Feuillade P Bistagnin Y Robert P Pringuey D Touari M Merdji Y Bensmaïl B Antidepressant effect of 5-methoxypsoralen: The melatonin synchronizer hypothesis Eur Psychiatry 1995 3 142 54 10.1016/0767-399X(96)80103-9
Elwan O A comparative study of viloxazine and imipramine in the treatment of depressive states J Int Med Res 1980 7 17 7358207
Elwan O Souief M Hasan MA Allam M Psychometric assessment of the therapuetic efficacy of antidepressant agents J Int Med Res 1976 2 118 24 16794
Dickersin K How important is publication bias? A synthesis of available data AIDS Educ Prev 1997 9 15 21 9083596
Youssef FA Adherance to therapy in psychiatric patients: An empirical investigation Int J Nurs Stud 1984 1 51 7 6562083 10.1016/0020-7489(84)90007-5
Okasha A Sadek A A controlled double-blind clinical trial between maprotiline and amitriptyline in depressive illness J Egypt Med Assoc 1976 557 62 801741
El-Anany FG Maghraby HA Shaker SE Abdul-Momeim AM Primary nocturnal enuresis: A new approach to conditioning treatment Urology 1999 405 8 9933064 10.1016/S0090-4295(98)00597-4
Mahmoud KZ El-Dakhli MR Fahmi IM Abdul-Aziz AB Comparative value of prostaglandin E1 and papaverine in treatment of erectile failiure: Double-blind crossover study among Egyptian patients J Urol 1992 147 623 6 1538443
Carter JP Grivetti LE Davis JT Nasiff S Mansour A Mousa WA Atta AE Patwardhan VN Abdel Moneim M Abdou IA Darby WJ Growth and sexual development of adolescent Egyptian village boys: Effects of zinc, iron and placebo supplementation Am J Clin Nutr 1969 59 78 4885009
Askar A Rye GT Thioridazine in treatment of psychiatric patients Acta Psychiatr Scand (supplementum) 1970 211 91
Loza N El-Dosoky AM Okasha TA Khalil AH Hasan NM Dossenbach M Kratky P Okasha A Olanzapine compared to chlorpromazine in acute schizophrenia Eur Neuropsychopharmacol 1999 S291 10.1016/S0924-977X(99)80354-X
El-Islam MF Ahmed SA Erfan ME The effect of unilateral ECT on schizophrenic delusions and hallucinations Br J Psychiatry 1970 539 447 8 5481209
Al-Waili NS Indomethacin suppository to treat primary nocturnal enuresis: A double-blind study J Urol 1989 1290 2 2681838
Al-Waili NS Diclofenac sodium in the treatment of primary nocturnal enuresis: Double-blind crossover study Clin Exp Pharmacol Physiol 1986 139 42 3519020
Zahr LK Therapeutic play for hospitalised pre-schoolers in Lebanon Paediatr Nurs 1998 5 449 54
De Deyn PP Carrasco MM Deberdt W Jeandel C Hay DP Feldman PD Young CA Lehman DL Breier A Olanzepine versus placebo in the treatment of psychosis with or without associated behavioral disturbances in patients with Alzheimer's disease Int J Geriatr Psychiatry 2004 2 115 26 14758577 10.1002/gps.1032
Tewfik GI Jain VK Harcup M Magowan S Effectiveness of various tranquilizers in the management of senile restlessness Gerontologia Clinica 1970 6 351 9 4926614
Ezzat DH Ibraheem MM Makhawy B The effect of Piracetam on ECT-induced memory disturbances Br J Psychiatry 1985 720 1 3913489
Demerdash A Mazhar N Elhanbali M AbouGhazala E A double-blind trial of fluspirilene versus flupenthixol decanoate in the treatment of schizophrenic inpatients J Kuwait Med Assoc 1981 1 7 16
See RE Fido AA Maurice M Ibrahim MM Salama GM Risperidone induced increase of plasma norepinephrine is not correlated with symptom improvement in chronic schizophrenia Biol Psychiatry 1999 12 1653 6 10376128 10.1016/S0006-3223(98)00199-1
El-Sadr A Sabry AA Abdul-Rahman M El-Barnachawy R Koraitim M Treatment of primary nocturnal enuresis by oral androgen mesteralone: A clinical and csytometric study Urology 1990 331 5 2219614 10.1016/0090-4295(90)80241-E
Naguib M Samarkandi AH Premedication with melatonin: A double-blind, placebo-controlled comparison with midazolam British Journal of Anaesthesia 1999 875 80 10562782
Al-Rafak H Bello LL Turkustani A Adenuby JO Intra-nasal midazolam in conscious sedation of young pediatric dental patients J Dent Res 1999 1154
Al-Faris E Al-Subaie A Khoja T Al-Ansary L Abdul-Raheem F Al-Hamdan N Al-Mazrou Y Abdul-Moneim H El Khwsky F Training primary health care physicians in Saudi Arabia to recognize psychiatric illness Acta Psychiatr Scand 1997 439 44 9421340
Wahass S Kent G The modification of psychological interventions for persistent auditory hallucinations to an Islamic culture Behavioral and Cognitive Psychotherapy 1997 351 64
African Trials Registry
Eastview
BiblioMedica
Eastern Mediterranean Research Organisation
ENSTINET
PASCAL
Hungarian Academic Journals
IndMED
KoreaMED
LILACS
Polish Medical Bibliography
MediaSphere
Russian Medical Journal +
Russian Medical Server
Saudi Medical Journal
Thai Index Medicus
Thai Abstract Database
Thai Thesis
Thai Mulinet Library Catalog
The Panteleimon Project
|
16045805
|
PMC1199527
|
CC BY
|
2021-01-04 16:33:03
|
no
|
BMC Psychiatry. 2005 Jul 26; 5:30
|
utf-8
|
BMC Psychiatry
| 2,005 |
10.1186/1471-244X-5-30
|
oa_comm
|
==== Front
Behav Brain FunctBehav Brain FunctBehavioral and Brain Functions : BBF1744-9081BioMed Central 1744-9081-1-111604579810.1186/1744-9081-1-11ResearchInduction of compulsive-like washing by blocking the feeling of knowing: an experimental test of the security-motivation hypothesis of Obsessive-Compulsive Disorder Woody Erik Z [email protected] Victoria [email protected] Lisa [email protected] Hilary [email protected] Markad [email protected] Henry [email protected] Dept of Psychology, University of Waterloo, Waterloo, Ontario, Canada2 Dept of Psychiatry and Behavioural Neurosciences, McMaster University, Hamilton, Ontario, Canada3 Dept of Medicine, McMaster University, Hamilton, Ontario, Canada2005 26 7 2005 1 11 11 19 4 2005 26 7 2005 Copyright ©2005 Woody et al; licensee BioMed Central Ltd.2005Woody et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Background
H. Szechtman and E. Woody (2004) hypothesized that obsessive-compulsive disorder results from a deficit in the feeling of knowing that normally terminates thoughts or actions elicited by security motivation. To test the plausibility of this proposed mechanism, an experiment was conducted to produce an analog of washing in obsessive-compulsive disorder by eliciting a scenario of potential harm and using hypnosis to block changes in internally generated feelings that would normally occur during washing.
Results
Participants reacted with increased disgust, anxiety, and heart rate to their mental images of contamination and potential danger. As predicted, high but not low hypnotizable participants showed a significant prolongation of washing when change in feelings during washing was blocked hypnotically.
Conclusion
Results show that blocking the affective signal that is normally generated during security-related behaviors, such as washing, leads to prolonged performance of these behaviors. This finding lends support to the plausibility of the proposed model of obsessive-compulsive disorder.
==== Body
Background
In obsessive-compulsive disorder (OCD), a sense of compulsion is associated with performing ritualistic thoughts or actions. There are two types of mechanism that might explain the intrusiveness and urgency characteristic of OCD symptoms. One possibility is that there is a pathological intensity of excitation in the system that initiates the particular thoughts or actions, such that they are elicited too readily and strongly [e.g., [1]]. A contrasting possibility is that there is a deficit in the system that normally terminates these thoughts or actions, such that they persist too long.
The idea that OCD symptoms stem from a pathologic intensity of excitation is intuitively appealing because it is consistent with the widespread notion of compulsion as a force that initiates behavior. However, Reed [[2], p. 127] found that only a tiny minority of OCD patients described their experience of compulsions in such a way. Instead, the great majority described their experience of compulsions in terms of an inability to stop – for example, "I keep wondering, and then I can't get it out of my mind," or "I can't move on because I can't convince myself that I've finished what I'm doing." Reed [[3], p. 384] concluded that "those who are trapped in a circle of repetitive behavior do not report that something forces them to continue, but that they lack something to make them stop."
Likewise, descriptive accounts of OCD behavior suggest that most patients engage in few but extended episodes of compulsive behavior during the day, rather than episodes of normal duration but excessive frequency [4]. Such a behavioral profile suggests a dysfunctional stop mechanism rather than activation mechanism.
Conceptualizations of OCD as a Cognitive Disorder
Some conceptualizations of OCD have focused on the hypothesis that there is an underlying disorder of cognition. There are various ways a cognitive disorder might explain the inability to terminate thoughts and actions normally. For example, Reed [2] suggested that OCD symptoms may be the result of a central cognitive deficit in the defining of categories, in the determination of boundaries and limits, in the establishment of criteria, and in the allocation of class members. He argued that the obsessional style and engagement in rituals of these patients represent attempts to compensate for their cognitive inability to define and put closure on experiences. Similarly, Pitman [5] referred to this cognitive inability as a failure in the sense of task completion, and Pélissier and O'Connor [6] described it as a dysfunctional pattern of inductive reasoning.
Other recent explanatory models of OCD have also been strongly cognitive; for example, a major line of theorizing has implicated dysfunction in the metacognitive regulation of one's own stream of thoughts [7]. Accordingly, Salkovskis [8-10], Rachman [11,12], and Wells [13] have suggested a causative role for various dysfunctional beliefs that OCD patients appear to have about the meaning and implications of their conscious thoughts – for example, the belief that thinking something bad is virtually the same as actually doing it (thought-action fusion). In other words, OCD patients may have difficulty terminating thoughts and actions because they accord them exaggerated and perhaps irrational significance.
OCD as a Disorder of Security Motivation
However, such cognitive models do not seem to account well for some of the key features of OCD. In particular, a striking feature of the disorder is the inability to feel reassured by seemingly obvious and compelling information from the senses. For example, although compulsive hand washers know objectively that their hands look clean, they cannot generate the normal subjective conviction that they are truly clean, and so continue to wash [14].
Somewhat in contrast to cognitive approaches, we have recently proposed a theory of OCD that focuses on its motivational underpinnings [15]. According to this theory, OCD patients are haunted by a sense of anxiety because their particular concerns and behaviors are invoked by a potent special motivation that handles potential threats to existence (e.g., predation) and protection from harm. Because the concerns of the system are potential rather than imminent threats, this motivational system is open-ended (in the sense that logical certainty about the absence of potential threat is unattainable); consequently, the system is not under immediate environmental control. Due to this lack of a terminating signal in the environment, goal completion in this system is normally signaled by an endogenously generated terminator (experienced as a feeling of knowing or task accomplishment), but OCD patients either cannot generate this emotional signal or it is inadequate to inhibit the invoked motivation.
To denote the particular feeling of knowing that serves as an essential terminator of the species-specific motivation concerned with protection from harm, we coined the term "yedasentience," [16] from the Hebrew yeda = knowing and Latin sentire = to feel. Our core hypothesis may then be stated as follows [[15], p. 116]:
An internally generated feeling of knowing (termed yedasentience) provides a phenomenological sign of goal-attainment and has as its consequence the termination of thoughts, ideas or actions motivated by concerns of harm to self or others. Failure to generate or experience this feeling produces symptoms characteristic of OCD.
The purpose of present study was to test the possibility that dysfunction of such a feeling of knowing is a plausible mechanism for OCD-like behavior. Our experimental approach was to block this feeling and see if the blockage leads to OCD-like behavior – specifically, prolonged washing. In this way, we hoped to demonstrate that we could temporarily create in non-patient individuals an OCD-like profile of behavior.
Design of the Experiment
To produce an experimental analog of OCD washing, we needed to address two major issues. The first was how to create a sense of potential harm and thus elicit the security motivation underlying OCD behavior.
In our pilot studies, we initially tried to generate a sense of potential harm by using the methodology of Jones and Menzies [17]. In this approach, the experimenter asks participants to immerse their hands in a noxious mix of wet dirt and other materials and tells them, "For ethical reasons I should inform you that in this sort of procedure there is always the possibility of picking up bacteria that will result in serious illness" [[17], p. 123]. However, debriefing revealed that our participants did not find this danger protocol credible, perhaps at least in part because the experiment was taking place in a university hospital (and, of course, they also knew it had received ethical approval). Hence their experience lacked the appropriate emotional quality and significance.
Therefore, instead of providing a physical stimulus, we allowed the participants to use their imagination and recall their own experience of being in contact with something contaminated. We instructed them to imagine not only this specific experience but also the emotional reactions, such as disgust, that would accompany it. The use of such mental images as stimuli is consistent with research showing that imagination activates many of the same neural systems as are evoked by actual stimuli. Indeed, based on this research Kosslyn [18] has advanced the reality simulation principle:
"An object seen in a mental image can have the same impact on the mind and body that the actual object would have. ... Once the brain systems are engaged, they don't know where the impetus came from. This means that they can produce the same effects whether you activated it endogenously (from information in memory) or exogenously (from looking at something)."
The second major issue in designing the experiment was how to block yedasentience, the endogenous signal that we hypothesize normally terminates security-motivation-driven washing behavior. We used hypnosis for this purpose, because in people who are high in hypnotic responsiveness this technique permits the induction of far-reaching alterations in the sense of reality, independent of objective sensory input [e.g., [19-21]]. For example, individuals high in hypnotic suggestibility are able, under hypnosis, to experience hallucinations in a variety of sensory systems; in addition to such positive hallucinations, they are also capable of experiencing striking negative hallucinations – that is, not experiencing something actually present to their senses [e.g., [22]]. In addition, with hypnosis one can dissociate emotional experience from sensory qualities, as shown for example in the hypnotic manipulation of the emotional experience of pain independent of the perception of its sensory qualities [23].
Thus, using hypnosis in appropriately preselected participants, it is quite possible to dissociate subjective experience from the objective input available to the senses, and independently manipulate subjective convictions. It is worth stressing that we are using hypnosis as an empirical method to obtain a preparation suitable for testing the working hypothesis; we are not asking whether high hypnotic ability does or does not make one prone to OCD.
In summary, our experiment attempted to produce an analog of OCD washing by eliciting the feeling of potential harm and then blocking the changes in feeling that would normally occur during washing. It follows from the security-motivation hypothesis of OCD that the combination of these two conditions should yield prolonged washing. In addition, we included both high and low hypnotically responsive participants in the experiment. Because blocking changes in feeling should only be possible for highly hypnotizable participants, the low participants serve as a control for demand effects (e.g., participants merely behaving differently because it was directly implied that they should). Thus, the results of the experiment should yield a three-way interaction involving potential harm, blocking of change in feeling, and hypnotic susceptibility.
Method
Overview
Participants preselected as High or Low in Hypnotizability came to the lab to take part in a study described as addressing the physiological changes that accompany everyday behaviors and emotions. Heart-rate electrodes were attached to participants, they engaged in an initial hand washing to familiarize them with the sink set-up, and then they were hypnotized. Participants in the Potential Harm Suggested condition were instructed to imagine an emotional experience of touching a disgusting, contaminated object, whereas those in the Potential Harm Absent condition were asked to imagine an emotional experience of calm and relaxation. Next, participants in the Yedasentience Blocked condition were told that when they washed their hands they would not experience a sense of satisfaction, whereas those in the Yedasentience Not Blocked condition were told they would experience the usual sense of satisfaction. The main dependent variable was the duration of the subsequent hand-washing behavior.
Participants
The sample consisted of 96 female and 53 male university students and other individuals who responded to notices posted in the teaching and hospital buildings of McMaster University or to recruitment in undergraduate classes. Participants were either paid or given partial course credit. All prospective participants were pre-screened with the Waterloo-Stanford Group C Scale (WSGC; [24,25]) or, in a minority of cases, the Harvard Group Scale of Hypnotic Susceptibility, Form A (HGSHS:A; [26]). For inclusion in the study, participants were required to score either high (9–12) or low (0–3) in hypnotizability on these scales. To maximize statistical power in the high hypnotizable cells, approximately one-third of participants selected were low hypnotizable (58, or 38%) and two thirds were high hypnotizable (91, or 63%). As a consequence, the four experimental conditions for low hypnotizables have a range of 14 to 15 participants each, and the four conditions for high hypnotizables have a range of 21–24. The mean age of the participants was 25, with a range from 16 to 67 years. Of the 149 participants, 83 (55.7%) were 16 to 19 years of age, 36 (24.2%) were 20 to 28, and 30 (21.1%) were over 30. The study received ethics approval at both McMaster University and the University of Waterloo.
It may be noted that hypnotic susceptibility has a modest inverse relation with non-dissociative psychopathology, such as mood and anxiety disorders [e.g., [27]]. Thus, it is unlikely that the high hypnotisability group would inadvertently consist of individuals with more OCD-like tendencies prior to the experimental manipulations. Likewise, the modest relationship does not preclude the generalization of obtained findings to OCD patients.
Apparatus
Hand washing took place at a sink installed with an automatic faucet and an automatic soap dispenser, both activated by the proximity of hands. The faucet was preset to deliver a flow of water at a constant rate and temperature that did not vary across participants; the delivery of soap was similarly constant. A video camera (Panasonic AG-456UP) mounted directly over the sink, approximately one meter above it, recorded all washing episodes during the experiment onto a videotape (Panasonic (PV-VS4821-K). The camera lens was zoomed to capture a clear view of hands, illuminated by a 500 W type "T" halogen light bulb. A second video camera (Hitachi VM-7500LA) mounted at another location away from the sink captured a view of the entire room and provided a record of the whole experimental session. For recording of heart rate, the ECG signal was digitized at a sampling rate of 500 Hz using a 12-bit analog-to-digital converter (DATAQ, Akron, Ohio, U.S.A.) connected to an IBM-compatible PC; the ECG signal was displayed on the computer monitor throughout the session and stored on a hard disk at the defined periods; the mean heart rate during each recording period was later calculated using a QRS complex detection algorithm.
Procedure
Participants took part in the study individually and remained seated in a comfortable swivel chair throughout its duration. To begin, the experimenter provided the following rationale:
"As we experience emotions, there are corresponding changes in our body. In this experiment, we want to study that connection between emotions and these bodily changes. Hence, one of the things I'm going to do is to attach you to this heart rate monitor that will sensitively measure changes in your body."
"Another important aspect of emotion is that people differ considerably from one another in their emotional responses. In this experiment we want to find out your particular pattern of response. Accordingly, I will ask you to engage in some everyday behaviors, such as washing your hands. I will also make some suggestions about your feelings. I will record your underlying responses for three minute periods between each of these behaviors or suggestions. In addition, we need to videotape all the participants so that we can review their overt behavior."
"Finally, as you know, I will be hypnotizing you at the beginning of the experiment. The hypnosis allows you to respond to the suggestions about your feelings. It also helps you to clear your mind and relax your body. Under these conditions, we can get a much better baseline against which to sensitively measure subtle emotional changes."
After the experimenter had attached the heart-rate electrodes to the skin over participants' collarbones and lower rib, she instructed them to turn to the sink and wash their hands, thus familiarizing them with the washing set-up and procedure, including a tap activated by an automatic sensor, an automatic liquid soap dispenser, and a supply of paper towels for drying. Once participants finished washing, the experimenter instructed them to move as little as possible for 3 minutes, with hands resting in lap and eyes closed, and during this period, their baseline heart rate was recorded.
Next, the experimenter administered each participant the standardized hypnotic induction from the WSGC, which includes instructions for focusing attention, eyes closing, relaxation, and count-based deepening. At the conclusion of the induction, the experimenter asked participants to remain deeply hypnotized, as still as possible with their eyes closed, and their heart rate was recorded for another three-minute period.
At this point, participants in the Potential-Harm-Suggested condition were given the following instructions:
"I want you to think of an emotional experience that I am about to describe. I want you to think of something you could touch that you would find really disgusting. Something that could be contaminated with germs and bacteria. Something like feces ... or dirty toilet water ... or vomit ... or worms ... bugs – whatever you find disgusting. When you think of that object, I want you to imagine that you have touched it – something that is disgusting and may be contaminated with germs and bacteria. You feel disgusted because you touched something that could be contaminated with germs and bacteria. Think how disgusted and contaminated you feel after touching this object."
"Now keep your eyes closed and your hands resting in your lap. Just keep them there, without further movement, and with your eyes closed, for three minutes while we take a heart rate recording. During the 3 minutes, I want you to think about how disgusted and contaminated your hands make you feel. Throughout this time, remain hypnotized, with your eyes closed, attending to how disgusted your hands make you feel."
In contrast, participants in the control (Potential-Harm-Absent) condition were instead given the following instructions:
"I want you to think of an emotional experience that I am about to describe. I want you to think of something that you could do that would be very relaxing. Something that would make you calm and relaxed. Something like reading a book ... or watching TV ... listening to quiet music – whatever you find relaxing. When you think of it, I want you to imagine that you are doing it – something that is relaxing and calming. You feel pleasantly relaxed and calm because this is something that you enjoy doing. Think of how relaxed and calm you feel."
"Now keep your eyes closed and your hands resting in your lap. Just keep them there, without further movement, and with your eyes closed, for three minutes while we take a heart rate recording. During the 3 minutes, I want you to think about how calm and relaxed you feel. Throughout this time, remain hypnotized, with your eyes closed, attending to how calm and relaxed you feel."
After heart rate had been recorded for another three-minute period, participants were very briefly reminded of the kind of experience they were supposed to keep in mind, either "how disgusted and contaminated your hands make you feel" or "how calm and relaxed you feel." Next, participants in the Yedasentience-Blocked condition were given the following instructions:
"Now listen closely to my words, because this is very important. As you know: usually when you wash your hands there is a feeling of satisfaction that comes with it... However, now when you wash your hands, you will find that you do not experience that feeling of satisfaction. There will be a lack of satisfaction as you wash your hands."
"Okay, now open your eyes and turn to face the sink. Now go ahead and wash your hands with soap. Continue to think of how disgusted and contaminated your hands make you feel. Keep in mind that as you wash your hands, you will feel little or perhaps even no sense of satisfaction. The usual sense of satisfaction from washing your hands will be weak, or even absent."
In contrast, participants in the control (Yedasentience-Not-Blocked) condition were instead given the following instructions:
"Now listen closely to my words, because this is very important. As you know: usually when you wash your hands there is a feeling of satisfaction that comes with it... And when you wash your hands, you will find that you experience that feeling of satisfaction as you normally would. There will be a normal sense of satisfaction as you wash your hands."
"Okay, now open your eyes and turn to face the sink. Now go ahead and wash your hands, with soap. Continue to think of how disgusted and contaminated your hands make you feel. Keep in mind that as you wash your hands, you will feel a normal sense of satisfaction. You will experience the usual sense of satisfaction from washing your hands."
Participants then completed the washing and drying of their hands, which was recorded by video camera to allow accurate, objective determination of response duration. The experimenter next asked participants to close their eyes and make themselves comfortable in the chair, deeply hypnotized, with hands resting in lap, while heart rate was recorded for the last three-minute period.
In the last stage of the study, the experimenter carefully cancelled potentially disturbing suggestions (having touched something disgusting, and the inability to experience a sense of satisfaction from washing hands) for those participants who had been given them, and all participants were given another opportunity to wash their hands to show that they were now "clean and normal." Next, the experimenter brought participants out of hypnosis using the count-down procedure from the WSGC. After removing the electrodes she asked participants to describe the emotional experience they had been thinking of during the middle part of the experiment. Participants then filled out a brief questionnaire about their feelings during the study. Specifically, they rated their feelings when they were thinking of an emotional experience on five-point scales, from "not anxious" to "very anxious, and from "not disgusted" to "very disgusted." They also rated the extent to which they had experienced a sense of satisfaction while washing their hands in the middle part of the experiment, from "not at all" to "very satisfied." Finally, all participants were fully debriefed, thanked for their participation, and paid or given credit.
Measurement of Dependent Variables
The duration of washing was measured from the videotapes as the amount of time in seconds from the beginning of hand washing, when participants made the initial contact with soap or water, to its end, when participants removed their hands from the flow of water just prior to drying them with paper towels. Due to technical reasons associated with recording a measurable ECG signal, somewhat fewer data are available for heart rate than for the duration of washing.
Results
Duration of Washing
The main dependent variable in this study is the duration of the hand washing following the experimental manipulations. An analysis for outliers indicated that three of the response durations fell more than 3.5 standard deviations above the overall mean, and therefore these data points were omitted from the following analysis. All three outliers occurred in the potential-harm, yedasentience-blocked cell (the one hypothesized to lead to exaggerated response duration); two of the participants were high hypnotizable and one was low hypnotizable. In two of the cases, the experimenter stopped the participant from engaging in further hand washing after about 5 minutes by saying, "That's fine." The other outlying response duration was also almost 5 minutes (253 s); in comparison, the next longest response duration in the sample was 72 seconds.
We performed a three-way between-subjects analysis of covariance of the duration of washing, using baseline washing time as the covariate. The factors were Hypnotizability (high vs. low), Potential Harm (present versus absent), and Yedasentience (blocked vs. not blocked). This analysis yielded the predicted three-way interaction, F(1, 137) = 7.125, p = .009. Other effects that were statistically significant were the two-way interactions of Hypnotizability by Yedasentience, F(1, 137) = 4.285, p = .04, and of Potential Harm by Yedasentience, F(1, 137) = 4.926, p = .028, and all the main effects: Hypnotizability, F(1, 137) = 13.908, p < .001; Potential Harm, F(1, 137) = 43.004, p < .001; and Yedasentience, F(1, 137) = 5.341, p = .022. Altogether, these effects, along with baseline washing, explained 52% of variance. An analysis of these factors together with Gender yielded no significant effects for Gender or its interaction with any other factors.
Figure 1 shows the adjusted means for this analysis. With regard to the significant three-way interaction, it is evident that blocking yedasentience significantly (p < .05) increased response duration only in the predicted cell, when potential harm had been suggested to high-hypnotizable participants. In contrast, blocking yedasentience had negligible and insignificant effects on response duration when potential harm was not suggested to highs, and when potential harm was suggested or not to lows. This pattern of results confirms the main hypothesis of the study. Also of some interest, the significant main effect of Potential Harm, together with the lack of any significant Hypnotizability by Potential Harm interaction, indicates that the suggestion of potential harm tended to increase washing time for all participants, regardless of their level of hypnotizability: For no suggestion of potential harm, the mean was 21.48 s, SE = 1.02, whereas for suggestions of potential harm, it was 31.07 s, SE = 1.04.
Figure 1 Adjusted mean washing duration as a function of Hypnotizability, Potential Harm, and Blocking of Yedasentience. Mean with an asterisk is significantly different from every other mean, p < .05. The combination of Potential Harm and blocked Yedasentience yielded prolonged hand washing in the highly hypnotizable participants, compared to all other conditions.
Self-Reported Feelings
Disgust and anxiety
On five-point scales, participants rated the levels of disgust and anxiety they had felt after being asked to think of an emotional experience but before their subsequent hand washing. A three-way analysis of variance was performed on disgust, again with the factors Hypnotizability (high vs. low), Potential Harm (present versus absent), and Yedasentience (blocked vs. not blocked). This analysis yielded a significant Hypnotizability by Potential Harm interaction, F(1, 138) = 14.377, p < .001, and also significant main effects for both these factors: Hypnotizability, F(1, 138) = 10.460, p = .002; and Potential Harm, F(1, 138) = 262.784, p < .001. Together, the effects explained 71% of the variance in disgust ratings. The corresponding analysis of anxiety ratings, explaining 41% of the variance, yielded the same three significant effects: the Hypnotizability by Potential Harm interaction, F(1, 139) = 8.602, p = .004, and the main effects for Hypnotizability, F(1, 139) = 5.210, p = .024, and Potential Harm, F(1, 139) = 62.342, p < .001.
Figure 2 shows the means for Hypnotizability by Potential Harm for both disgust and anxiety. The manipulation of potential harm significantly (p < .05) increased disgust and anxiety levels for both low and high hypnotizable participants, indicating the success of this manipulation. However, the significant interactions indicate that the increases in disgust and anxiety were significantly greater for high hypnotizable participants than for their low hypnotizable counterparts.
Figure 2 Disgust and Anxiety as a function of Hypnotizability and Potential Hrm. Means with an asterisk are each significantly different from the adjacent mean for No Potential Harm, p < .05. The suggestion of Potential Harm was effective in generating higher self-ratings of Disgust and Anxiety in both Low and High Hypnotizable participants, although significantly more so in the High Hypnotizable participants.
Satisfaction while washing hands
Also on a five-point scale, participants rated the level of satisfaction they had experienced while subsequently washing their hands. The corresponding three-way analysis of variance of these ratings, explaining 40% of the variance, yielded a significant Hypnotizability by Yedasentience interaction, F(1, 139) = 20.246, p < .001, and a significant main effect of Yedasentience, F(1, 139) = 49.781, p < .001. Figure 3 provides the associated means. For the high hypnotizable participants, blocking yedasentience significantly (p < .05) reduced their experience of satisfaction while washing their hands; whereas for the low hypnotizable participants, this effect was negligible and statistically insignificant. The implication is that, as anticipated, only high hypnotizables can effectively enact the suggestion to block yedasentience.
Figure 3 Satisfaction as a function of Hypnotizability and Blocking of Yedasentience. Mean with an asterisk is significantly different from the adjacent mean for No Blocking of Yedasentience, p < .05. Blocking Yedasentience significantly reduced self-ratings of satisfaction during the hand-washing in the High Hypnotizable participants, but not in the Low Hypnotizable participants.
Heart Rate
The study also included a more covert index of how participants were feeling, namely their heart rate. We submitted the heart-rate data to a four-way mixed-model analysis of covariance, using baseline heart rate as the covariate. The three between-subject factors were Hypnotizability (high vs. low), Potential Harm (present versus absent), and Yedasentience (blocked vs. not blocked). The within-subject factor was Trials, with three times of measurement: Trial 1 was measured just after the hypnotic induction; Trial 2 was measured just after the suggestion of an emotional experience (e.g., a situation of potential harm); and Trial 3 was measured just after the completion of hand washing. This analysis yielded one significant effect, the two-way interaction of Trials by Potential Harm, multivariate F(2, 117) = 5.803, p = .004, which explained 9% of the variance (Wilk's Lambda = .910). Figure 4 shows the relevant means. The mean for Trial 2 in the potential-harm-suggested condition is significantly higher (p < .05) than each of the three other means, which in turn do not differ significantly from one another. Thus, the suggestion of an experience of potential harm increased participant's heart rates, whereas the control suggestion of a positive experience did not; in addition, this potential-harm-related increase dissipated fully once the participants had been allowed to wash their hands. (Note that it makes sense for the experimental factor of Yedasentience not to be involved in this effect: Its manipulation took place between Trial 2 and Trial 3, and heart rate at Trial 3 was measured after the completion of handwashing, when participants had been able to take as long as they wanted to clean their hands.)
Figure 4 Heart Rate as a function of Time of Measurement and Potential Harm. Mean with an asterisk is significantly different from every other mean, p < .05. The suggestion of Potential Harm increased participants' heart rates compared to the control suggestion; this increase disappeared once the participants had washed their hands.
Discussion
Although high hypnotizables showed a particularly strong emotional response to their mental images of contamination and potential harm, all participants tended to respond with increased disgust and anxiety. In addition, all participants, regardless of their level of hypnotizability, tended to react to their images of potential harm with elevated heart rate and increased washing time, and this elevated heart rate returned to baseline when they had washed. Taken together, these self-report, heart-rate, and behavioral data indicate that both high and low hypnotizable participants succeeded in imagining a situation of potential harm in a vivid and involving way. This is an essential precondition for the meaningfulness of the central manipulation of the experiment, which was the blocking of yedasentience.
The effect of the yedasentience-blocking suggestion was highly specific: It had the predicted effect of prolonging the duration of washing only in the predicted condition, in which potential harm had been suggested to high-hypnotizable individuals. This key result supports our hypothesis that the dysfunction of such a feeling of knowing is a plausible mechanism for OCD-like behavior.
The pattern of results obtained also helps to discount certain alternative explanations of the results. For example, although we did not directly tell participants to wash longer, it might be argued that we simply implied it in the suggestion for a lack of a feeling of satisfaction. However, only the high hypnotizables showed prolonged washing in response to this suggestion, and they showed it only after potential danger had been invoked. Thus, their extra washing would appear to be an integrated, natural response to the blocking of yedasentience, rather than merely some reflection of demand characteristics. Similarly, another possible alternative explanation would be that the yedasentience-blocking suggestion acted inadvertently as an additional suggestion about the state of dirtiness of the participants' hands. However, contrary to such an interpretation, the effects of the Potential Harm and Yedasentience manipulations were not additive: For high hypnotizables, when yedasentience was not blocked, potential harm had no significant effect on washing time, and when potential harm was low, the blocking of yedasentience had no effect on washing time.
One might also question whether this increase in washing time, which was fairly modest in magnitude (about 20 s), was sufficiently long to represent an analogue of OCD-like washing. The 20-s increase needs to be put into perspective: It may be compared with the 42-s increase due to a high-danger manipulation that Jones and Menzies [17] obtained in the top 10% of scorers on an OCD-screening instrument, who had put their hands for 5 minutes in a garbage can of dirt, animal hair, raw meat, and household food scraps. In addition, it is noteworthy that in our study the participants' hands were never actually dirty (indeed, they had just been washed a few minutes previously). Finally, it is worth mentioning that three participants showed a far more prolonged response to the yedasentience-blocking suggestion, continuing to wash their hands for about 5 minutes, or possibly longer if they had not been stopped. What made these participants different from the others in this study is unknown, but it is relevant that they appeared quite anxious and uncomfortable during their hand washing.
Limitations of the present study
There are two important limitations of the present study. First, the study pertains most directly to an understanding of compulsive behavior rather than obsessive thoughts. Second, the study addresses only one form of compulsive behavior, namely, washing, but there are other kinds of compulsive behaviors such as checking or hoarding. Nevertheless, it is important to note that the underlying model addresses a broad range of OCD phenomena, including obsessional symptoms, as discussed elsewhere [15,28].
Other potential limitations of the present study merit attention. Because hypnosis is sometimes considered to be an altered state of consciousness, it could be argued that washing behavior in this state has limited relevance to the behavior of OCD patients. For example, it might be thought that hypnosis would interfere with the experience of anxiety that characterizes the experience of OCD patients. However, as the presented self-reports (Figure 2) and heart rate data (Figure 4) clearly showed, the participants in the relevant groups did report anxiety in response to the suggestion of potential harm and this anxiety dissipated when they washed their hands. In fact, the state of hypnosis did not limit the extent of anxiety as evidenced by the observation that in the high hypnotizable participants anxiety levels were just as high as in low hypnotizables. Thus, the state of hypnosis is not incompatible with the experience of anxiety. Similarly, it might be thought that participants in hypnosis become incapable of making conscious decisions. However, as many studies have indicated, such a view is incorrect [29]. Overall, the demonstration of the effects of yedasentience blockade under hypnosis should apply to similar behavioral effects of yedasentience blockade in OCD patients.
Another potential limitation is that the study lacks a manipulation check for the success of yedasentience blockage. Two pieces of data address this issue. First, Figure 3 shows that self-ratings of satisfaction are consistent with the intended purpose of the manipulation to block yedasentience. Second, Figure 1 illustrates that the experimental manipulation produced the expected 3-way interaction, again providing support for the effectiveness of the manipulation. Thus, the effectiveness of yedasentience manipulation is not simply assumed and in fact the findings noted above constitute the empirical evidence that the manipulation was effective.
Finally, it might be objected that hypnotically induced behaviors are simply socially sanctioned role playing. The widely accepted control for this potential problem is to include low hypnotizable subjects, who are exposed to exactly the same role demands. The fact that the low hypnotizable participants in our study did not show the same response suggests that role playing is not the key explanation for the observed results.
Implications for Future Research
Our instructions for imagining a scenario of potential danger were double-barrelled: They involved both the idea of potential danger (contamination) and the emotion of disgust. One may ask about the respective roles of these two aspects, and whether both are actually important in eliciting the relevant security motivation.
Along these lines, some recent work indicates that the emotion of disgust may be of special importance in OCD [30,30-33]. Nonetheless, although the relevance of disgust to compulsive washing seems clear, it is much more difficult to see its relevance to some other OCD behaviors – for example, compulsive checking.
There is some evidence that subtypes of OCD exist [34-36] and that checkers may be different from washers [37]. Accordingly, we would propose that the special role of disgust is as follows: If associated with the signal of potential danger there is an induced feeling of disgust, then washing responses are potentiated. Thus, although we would argue that the invocation of disgust is not the pathogenic characteristic of OCD (in our model, absence of yedasentience is pathogenic), the presence of disgust may be a factor that biases OCD symptoms towards washing compulsions. Substantiating the possibility that different subtypes of OCD may have different special emotions is an important topic for further research.
Similarly, another important task for future research is to show that blocking the feeling of knowing, as was done in the present experiment to elicit OCD-like prolongation of hand-washing behavior, can also elicit other major types of OCD-like behavior, including checking behavior. Such research could not only help to evaluate the generality of our findings, but also help to elucidate the differences between separable classes of OCD behavior – for example, whether there is another particular affect, paralleling the role of disgust in washing, that is specific for the invocation of checking behavior.
Finally, the security-motivation hypothesis of OCD has other important implications. For example, we have provided a detailed provisional model of its hypothesized neural underpinnings and speculated on its implications for treatment [15]. We hope the present demonstration of its plausibility stimulates wider interest in this hypothesis.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
EZW and HS were the principal investigators, who designed the experiment, performed the data analysis, and wrote the manuscript. VL, LS, and HG participated in the design of the study and helped to collect the data. MK assisted with carrying out the heart-rate investigations.
Acknowledgements
This study was supported by operating funds from the Ontario Mental Health Foundation and the Canadian Institutes of Health Research (MOP 74553).
==== Refs
Oltmanns TF Gibbs NA Emotional responsiveness and obsessive-compulsive behavior Cognition & Emotion 1995 9 563 578
Reed GF Obsessional Experience and Compulsive Behaviour: A Cognitive-Structural Approach 1985 Orlando, Florida: Academic Press, Inc
Reed GF The obsessional-compulsive experience: a phenomenological reemphasis PPR 1977 37 381 384
Neziroglu F Yaryura-Tobias JA Over and Over Again: Understanding Obsessive-Compulsive Disorder 1991 Toronto: Lexington Books
Pitman RK Animal models of compulsive behavior Biol Psychiatry 1989 26 189 198 10.1016/0006-3223(89)90022-X 2660914
Pelissier MC O'Connor KP Deductive and inductive reasoning in obsessive-compulsive disorder Br J Clin Psychol 2002 41 15 27 10.1348/014466502163769 11931675
Purdon C Clark DA Meta-cognition and obsessions Clinical Psychology and Psychotherapy 1999 6 96 101 10.1002/(SICI)1099-0879(199905)6:2<102::AID-CPP191>3.3.CO;2-X
Salkovskis PM Obsessional-compulsive problems: a cognitive-behavioural analysis Behav Res Ther 1985 23 571 583 10.1016/0005-7967(85)90105-6 4051930
Salkovskis PM Cognitive-behavioural factors and the persistence of intrusive thoughts in obsessional problems Behav Res Ther 1989 27 677 682 10.1016/0005-7967(89)90152-6 2610662
Salkovskis PM Swinson RP, Antony MM, Rachman S, Richter MA Psychological approaches to the understanding of obsessional problems Obsessive-compulsive disorder: Theory, research, and treatment 1998 New York, NY: The Guiford Press 33 50
Rachman S A cognitive theory of obsessions Behav Res Ther 1997 35 793 802 10.1016/S0005-7967(97)00040-5 9299799
Rachman S A cognitive theory of obsessions: elaborations Behav Res Ther 1998 36 385 401 10.1016/S0005-7967(97)10041-9 9670600
Wells A Cognitive therapy of anxiety disorders: A practice manual and conceptual guide 1997 New York, NY, US: John Wiley and Sons, Inc
Rapoport JL The Boy Who Couldn't Stop Washing: The Experience and Treatment of Obsessive-Compulsive Disorder 1989 New York: E.P. Dutton
Szechtman H Woody E Obsessive-compulsive disorder as a disturbance of security motivation Psychol Rev 2004 111 111 127 10.1037/0033-295X.111.1.111 14756589
Woody E Szechtman H Hypnotic hallucinations and yedasentience Contemporary Hypnosis 2000 17 26 31
Jones MK Menzies RG Role of perceived danger in the mediation of obsessive-compulsive washing Depress Anxiety 1998 8 121 125 10.1002/(SICI)1520-6394(1998)8:3<121::AID-DA4>3.0.CO;2-P 9836063
Kosslyn SM What shape are a German shepherd's ears? http://www.edge.org/documents/archive/edge103.html
Hilgard ER Hypnotic susceptibility 1965 New York, NY: Harcourt, Brace & World
Szechtman H Woody E Bowers KS Nahmias C Where the imaginal appears real: a positron emission tomography study of auditory hallucinations Proc Natl Acad Sci U S A 1998 95 1956 1960 10.1073/pnas.95.4.1956 9465124
Woody E Szechtman H Hypnotic hallucinations: towards a biology of epistemology Contemporary Hypnosis 2000 17 4 14
Kosslyn SM Thompson WL Costantini-Ferrando MF Alpert NM Spiegel D Hypnotic visual illusion alters color processing in the brain Am J Psychiatry 2000 157 1279 1284 10.1176/appi.ajp.157.8.1279 10910791
Rainville P Duncan GH Price DD Carrier B Bushnell MC Pain affect encoded in human anterior cingulate but not somatosensory cortex Science 1997 277 968 971 10.1126/science.277.5328.968 9252330
Bowers KS The Waterloo-Stanford Group C (WSGC) scale of hypnotic susceptibility: normative and comparative data Int J Clin Exp Hypn 1993 41 35 46 8425774
Bowers KS Waterloo-Stanford Group Scale of Hypnotic Susceptibility, Form C: manual and response booklet Int J Clin Exp Hypn 1998 46 250 268 9650438
Shor RE Orne EC Harvard Group Scale of Hypnotic Susceptibility 1962 Palo Alto, CA: Consulting Psychologists Press
Spiegel D Detrick D Frischholz E Hypnotizability and psychopathology Am J Psychiatry 1982 139 431 437 7065288
Woody EZ Szechtman H Motivation, time course, and heterogeneity in obsessive-compulsive disorder: Response to Taylor, McKay, and Abramowitz (2005) Psychol Rev 2005 112 658 661 10.1037/0033-295X.112.3.658 16060757
Bowers KS Hypnosis for the Seriously Curious 1983 New York: W.W. Norton & Company, Inc
Sprengelmeyer R Young AW Pundt I Sprengelmeyer A Calder AJ Berrios G Winkel R Vollmoeller W Kuhn W Sartory G Disgust implicated in obsessive-compulsive disorder Proc Biol Sci 1997 264 1767 1773 10.1098/rspb.1997.0245 9447734
Stein DJ Liu Y Shapira NA Goodman WK The psychobiology of obsessive-compulsive disorder: how important is the role of disgust? Current Psychiatry Reports 2001 3 281 287 11470034
Shapira NA Liu Y He AG Bradley MM Lessig MC James GA Stein DJ Lang PJ Goodman WK Brain activation by disgust-inducing pictures in obsessive-compulsive disorder Biol Psychiatry 2003 54 751 756 10.1016/S0006-3223(03)00003-9 14512216
Woody SR Teachman BA Intersection of disgust and fear: Normative and pathological views Clinical Psychology: Science and Practice 2000 7 291 311 10.1093/clipsy/7.3.291
Leckman JF Grice DE Boardman J Zhang HP Vitale A Bondi C Alsobrook J Peterson BS Cohen DJ Rasmussen SA Symptoms of obsessive-compulsive disorder Am J Psychiatry 1997 154 911 917 9210740
Miguel EC Baer L Coffey BJ Rauch SL Savage CR Osullivan RL Phillips K Moretti C Leckman JF Jenike MA Phenomenological differences appearing with repetitive behaviours in obsessive-compulsive disorder and Gilles de Tourette's syndrome Br J Psychiatry 1997 170 140 145 9093502
Antony MM Swinson RP Swinson RP, Antony MM, Rachman S, Richter MA Diagnostic issues and epidemiology in obsessive-compulsive disorder Obsessive-compulsive disorder: theory, research, and treatment 1998 New York, NY: The Guiford Press 3 32
Pigott TA Obsessive-compulsive disorder: symptom overview and epidemiology Bull Menninger Clin 1998 62 A4 32 9810775
|
16045798
|
PMC1199578
|
CC BY
|
2021-01-04 22:25:38
|
no
|
Behav Brain Funct. 2005 Jul 26; 1:11
|
utf-8
|
Behav Brain Funct
| 2,005 |
10.1186/1744-9081-1-11
|
oa_comm
|
==== Front
BMC BiochemBMC Biochemistry1471-2091BioMed Central London 1471-2091-6-141609554210.1186/1471-2091-6-14Research ArticleLow-magnesium, trans-cleavage activity by type III, tertiary stabilized hammerhead ribozymes with stem 1 discontinuities Burke Donald H [email protected] S Travis [email protected] Department of Chemistry, Indiana University, Bloomington, IN 47405-7102 U.S.A2 Department of Molecular Microbiology and Immunology, 471h Life Sciences Center, University of Missouri-Columbia, School of Medicine, 1201 Rollins Dr., Columbia, MO 65212-7310 U.S.A2005 12 8 2005 6 14 14 4 6 2005 12 8 2005 Copyright © 2005 Burke and Greathouse; licensee BioMed Central Ltd.2005Burke and Greathouse; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Low concentrations of free magnesium in the intracellular environment can present critical limitations for hammerhead ribozymes, especially for those that are designed for intermolecular (trans) cleavage of a host or pathogen RNA. Tertiary stabilizing motifs (TSM's) from natural and artificial ribozymes with a "type I" topology have been exploited to stabilize trans-cleaving hammerheads. Ribozymes with "type II" or "type III" topologies might seem incompatible with conversion to trans-cleavage designs, because opening the loop at the end of stem 1 or stem 2 to accommodate substrate binding is expected to disrupt the TSM and eliminate tertiary stabilization.
Results
Stem 1, together with single-stranded segments capping or internal to this stem, contains both the substrate-binding and tertiary stabilization functions. This stem was made discontinuous within the sTRSV hammerhead ribozyme, thereby separating the two functions into discrete structural segments. The resulting ribozyme, designated "RzC," cleaved its 13 nucleotide target substrate at MgCl2 concentrations as low as 0.2 mM at 25°C and 0.5 mM at 37°C. Under multiple-turnover conditions, nearly thirty turnovers were observed at the highest substrate:RzC ribozyme ratios. Similar stabilization was observed for several derivatives of RzC. Catalytic activity was diminished or eliminated at sub-millimolar MgCl2 concentrations for ribozymes with weakened or deleted tertiary interactions. Eadie-Hofstee analysis revealed that the stabilized and non-stabilized ribozymes bind their substrates with equivalent affinities, suggesting that differences in observed activity are not the result of diminished binding. Some of the stabilized and non-stabilized ribozymes appear to fold into a heterogeneous collection of conformers, only a subset of which are catalytically active.
Conclusion
Hammerhead ribozymes with the "type III" topology can be converted to a tertiary, trans-cleavage design. Separating the stabilization and substrate recognition functions of stem 1 increases cleavage activity at physiological concentrations of divalent magnesium while retaining recognition of exogenous targets. Trans-cleaving ribozymes that exploit the tertiary stabilizing motifs of all natural hammerhead topologies can therefore be used in intracellular applications.
==== Body
Background
Self-cleaving hammerhead ribozymes contain three base-paired stems joined by a highly conserved core. Tertiary stabilizing motifs (TSM) of diverse morphologies between single-stranded elements at the ends of, or within, stems 1 and 2 increase cleavage activity at physiological concentrations of divalent magnesium ions in vitro and in cells [1-8]. This discovery has propelled a resurgence of interest in metal ion binding by hammerhead ribozymes [9,10] and in the use of intracellularly expressed ribozymes as gene-knockdown agents. Low magnesium concentrations in the intracellular environment can be a critical limitation for hammerhead ribozymes. Although the total intracellular concentration of divalent magnesium is approximately 3.5 to 8.5 mM, analysis of 31P chemical shift indicates that free Mg2+ ranges from 0.2 to 1.2 mM and is generally between 0.4 to 0.8 mM depending on tissue type and physiological state [11-15]. Consistent with this view, the intracellular kinetic behavior of a hairpin ribozyme is more closely approximated by in vitro assays carried out at 2.0 mM MgCl2 than at 10 mM MgCl2 [16]. It is therefore important to define the ribozyme topologies and sequences that confer low magnesium activity.
Hammerheads are classified as being of type I, II or III according to whether the 5' and 3' termini reside within stem 1, 2 or 3, respectively (Figure 1A). The distinct connectivity patterns make these three types topologically non-equivalent. Tertiary-stabilized type I hammerheads, such as the SMαl ribozyme from Schistosoma mansoni, are readily adapted for trans-cleavage by opening the loop at the end of stem 3. In the SMαl ribozyme, however, nucleotides from both the substrate and ribozyme strands contribute to establishing stable tertiary interactions, significantly limiting the range of substrates that can be targeted for cleavage at physiological concentrations of Mg2+. We recently described hammerhead ribozyme RzB, which was derived from in vitro selections from a library of type I self-cleaving hammerheads. RzB carries an artificial TSM that is nearly independent of the sequence of the RNA fragment to be cleaved, freeing the experimental design from constraints encountered in ribozymes based on SMαl [4].
Figure 1 A. Types I, II and III hammerhead ribozymes. Peripheral regions shown as dotted lines contain the tertiary stabilizing motifs and can be of arbitrary sizes. Stems 1, 2 and 3 are indicated. B. The type III hammerhead ribozyme from sTRSV, showing tertiary interactions predicted from comparative sequence analysis, mutational data and computational modeling [2] (pairwise interactions depicted according to ref [22]).
There have not been reports of using the type II or type III topology for low-magnesium trans-cleavage. Opening the loop at the end of stem 1 in these ribozymes to accommodate substrate binding is expected to disrupt the TSM and eliminate tertiary stabilization. We reasoned that type II and type III hammerhead ribozymes could nevertheless be used for trans-cleavage at physiological magnesium concentrations if the substrate binding function of stem 1 could be separated from the tertiary stabilizing function of the TSM carried within loop 1. To this end, we constructed trans-cleaving versions of the type III self-cleaving hammerhead ribozyme from the Tobacco Ring-Spot Virus satellite RNA (sTRSV) [17]. The functional separation was achieved by placing both the 5' end of the ribozyme and the 3' end of the cleavage substrate within stem 1. Ribozymes with a discontinuous stem 1 exhibited tertiary stabilization at physiological magnesium. We further demonstrate that this stabilization extends to cleavage of a human ras oncogene mRNA fragment.
Results & Discussion
Low-magnesium activity of trans-cleaving ribozyme with discontinuous stem 1
Ribozyme "RzC" was designed to preserve the endogenous tertiary interactions of the sTRSV hammerhead ribozyme. The distal half of stem 1 and all of loop 1 (terminal three base pairs and seven-nucleotide loop) are continuous sTRSV sequence. Substrate is recognized through base pairing with the proximal half of stem 1 (three base pairs) and all of stem 3. Stem and loop 2 are native sTRSV sequence, again to preserve tertiary interactions (Figure 2). Thus ribozyme RzC separates the two functions of stem 1 – substrate recognition and tertiary stabilization – into discrete structural domains.
Figure 2 Ribozymes described in this study. Stems la, Ib, 2 and 3 are indicated. Substrate strands are shaded. Nucleotides involved in establishing the roles of tertiary stabilization in RzC, RzCAΔ1 and RzCΔ2 are boxed.
The catalytic activity of RzC was measured in 10 mM MgCl2/pH 7.5/37°C for trans-cleavage of a 13-mer oligoribonucleotide denoted "substrate C." Greater than 40% of this substrate was cleaved within the initial 15 seconds, yielding an initial rate of at least 3.2 min-1 (Figure 3, diamonds). Similar rapid cleavage was observed at 5 mM and at 2 mM MgCl2 (not shown). Importantly, RzC also cleaves substrate C at sub-millimolar Mg2+ concentrations, yielding an initial rate of 0.3 min-1 in 0.5 mM MgCl2 (Figure 3, triangles). When activity was measured at 25°C, ribozyme RzC continued to cleave rapidly at 0.2 mM MgCl2 (0.2 min-1) (Table 1). Much higher concentrations of MgCl2 are required to observe comparable rates for minimal ribozymes lacking TSM, suggesting that the TSM from sTRSV is functional in the context of a discontinuous stem 1.
Table 1 Kinetic parameters for discontinuous ribozymes derived from sTRSV
[MgCl2], mM f∞ fa ka kb net init rate, min-1
RzC [25] 1.0 0.59 0.70 4.23 0.040 1.75
0.5 0.54 0.58 2.03 0.070 0.65
0.2 0.42 0.23 3.57 0.091 0.37
0.1 0.11 0.62 0.92 0.010 0.06
RzC [37] 10 0.74 0.67 6.49 0.013 3.2
5 0.63 0.77 2.31 0.022 1.1
2 0.71 0.66 3.52 0.023 1.7
0.5 0.73 0.38 0.98 0.040 0.3
RzCΔl [37] 10 0.68 0.78 3.92 0.12 2.09
5 0.76 0.71 5.22 0.033 2.82
2 0.76 0.57 1.51 0.062 0.67
0.5 0.58 0.074 1.77 0.028 0.09
RzCΔ2 [37] 10 0.51 0.55 7.1 0.081 2.0
5 0.66 0.41 3.1 0.025 0.85
2 0.58 0.229 3.1 0.036 0.43
0.5 n. d. n. d. n. d. n. d. n. d.
Kinetic parameters were determined for discontinuous ribozymes as described in Methods. Single-turnover reactions were performed at 25°C or 37°C (square brackets next to ribozyme designations), and the data were fit to a double exponential curve. n.d., no cleavage detected. No cleavage was observed for RzC at 0.075 and 0.05 mM MgCl2 (not shown).
Figure 3 Magnesium dependence of substrate cleavage by discontinuous ribozymes with varying degrees of tertiary stabilization. Traces shown are for cleavage at 37°C in 10 mM (diamonds) or 0.5 mM (triangles) MgCl2. Data were fit to double exponential kinetic model as described in Methods, using best fit parameters given in Table 1.
Two variants of RzC were built to assess the contribution of the tertiary interaction modules to the observed magnesium dependence. Ribozymes RzCΔ1 and RzCΔ2 are identical to RzC outside of stem 1. In ribozyme RzCΔ1, the seven nucleotide terminal loop sequence 5' UGUGCUUU 3' is replaced with a stable tetraloop 5' UUCG 3', while in ribozyme RzCΔ2, the loop and terminal three base pairs are removed altogether. When reactions were monitored in 10 mM MgCl2, the deletions had little effect on initial rates, as all three ribozymes cleaved substrate C with initial rates of 2–3 min-1. Decreasing the MgCl2 concentration to 2.0 mM, however, produced a clear difference in initial cleavage rates, which were greatest for RzC and slowest for RzCΔ2. In 0.5 mM MgCl2, the initial rate of RzC was 3 times greater than that of RzCΔ1 (0.3 vs. 0.09 min-1), and no substrate cleavage at all was detected for RzCΔ2 (Figure 3 and Table 1). Interestingly, the rates observed for RzCΔ1 were closer to those of RzC than to those of RzCΔ2. It is possible that the intermediate activity seen for RzCΔ1 at low magnesium may be due to weak tertiary stabilization using alternative interactions between loop 2 and the stable tetraloop at the end of loop 1, although this possibility was not further explored.
The enhanced activity of RzC relative to RzCΔ1 and RzCΔ2 at submillimolar MgCl2 support the underlying hypothesis that the TSM stabilizes productive RNA folding. The fraction of the pre-annealed ribozyme-substrate complex that participates in the rapid phase of the cleavage reaction (fa) decreases only slightly for RzC at sub-millimolar MgCl2, while it drops precipitously for RzCΔ1 and RzCΔ2, suggesting that that fraction of the ribozyme that folds into a productive conformation is greater for RzC than for the other two ribozymes. The fraction of the ribozyme-substrate complex that can access the active conformation is given by f∞. (This estimation is actually a lower limit, as it assumes that the back reaction is negligible; a reasonable assumption given that the three-nucleotide 3' cleavage product is expected to dissociate rapidly, minimizing the possibility of re-ligation.) For reactions in which cleavage is observed, the value of f∞ is not strongly dependent upon MgCl2 concentration, although it is slightly lower for RzCΔ2 than for the other two. Together, these results indicate that the TSM sequence elements from the sTRSV hammerhead are providing similar tertiary stabilization in RzC.
Substrate affinity unaffected by TSM deletions
To rule out the possibility that the rate differences among RzC, RzCΔ1 and RzCΔ2 might be due to differential substrate-binding affinity, multiple turnover cleavage kinetics were measured as a function of substrate concentration using 1 nM ribozyme and excess substrate (10 to 200 nM). These conditions yielded between 4 and 28 nM cleaved product, indicating between four and twenty-eight turnovers. The slope of an Eadie-Hofstee plot of the rate data gives the apparent affinity of the ribozyme-substrate interaction as the Michaelis-Menten constant, Km. When multiple-turnover reaction kinetics were monitored in 10 mM MgCl2, all three ribozymes gave comparable Km values (~40 nM for RzC vs. ~70 nM for RzCΔ1 and RzCΔ2). Thus, substrate affinity is dominated by base paring within stem 3, and is not substantially affected by either the tertiary docking interactions or by the stacking between stems la and lb. Because the single-turnover reactions were performed using 10-fold excess ribozyme (1000 nM) over substrate (100 nM) at concentrations that are more than an order of magnitude above the estimated Km values (40–70 nM), the substrates are assumed to have been fully bound to ribozyme. Differences in the magnesium sensitivities among RzC, RzCΔ1 and RzCΔ2 are therefore not due to relative occupancy of the enzyme, but rather to conformational differences arising from differential tertiary stabilization. It was not possible to compare affinities of the three ribozymes at 0.5 mM MgCl2, as only RzC was active under these conditions. The slight curvature in its Eadie-Hofstee plot could be interpreted as experimental noise, or as reflecting higher affinity at low concentrations of substrate than at high concentrations.
Effects of sequence context on cleavage in sub-millimolar MgCl2
Two new ribozymes were generated to determine the generality of the discontinuous stem 1 hammerhead design (Figure 2). First, in construct RzC4, the individual base pairs of stems 1 and 3 in RzC were inverted to their Watson-Crick opposites except for the GUC at the cleavage site. The magnesium dependence of RzC4 was quite similar to that of RzC, declining 4-fold as MgCl2 was reduced from 5 mM to 0.5 mM (Figure 5, Table 2). Second, ribozyme RzC-ras was designed to cleave the fifteen nucleotide fragment of the human ras oncogene. In published cellular assays, ribozyme-mediated cleavage at this site reduced expression of a ras-luciferase fusion by 60% [18]. The 3' end of the RzC-ras substrate extends beyond the ribozyme, generating a branched junction within stem 1. Cleavage of the ras substrate by ribozyme RzC-ras is robust at both 2 mM and 5 mM MgCl2, proceeding at initial rates of 0.48 and 0.83 min-1, respectively, and reaching a calculated plateau of approximately 70%. At 0.5 mM MgCl2, the initial cleavage rate drops off sharply (40-fold reduced), although the calculated plateau is still approximately 70%, (Figure 5 and Table 2). It is possible that ribozyme RzC-ras initially folds into an active conformation when assayed at the highest concentrations of MgCl2, but that at the lowest concentrations of MgCl2 it initially assumes an inactive conformation. The rate-limiting step of the majority species under these conditions is then the inactive-to-active conformational conversion, which is facilitated by the presence of the TSM. In sum, several of the stabilizing features observed in RzC are also seen in other sequence contexts, including ones in which the substrate includes flanking nucleotides not paired with the ribozyme.
Figure 5 Summary of magnesium dependence of initial rates. Initial rates observed in 10 (gray bars), 5 (diagonally hatched bars), 2 (white bars) and 0.5 (black bars) mM MgCl2 are normalized to the rate observed in 5 mM MgCl2. The down arrow for RzCΔ2 indicates that no cleavage was observed under these conditions. Symbols above the bars indicate magnesium sensitivity: "++," moderate magnesium dependence indicative of tertiary stabilization; "+/-," intermediate magnesium dependence indicative of modest tertiary stabilization; "-," strong magnesium sensitivity indicative of there being no tertiary stabilization.
Table 2 Generalization of discontinuous hammerhead ribozyme design
[MgCl2], mM F∞ Fa ka kb init rate, min-1
RzC4 [37] 5 0.32 0.24 0.077
2 0.27 0.13 0.036
0.5 0.35 0.055 0.019
RzC-ras [37] 5 0.69 1.21 0.83
2 0.69 0.45 1.44 0.066 0.48
0.5 0.71 0.017 0.012
RzD [37] 5 0.52 0.125 0.065
2 0.0135 0.0175 0.0024
0.5 0.812 0.00032 0.00022
Generalization of discontinuous hammerhead design, following same conventions as in Table 1. Most data fit well to a single-exponential rate equation except for RzC-ras at 2 mM MgCl2, for which a double-exponential rate equation was used.
Importance of stem 1b stability
The lengths of stems 1 and 2 critically determine the positioning of the interacting nucleotides at the ends of each stem. For a given stabilized hammerhead ribozyme, variants with shorter or longer stems 1 are not expected to retain tertiary stabilization – their tertiary interaction elements would be nearer to (or farther from) the core, and the nucleotides in loop 1 would be rotated by approximately 30° around the helical axis for each base pair change in helix length. The sTRSV and RzC hammerhead ribozymes have 6 total base pairs in stem 1 (three each in stems la and 1b). In contrast, the hammerhead ribozyme from peach latent mosaic viroid (PLMVd) has only five base pairs in its stem 1. In this case a three-nucleotide loop 1 (UAA) interacts with a hexaloop (UAGAGU) in loop 2 [2,4]. Ribozyme RzD was designed to determine whether hammerhead ribozymes with 5 nucleotides in stem 1 could be converted from cis-cleavage to trans-cleavage by following the design strategy used to generate ribozyme RzC. Specifically, stem 1 was divided into a three base pair stem la (to promote intermolecular substrate binding) and a two base pair stem 1b (to preserve the original PLMVd tertiary interaction). Ribozyme RzD showed markedly sharper dependence on divalent magnesium than had been observed for RzC, with initial cleavage rates declining ~300-fold as MgCl2 was decreased from 5 mM to 0.5 mM. The 2 bp of stem 1b thus appear to be insufficiently stable to support the discontinuous design.
Conformational heterogeneity and generalizability of the design
It is common for hammerhead ribozymes to fold into heterogeneous populations in which a subset cleaves rapidly while the remaining RNAs convert over time into active fold which then cleave. This pattern is evident in the biphasic kinetics seen with several of the constructs described here (Figures 3 and 5; Tables 1 and 2). The tertiary stabilized, type 1 hammerhead ribozyme from the intestinal fluke parasite, Schistosoma mansoni (SMα1), also exhibits multiphasic kinetics indicative of conformational heterogeneity [19]. We observed similar results with the tertiary stabilized type 2 hammerhead ribozyme from the Dolichopoda cave cricket, and found that this heterogeneity was largely eliminated by changing a few nucleotides within stems 1 and 3 (M. Roychowdhury & D. Burke, in preparation). While conformational heterogeneity affects initial rates and the chemical interpretation of catalytic mechanism, it is of minimal relevance to modulation of gene expression by intracellularly expressed ribozymes so long as the target RNA can be cleaved fast enough to exert the desired biological effect. The discontinuous design used in ribozyme RzC yields rapid, multiple-turnover cleavage at physiological concentration of divalent magnesium, thereby demonstrating its potential utility for use inside cells.
While this manuscript was in preparation, Weinberg and Rossi described a slightly different design for hammerheads with discontinuous stems 1, in which additional base pairs are allowed to form between ribozyme and substrate by introducing a new stem at the discontinuity [20]. Although those authors only evaluated single-turnover cleavage under conditions of high divalent magnesium (10 mM), the results described here suggest that the Weinberg and Rossi design may also allow cleavage at sub-millimolar concentrations of magnesium.
Conclusion
The "Discontinuous Stem 1" design described here takes advantage of natural stabilizing tertiary interactions in approximately native structural contexts to enable trans-cleavage of model substrates at physiological concentrations of MgCl2. The design was particularly effective for ribozymes derived from sTRSV hammerhead ribozyme, and markedly less effective for a ribozyme derived from PLMVd. Hammerhead ribozyme RzC cleaved its 13 nucleotide target substrate effectively at MgCl2 concentrations as low as 0.2 mM at 25°C and 0.5 mM at 37°C. Catalytic activity was reduced or eliminated at sub-millimolar MgCl2 concentrations for ribozymes in which tertiary interactions are diminished (RzCΔ1) or removed (RzCΔ2). Two additional ribozymes that altered the internal guide sequence, including one targeted to the human ras oncogene, were active in low concentrations of MgCl2, demonstrating that this design could be adapted for use against other targets. We envision that trans-cleaving ribozymes that exploit the tertiary stabilizing motifs of sTRSV or other type II or type III hammerheads could be used in gene therapies or other intracellular applications.
Methods
Ribozymes, RNA substrates, and oligonucleotides
RNA substrates were synthesized by Dharmacon (Lafayette, CO) and DNA oligonucleotides by Integrated DNA Technologies (Coralville, IA). Ribozymes were synthesized by transcription in vitro from synthetic DNA templates, then radiolabeled and purified as described [21].
Determination of single-turnover kinetic parameters and initial rates
Kinetic analysis was carried out essentially as described [4]. Briefly, for single-turnover reactions, 10 pmol of end-labeled substrate RNA was mixed with 100 pmol ribozyme in Tris-HCl, pH 7.5 (final 50 mM in reaction), heated to 90°C for 1 min, then allowed to cool slowly to the reaction temperature. After equilibrating at either 25°C or 37°C for 5 min, one-tenth of the sample was removed as a zero time point and quenched in an equal volume of stop buffer (95% formamide, 20 mM EDTA, 0.5% bromophenol blue, 0.5% xylene cyanol). Cleavage reactions for the remainder of the sample were initiated by adding MgCl2 to the desired concentration. Aliquots were removed at various times and quenched in stop buffer. Cleaved and uncleaved substrate were separated on by denaturing (7 M urea) 12% polyacrylamide gel electrophoresis. Bands in the gels were quantified using the ImageQuant software from Molecular Dynamics. The data were first modeled by a single-exponential equation , where ft = fraction cleaved at a given time (cleaved/(cleaved + uncleaved)), f0 = zero point correction (essentially zero), f∞ = estimated plateau value at infinite time, and kobs = observed first-order rate constant. Those data sets that could not be adequately modeled as single-exponential processes were fit to a double-exponential process using the equation , where ka and kb are the kobs values for two exponentially decaying processes, and "a" is the fraction of the active ribozyme that partitions into the faster of the two processes. Initial cleavage rates were calculated by taking the first derivative of the single- or double-exponential equation above and solving for t = 0 to yield. For example, ignoring the f0 term, the initial rate for the double-exponential process is given by rate = f∞•a•ka + f∞•(1-a)•kb. Multiple-turnover reactions were carried out in a similar manner using 1 nM ribozyme and excess substrate (10 to 200 nM).
Abbreviations
TSM, tertiary stabilizing motif; sTRSV, satellite RNA of the Tobacco Ring-Spot Virus; PLMVd, peach latent mosaic viroid.
Authors' contributions
STG carried out the enzyme kinetic analysis. DHB conceived of the study, participated in its design and drafted the manuscript. Both authors read and approved the final manuscript.
Figure 4 Eadie-Hofstee plots for determining substrate binding affinities. Reaction velocity (vo) in micromoles•min-1 is normalized to total ribozyme concentration (1 nM). Indicated on each plot are the ribozyme species, the concentration of MgCl2 in millimolar (in parentheses) and the calculated Km value (negative of the slope). Gray trendlines for RzC at 0.5 mM MgCl2 subdivide the data set into regions of steep and shallow slopes with Km values of 90 nM and 12 nM.
Acknowledgements
The authors wish to thank Peter Unrau (Simon Frasier University), Manami Roychowdhury-Saha and Vanvimon Saksmerprome for comments on the manuscript, and Sugata Roychowdhury for insightful discussions early in the project. This work was supported by undergraduate research grants to S.T.G. from the Earl G. Sturdevant, Harry G. Day, Indiana University Honors College and the HHMI scholarship funds, and by grants to D.H.B. from the NIH under award AI45344 and from the David and Lucille Packard Interdisciplinary Science program.
==== Refs
Blount K Uhlenbeck O The structure-function dilemma of the hammerhead ribozyme Annu Rev Biophys Biomol Struct 2005 34 415 440 15869397 10.1146/annurev.biophys.34.122004.184428
Khvorova A Lescoute A Westhof E Jayasena SD Sequence elements outside the hammerhead ribozyme catalytic core enable intracellular activity Nat Struct Biol 2003 10 708 712 12881719 10.1038/nsb959
De la Peña M Gago S Flores R Peripheral regions of natural hammerhead ribozymes greatly increase their self-cleavage activity EMBO J 2003 22 5561 5570 14532128 10.1093/emboj/cdg530
Saksmerprome V Roychowdhury-Saha M Khvorova A Jayasena S Burke DH In vitro selection of trans-cleaving hammerhead ribozymes active in sub-millimolar magnesium RNA 2004 10 1916 1924 15547137 10.1261/rna.7159504
Penedo J Wilson T Jayasena S Khvorova A Lilley D Folding of the natural hammerhead ribozyme is enhanced by interaction of auxiliary elements RNA 2004 10 880 888 15100442 10.1261/rna.5268404
Canny M Jucker F Kellogg E Khvorova A Jayasena S Pardi A Fast cleavage kinetics of a natural hammerhead ribozyme J Am Chem Soc 2004 126 10848 10849 15339162 10.1021/ja046848v
Uhlenbeck O Less isn't always more RNA 2003 9 1415 1417 14623996 10.1261/rna.5155903
Hammann C Lilley D Folding and activity of the hammerhead ribozyme ChemBiochem 2002 3 690 700 12203967 10.1002/1439-7633(20020802)3:8<690::AID-CBIC690>3.0.CO;2-C
Kisseleva N Khvorova A Westhof E Schiemann O Binding of manganese(II) to a tertiary stabilized hammerhead ribozyme as studied by electron paramagnetic resonance spectroscopy RNA 2005 11 1 6 15611296 10.1261/rna.7127105
Suzumura K Takagi Y Orita M Taira K NMR-based reappraisal of the coordination of a metal ion at the pro-Rp oxygen of the A9/G10.1 site in a hammerhead ribozyme J Am Chem Soc 2004 126 15504 15511 15563179
Willcocks J Mulquiney P Ellory J Veech R Radda G Clarke K Simultaneous determination of low free Mg2+ and pH in human sickle cells using 31P NMR spectroscopy J Biol Chem 2002 277 49911 49920 12297506 10.1074/jbc.M207551200
Uetani T Matsubara T Nomura H Murohara T Nakayama S Ca2+-dependent modulation of intracellular Mg2+ concentration with amiloride and KB-R7943 in pig carotid artery J Biol Chem 2003 278 47491 47497 12958315 10.1074/jbc.M307898200
Cohen S Burt C 31P nuclear magnetic relaxation studies of phosphocreatine in intact muscle: determination of intracellular free magnesium Proc Natl Acad Sci U S A 1977 74 4271 4275 270670
Millart H Durlach V Durlach J Red blood cell magnesium concentrations: analytical problems and significance Magnes Res 1995 8 65 76 7669509
Mildvan A Role of magnesium and other divalent cations in ATP-utilizing enzymes Magnesium 1987 6 28 33 3029516
Yadava R Mahen E Fedor M Kinetic analysis of ribozyme-substrate complex formation in yeast RNA 2004 10 863 879 15100441 10.1261/rna.5234204
Prody G Bakos J Buzayan J Schneider I Bruening G Autolytic processing of dimeric plant virus satellite RNA Science 1986 231 1577 1580
Scherr M Grez M Ganser A Engels J Specific hammerhead ribozyme-mediated cleavage of mutant N-ras mRNA in vitro and ex vivo. Oligoribonucleotides as therapeutic agents J Biol Chem 1997 272 14304 14313 9162065 10.1074/jbc.272.22.14304
Osborne E Schaak J DeRose V Characterization of a native hammerhead ribozyme derived from schistosomes RNA 2005 11 187 196 15659358 10.1261/rna.7950605
Weinberg M Rossi J Comparative single-turnover kinetic analyses of trans-cleaving hammerhead ribozymes with naturally derived non-conserved sequence motifs FEBS Lett 2005 579 1619 1624 15757651 10.1016/j.febslet.2005.02.016
Burke DH Ozerova NDS Nilsen-Hamilton M Allosteric hammerhead TRAP ribozymes Biochemistry 2002 41 6588 6594 12022862 10.1021/bi0201522
Leontis NB Westhof E Geometric nomenclature and classification of RNA base pairs RNA 2001 7 499 512 11345429 10.1017/S1355838201002515
|
16095542
|
PMC1199579
|
CC BY
|
2021-01-04 16:26:24
|
no
|
BMC Biochem. 2005 Aug 12; 6:14
|
utf-8
|
BMC Biochem
| 2,005 |
10.1186/1471-2091-6-14
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-1821602949910.1186/1471-2105-6-182Methodology ArticleComparison of codon usage measures and their applicability in prediction of microbial gene expressivity Supek Fran [email protected]ček Kristian [email protected] Department of Molecular Biology, Division of Biology, Faculty of Science, Zagreb University, Rooseveltov trg 6, 10000 Zagreb, Croatia2 Protein Structure and Bioinformatics, International Centre for Genetic Engineering and Biotechnology, Padriciano 99, 34012 Trieste, Italy2005 19 7 2005 6 182 182 11 2 2005 19 7 2005 Copyright © 2005 Supek and Vlahoviček; licensee BioMed Central Ltd.2005Supek and Vlahoviček; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
There are a number of methods (also called: measures) currently in use that quantify codon usage in genes. These measures are often influenced by other sequence properties, such as length. This can introduce strong methodological bias into measurements; therefore we attempted to develop a method free from such dependencies. One of the common applications of codon usage analyses is to quantitatively predict gene expressivity.
Results
We compared the performance of several commonly used measures and a novel method we introduce in this paper – Measure Independent of Length and Composition (MILC). Large, randomly generated sequence sets were used to test for dependence on (i) sequence length, (ii) overall amount of codon bias and (iii) codon bias discrepancy in the sequences. A derivative of the method, named MELP (MILC-based Expression Level Predictor) can be used to quantitatively predict gene expression levels from genomic data. It was compared to other similar predictors by examining their correlation with actual, experimentally obtained mRNA or protein abundances.
Conclusion
We have established that MILC is a generally applicable measure, being resistant to changes in gene length and overall nucleotide composition, and introducing little noise into measurements. Other methods, however, may also be appropriate in certain applications. Our efforts to quantitatively predict gene expression levels in several prokaryotes and unicellular eukaryotes met with varying levels of success, depending on the experimental dataset and predictor used. Out of all methods, MELP and Rainer Merkl's GCB method had the most consistent behaviour. A 'reference set' containing known ribosomal protein genes appears to be a valid starting point for a codon usage-based expressivity prediction.
==== Body
Background
As the numbers of sequenced genes grew, it became evident that synonymous codons are not used equally [1-3]. Codon frequencies were found to vary on 3 levels: between genomes, between genes in the same genome, and within a single gene [4]. Many factors have been shown to influence codon usage patterns, the most important being: (i) overall nucleotide composition of the genome, reflecting mutational biases; (ii) selective forces acting on highly expressed genes to improve efficiency of translation [5]; and (iii) horizontal gene transfer, with transferred genes retaining the codon frequencies of their former host [6]. Connections have also been demonstrated between codon usage and: (i) gene length [7]; (ii) location on the chromosome [8]; (iii) the strand it resides on [9]; (iv) need for specific secondary structures in mRNA [10]; and (v) characteristics of the gene's protein product, such as its hydrophobicity [11] or secondary structure elements [12].
Moreover, the relative influence of each of these factors varies from genome to genome, and from gene to gene. For example, selection for translation efficiency shapes codon usage more in fast-growing microbes [13] than in slow-growing ones [14]. In contrast, codon usage of human genes depends largely on GC richness of the chromosomal region (isochore) [15]. It is still unclear to what extent other elements contribute to the genes' codon usage patterns [16]. The multitude of influences on codon preferences, as well as high dimensionality of codon usage data, necessitated the development of various measures (also called: statistics) of codon usage.
Many researchers in this field formulated their own measures, which led to a large number of available methods [17,18] for codon usage analysis. Unfortunately, these methods are not universally applicable, as their behaviour tends to be context-dependant. They may exhibit strong artefacts with varying (i) sequence length, (ii) overall amount of codon bias and (iii) codon bias discrepancy (see Results and Discussion for an explanation). Previous works [19,20] discussed this issue and compared some of the commonly used measures available at the time. Our aim was to develop and test a measure that would be free from dependence on the aforementioned contexts. Also, we attempted to verify the usefulness of such a measure by employing it to predict gene expressivity in microbial genomes.
Results & discussion
The "Measure Independent of Length and Composition" (MILC)
Our primary motivation in developing this novel method was to correct for possible artefacts due to sequence length variability. The measure should be able to quantify the distance in codon usage between a gene and some expected distribution of codons. The codon distribution could either be calculated from the background nucleotide composition, or derived from a single gene or a gene group. Therefore, MILC is conceptually similar to Karlin and Mrazek's B [21], Novembre's ENC' [19] or Urrutia and Hurst's MCB method [22].
Mathematically, the measure is based on a log-likelihood ratio score used in the statistical G-test for goodness-of-fit. This methodology yields numerically similar results to the more commonly used χ2 test, but may hold theoretical advantages over it in statistical analyses [23]. Both of the methods have been used in past examinations of codon usage patterns [24,25].
The individual contribution Ma of each amino acid a to the MILC statistic is calculated as
where Oc denotes the actual observed count of the codon c in a gene, and Ec stands for the expected count of the same codon. The Oc/Ec ratio is mathematically equal to, and can be replaced by fc/gc, where fc is the frequency of the codon c in a gene, and gc is the expected frequency of the same codon. The sum of f or g over all codons for each amino acid should equal 1. The total difference in codon usage is then assessed by the following formula:
The sum of contributions of all amino acids (stop codons are excluded from calculation) is divided by L, the gene length in codons, in attempt to compensate for the expected increase with total number of codons. This is analogous to the procedure described in [25]. However, such a „scaled χ2" statistic still depends on gene length [20], greatly overestimating the overall amount of bias in shorter sequences. The correction factor C in Equation 2 attempts to correct for this overestimation.
The cause for the abovementioned effect are sampling errors: a relatively small number of observations (counted codons) cannot exactly fit the expected distribution, leading to a higher perceived χ2 score. In order to demonstrate the effect, let us presume that the expected codon frequencies for two cysteine codons are g(UGU) = 0.5 and g(UGC) = 0.5; and that our hypothetical gene complies with these codon frequencies. However, a short gene might have only a single codon for Cys, thus the observed counts can be only OUGU = 1 and OUGC = 0, or vice versa. Either way, instead of being equal to 0, the cysteine's contribution to the χ2 score will be:
In case the gene has two cysteines, there is a 50% chance that OUGU = OUGC = 1, which would yield a (correct) χ2 score of 0; and a 50% chance that one of them will be 2, and the other 0, which gives a χ2 score of 2. The weighted average of these scores will again be equal to 1. Moving on to cases with 3, 4 or more cysteines we see that always MCys = 1, and it can be shown that for each amino acid in this case Ma is equal to its degree of redundancy minus 1 (e.g. MIle = 2, MPro = 3). In fact, this is the expected value of the χ2 statistic under the null hypothesis (observed frequencies match the expected frequencies), which equals the number of degrees of freedom. The calculation can be generalized to cases when the observed frequencies do not match the expected codon distribution, and is also applicable to the G statistic MILC is based upon. Further examples to better illustrate this point are given in the material accompanying this paper [see Additional file 1].
To reiterate, in a situation where the gene's codon usage matches the expected distribution, with all amino acids present, the sampling errors will increase the χ2 score by 41, and the „scaled χ2" by 41/L. The correction factor C is therefore calculated as:
where ra is the number of possible codons for the amino acid a – its degeneracy class. Only the amino acids actually present at least once in the sequence contribute to C, e.g. if a gene missed one of the four-fold amino acids, C would be 38/L + 0.5. When the observed frequencies match the expected codon distribution closely, MILC can assume negative values. In order to compensate, a constant of 0.5 is added to the correction factor C (see Equation 4). Regarding minimum sequence length, we recommend that only sequences of 80 codons or longer be analysed using MILC (or any other measure of codon usage); many researchers set this threshold to even higher values, such as 100.
Behaviour of codon usage measures under varying conditions
A multitude of methods to measure codon usage has been published, including "scaled χ2" [25], "effective number of codons" ENC [26], "codon bias index" CBI [27], "intrinsic codon bias index" ICDI [28], two versions of "codon bias" B [21,29], "maximum likelihood codon bias" MCB [22], "effective number of codons prime" ENC' [19], and "synonymous codon bias orderliness" SCUO [30]. Among those, we chose to test the methods that have been either frequently used in codon usage examinations, or that are new and haven't been extensively tested [20].
ENC is an older, widely accepted measure that quantifies the degree of deviation from equal use of synonymous codons; ENC' gives results comparable to ENC but allows comparison to any desired codon distribution; the 1998 version of Karlin and Mrazek's B has been used extensively in later research of microbial genomes by the same authors; MCB is a method conceptually similar to B, used in examinations of human genes; and SCUO is a representative of the information theory-based measures, which have recently been used on several occasions [31,32] to analyze codon usage. Finally, the method proposed in this paper, MILC, is compared in performance to the aforementioned methodologies.
Figure 1 demonstrates the behaviour of the methods when examining genes of differing lengths. Pseudorandomly generated sequences (or 'genes') obtained using INCA [33] were used for testing under varying conditions (see Methods): Figures 1a, 1c and 1e show the performance (degree of misestimation) for chosen measures at 5 different lengths, with 1b, 1d and 1f showing the standard deviations for the 10000 measurements performed at each length. In this aspect, our testing conditions resemble the ones previously used by Comeron and Aguade [20] or Novembre [19], the essential difference being the normalization and comparison of the results. Here, the values are presented as percentages of the 'dynamic range' of a measure (the largest difference between its high and low values under realistic conditions, see Methods). We feel this is more reasonable than e.g. normalizing a mean of the sample at a certain length by simply dividing it by the value at 2500 codons, which (i) unfairly penalizes measures which approach zero as bias lessens, as opposed to those approaching an arbitrary value, e.g. 61 for ENC and ENC', and (ii) among the measures approaching zero, favours those displaying larger values at 2500 codons, in spite of this being an undesirable quality – the value should be as close to zero as possible. For instance, both B and MCB are meant to equal 0 when expected and observed codon frequencies match, however in practice at the length of 2500 codons B assumes the value of approx. 0.1, and MCB of 0.033 (Table 2, "None" dataset). Dividing the misestimation of each measure by the above values would be unfairly advantageous for B; a more extreme example is ENC with its baseline value of 60.9. These issues are addressed by expressing the results as percentages of the dynamic range – a simple linear transformation essential for objective comparison of the methods' performances. However, when using a single measure to compare genes (or gene groups), or to determine association with other genomic data, it should not matter if the normalization is performed or not. The relative distances of codon usage in two genes (gene groups) would remain equal in both cases, and the degree of correlation with other genomic data would also not change.
Table 2 Determining the 'dynamic range' for measures of codon usage
dataset method max mean coef var dataset method min mean coef var dyn range
High-2 ENC 26.1757 0.3073 None ENC 60.9141 0.1390 -34.738
High-2 B | None 1.0250 0.0155 None B | None 0.0998 0.0118 0.925
High-2 MCB | None 3.0810 0.0783 None MCB | None 0.0330 0.0078 3.048
High-2 ENC' | None 26.1757 0.3073 None ENC' | None 60.9141 0.1390 -34.738
High-2 MILC | None 1.9410 0.0389 None MILC | None 0.5000 0.0037 1.441
High-2 SCUO 0.5470 0.0146 None SCUO 0.0068 0.0016 0.540
Figure 1 Effect of sequence length on behaviour of codon usage measures. Figure 1a, 1c and 1e illustrate the degree of misestimation of a measure at varying sequence lengths (x axis), compared to the values at 2500 codons.. The values were obtained by calculating means for 10000 randomly generated sequences per method per length, and are expressed as percentages of the measures' dynamic range (see Methods). Figures 1b, 1d and 1f display standard deviations of the same 10000 measurements, as percentage of the dynamic range; higher values mean a method is more 'noisy'. In Figures 1a and 1b we generated sequences unbiased in use of codons and compared them to a frequency table also assuming equal use ('None', see Methods'). In 1c and 1d both the sequences and the expected frequency were, on overall, biased ('Med-1'); in 1e and 1f the sequences were biased ('Med-1'), but were compared to an unbiased expected frequency table ('None').
We designed three experiments to determine to what extent changing gene length affects each measure. In the first experiment (Figures 1a and 1b) the expected distribution assumes equal codon frequencies ("None", see Methods) and the generated sets of genes attempt to mimic that distribution. Therefore, the methods should ideally report a minimal distance between the observed and the expected distribution. ENC, ENC', MILC and MCB are generally well behaved under these conditions and tend to somewhat overestimate the amounts of bias in short sequences, MCB overestimates bias also in longer sequences. In contrast, B and SCUO greatly overestimate the bias in shorter genes (by "shorter" we assume a range of gene lengths most frequent in genomes, e.g. 100–500 codons). For example, using B on sequences 250 and 500 codons long would result in the first sequence being seemingly different twice as much from the expected distribution as the second one. Moreover, the overestimation at 250 codons may amount to as much as a quarter of the dynamic range of B. As anticipated, the variability of all measures (Figure 1b) decreases with an increase in gene length. It must be noted that MCB measurements introduce significantly less noise than the rest of the methods, particularly in short genes.
The second experiment, where the overall amount of bias in both the generated sequences and the expected distribution increases (Figure 1c) shows little change regarding length dependence – all methods see a very modest improvement in performance. ENC now tends to slightly underestimate bias, however, the variability chart (Figure 1d) shows that here it becomes noticeably less reliable than other methods, and so does SCUO. MCB is still the best performer, followed by MILC and B for shorter sequences, and ENC' for longer ones.
Figures 1e and 1f, representing the third experiment, demonstrate what happens when a gene unbiased in codon usage differs from the biased expected codon frequencies, derived from the "Med-1" dataset (see Methods). This is, in fact, a situation more likely to occur in real-life applications, as a gene would probably show at least some deviation from the expected codon distribution. ENC and SCUO expectedly behave precisely the same as in 1a and 1b, because they by definition always assume an unbiased expected distribution. Interestingly, B improves significantly and does not feel as much influence of gene length when the observed and expected codon distributions differ. It now performs on par with ENC' and MCB, both of which show a detrimental effect of increasing distance between the observed and the expected distribution. This factor also increases the amount of variation introduced by measures (excluding ENC and SCUO), most of all ENC', and causes MCB to lose its advantage over MILC and B.
We have shown that ENC and ENC' display a drop in reliability as the overall amount of bias (measured by ENC, Figure 1d), or the difference in bias (measured by ENC', 1f) increases. The explanation is the cutoff value that both measures introduce [19,26], causing the distribution of the measurements to become asymmetrical and therefore artificially reducing the measures' variance when the observed codon distribution is close to the expected one. Having such a threshold might, in theory, mask biologically relevant information; for an example, see the ENC' plot in Figure 2
Figure 2 Plots of the E. coli genome made using different measures of codon usage. The four plots were made by using measures that allow an expected codon distribution to be specified: B, MCB, ENC' or MILC. The distance of codon usage of a gene from E. coli ribosomal genes was plotted on the x axis, and the distance of codon usage of a gene from the average codon usage of E. coli was plotted on the y axis. A characteristic 'crescent moon' shape is seen on all four plots. White square represent ribosomal protein genes, while all other genes are represented by grey squares.
Measures of codon usage introduce different levels of statistical bias in shorter genes; however, it must be noted that even if this influence were completely eliminated, there might still exist a connection between codon bias and length caused by the inherent properties of the sequences. Selection might be acting to optimize codon usage patterns (and therefore translational efficiency) in energetically costly longer genes; on the other hand it might also act to reduce the size of highly expressed (and strongly biased) proteins [7]. The only way to nullify these length effects – if this is desired – is to use regression, while employing a length-insensitive measure.
In addition to being resistant to length variation, the methods should ideally be invariant to both overall bias and the relative difference in codon usage. Moreover, the measures should be commutative with respect to properties of the observed and expected distributions. We designed two experiments to investigate these issues.
Figure 3a shows the influence of overall amount of codon bias ('background nucleotide composition') on performance of the individual methods: we examined sets of 10000 sequences generated to match the expected frequencies at varying degrees of bias; the sequences were 2500 codons long to eliminate gene length effects. The baseline value was determined by comparing unbiased ("None") genes to unbiased ("None") expected frequencies. ENC and SCUO report higher differences from the baseline as the overall bias increases, which is anticipated since overall bias is exactly what the two methods attempt to quantify. The other methods' results should not vary between datasets. Indeed, ENC', MILC and MCB have proven to be independent of this factor, while B only slightly decreases as overall bias rises.
Figure 3 Effect of overall amount of bias on behaviour of codon usage measures. Figure 3a describes the change in behaviour of each measure as the overall bias increases from unbiased ('None') to a nucleotide composition noted on the x axis. The values were obtained from 10000 randomly generated sequences, 2500 codons long, per frequency table (None, Low-1, Low-2 etc.) per measure. Figure 3b demonstrates how the measures react when the nucleotide compositions of the generated sequences and the expected codon frequency table are interchanged (commutative property). See Results and Discussion for further explanation. In both figures, the values on the y axis are expressed as percentages of the measures' dynamic ranges.
Furthermore, in order to test the commutative property, using each measure we compared datasets with varying levels of bias to the "None" expected distribution, and vice versa. Theoretically, when using many long sequences, comparing "None" genes to, for instance, "Med-1" expected distribution should yield the same result as comparing "Med-1" genes to the "None" expected distribution. In Figure 3b we show that among the measures that allow comparisons, the only one handling this appropriately was Karlin and Mrazek's B. MILC is less sensitive than ENC' and especially MCB, which displays a polar effect, being more strongly influenced by changes in the overall bias in the expected frequencies.
In genomes, individual amino acids may vary in amount of codon bias, an occurrence termed 'codon bias discrepancy', best described by the phrase "some codons are more optimal than others" in Fuglsang's paper [34]. For instance, in E. coli the CGU and CGC codons for arginine are strongly preferred over the other four codons, while six codons for serine are chosen more uniformly, with a mild preference for AGC over the others.
It has been implied that ENC may be dependant on the strength of the codon biasdiscrepancy [35], and the same limitations are expected to apply to the ENC' due to the similarities in calculation of the two statistics. Based on two frequency tables adopted from Fuglsang [35], representing examples of moderately biased codon distributions with and without discrepancy, we generated genes of varying lengths and compared them to a uniform distribution of codons. Figure 4a demonstrates that this amount of discrepancy causes most of the methods to moderately overestimate overall bias (10–15% of the dynamic range), while B is less affected by this change. Figure 4b illustrates a similar situation, however this time we performed the test using our own codon distribution, "Med-1d", that preserves the GC3s content of the "Med-1" while introducing discrepancy (see Methods). All of the methods again overestimated bias, although to a lesser degree; relations between methods remain similar. It is still undetermined to which extent amino acids differ in degree of bias in real genomes, and our tests do not indicate too strong an influence of this issue on measures of codon usage.
Figure 4 Effect of codon bias discrepancy on behaviour of codon usage measures. The figure shows how the measures react to codon discrepancy, i.e. when the amino acids within a sequence differ in amounts of bias. The value on the y axis is the amount of overestimation (in % of the methods' dynamic ranges) that occurs as discrepancy is introduced; this was determined by examining 10000 generated sequences for each length (x axis) and method. Figure 4a uses frequency tables adopted from Fuglsang [35], and 4b uses the authors' own frequency tables.
Improving prediction of microbial gene expressivity
Analogous to Karlin and Mrazek's method of predicting expression levels of genes [36], we formulate a statistic named MELP (MILC-based Expression Level Predictor), computed simply as the ratio of respective distances of a gene's codon usage from the genomic average, and a predefined reference set:
This novel method of quantitatively predicting gene expressivity is then compared to existing methods: CAI [37], Fop [1], E [36] and GCB [38]. Instead of testing for context-independence, as we did with general measures of codon usage, we chose to rate the expression level predictors by how well they approximate real-world observations. We have collected datasets, listed in Table 3 (Methods), which consist of either mRNA or protein abundance data for unicellular organisms obtained by different methods – mostly cDNA microarrays, but also by Affymetrix arrays (Pfa-2, and partly Sce-3 data), SAGE (also partly in Sce-3), and a number of quantitative proteomics techniques. This was done in order to assemble a collection of heterogeneous data large enough to allow a rough comparison of codon usage-based predictors of gene expression. Since we wanted to avoid making any assumptions about the distributions of data in each dataset, we used a nonparametric statistic, Spearman's (rank) correlation coefficient, to quantify agreement with predicted expression levels (Figure 5). We also tried calculating Pearson (linear) correlation coefficients for the data, which in some cases showed significant improvement by log-transforming the data, however this effect was not observed consistently among datasets or expression predictors [see Additional file 1].
Table 3 Transcript/protein abundance data used for validation of expression level predictors
name type N ref Web source Files / accessions medium
Saccharomyces cerevisiae
Sce-1 prot 2014 [51] 1.ref-abund.xls, column G rich
Sce-2 prot 3960 [52] nature02046-s2.xls rich (YEPD)
Sce-3 mRNA 5432 [51] 1.ref-abund.xls, column B combined data
Escherichia coli K-12 MG1655
Eco-1 prot 138 [46] tables A1, A2, A3 minimal
Eco-2 prot [79] 47 columns AB, RIC rich
Eco-3 prot 69 [47] columns PHNppm, PSppm, NSppm minimal (MOPS, glucose)
Eco-4 mRNA 2597 [53] 3181Table6.xls, column D rich (LB)
Eco-5 mRNA 3685 [54] EXPSET003: PALSP01-PALSP11 minimal (MOPS, glucose)
Escherichia coli K-12 W3110
Ecj-6 mRNA 3788 [55] ex298 – ex320, ex328-ex334
Bacillus subtilis
Bsu-1 mRNA 3581 [56] ex745 – ex749 rich (LB)
Bsu-2 mRNA 3590 [57] ex264, ex265, ex272, ex273, ex275, ex276, ex278 – ex286 rich (LB)
Bsu-3 mRNA 3577 [58] ex940 – ex945 DSM
Synechocystis sp. PCC6803
Syn-1 mRNA 2840 [59] ex832 – ex839 low light conditions
Syn-2 mRNA 2840 [60] ex22, 23, 24, 44
Plasmodium falciparum 3D7
Pfa-1 prot 1068 [61] nature01107-s1.xls average of 4 life stages
Pfa-2 mRNA 2081 [62] Table_1, columns I, K, Q, AB, AD, AJ, AO, AQ average of 4 life stages
Figure 5 Performance of codon usage-based expression level predictors. Height of the columns shows the Spearman's (rank) correlation coefficient for each gene expression dataset / predictor combination. Error bars illustrate the change in success of the prediction when the default reference set (consisting of ribosomal protein genes >100 codons) is replaced by a computationally generated one [44].
The agreement of predicted and actual protein/transcript levels varied greatly between all examined combinations of prediction method and dataset. The cause may lie in the quality of experimental data; for instance, mRNA abundances and protein 2D-PAGE data have been shown not to agree well in certain cases [39]; 2D-PAGE as a method may only be suitable for detection of abundant proteins [40], while microarray data tends to suffer from noise introduced at each step of different experimental protocols [41]. The other probable reason for relatively incoherent results is that a model for predicting gene expression from genomic data, based solely on codon usage, is oversimplified. Other factors, such as promoter strength and gene copy number should also be taken into account. Fortunately, optimal codon usage in genes seems to coincide with factors enhancing transcription – this is why it is possible to observe a correlation between codon usage (acting at translation level) and transcript abundances. Keeping these limitations in mind, it seems safe to say that, in comparison to other predictors, GCB and MELP behave more consistently throughout all datasets.
Transcript and/or protein levels in a cell are normally subject to regulation, as opposed to codon usage patterns, which are 'hard-coded' in the genome sequence. If we suppose the major force shaping gene-specific codon usage patterns in microbes is selection for translation efficiency, which operates in periods of fast competitive growth, it follows that codon usage will be 'optimised' for genes highly expressed in such periods. For that reason we chose datasets of organisms harvested in exponential growth phase, and without severe nutritional restrictions in the medium. For instance, the Bsu-2 datasets describes Bacillus harvested at OD600 ≅ 0.4 – 0.6; an analogous dataset [see Additional file 1] for bacteria harvested at OD600 ≅ 1.1 does not correlate so well with predicted expression levels (Pearson's correlation coefficient for MELP = 0.234 vs. 0.187, for GCB = 0.277 vs. 0.185). In addition, the growth conditions should match the organism's natural habitat. For instance, E. coli grown in a rich medium has gene expression levels closer to the predicted values than E. coli in a defined medium; should the data in Eco-2 dataset be replaced with data from MOPS+glucose grown cells [see Additional file 1], the Pearson's correlation coefficient for log-transformed data drops from 0.720 to 0.663 (MELP), or from 0.708 to 0.642 (GCB). Furthermore, nitrogen or phosphorus starvation of E. coli in the Eco-3 dataset reduces the correlation with predicted values (data not shown). Such connections between codon usage and gene expression under different conditions can be used to hypothesize about the exact 'natural' environment of a microbe [42].
Any codon usage-based prediction of gene expression relies on a prior definition of a 'reference set', consisting of highly expressed genes. Our reference sets were defined as all genes coding for ribosomal proteins, longer than 100 codons; other approaches to this issue exist. For instance, the original definition for CAI [37] listed a set of genes which have been empirically proven to be highly expressed in yeast and E. coli; Karlin and Mrazek [36] included transcription/translation related factors and chaperones in the reference set, in addition to the ribosomal protein genes; attempts have been made to detect major trends in codon usage by iterative computational methods [38,43] and use the results to define a reference set. We investigated to what extent reference set composition affects prediction of gene expression; the alternative reference sets used were obtained from Merkl [44] and generated by computationally detecting the major trend in codon usage in a genome. The sets normally contained ribosomal protein genes, elongation factors and energy metabolism genes; also photosynthesis genes in Synechocystis and histones in P. falciparum; such functional assignments for reference set genes were not unexpected. Under the assumption that the major trend is due to translational selection, the change in reference set composition should have theoretically resulted in improved prediction. However, the outcome was highly dependent on the genome examined, and the predictor used (shown as error bars in Figure 5). In some instances, the use of the alternative reference set resulted in poorer correlation. More high-quality transcript/protein abundance data would be required to reach a definite recommendation on forming a reference set.
Conclusion
We introduce a novel method, based on a corrected log-ratio chi-squared statistic, of measuring codon usage bias in genes or gene groups – MILC. By comparing its performance to other commonly used measures of codon usage in a variety of contexts, we have established that MILC is a generally applicable method, being resistant to changes in gene length and overall nucleotide composition, and introducing little noise into measurements. Other measures, however, may also be appropriate for specific purposes: B, when comparing very long sequences (groups of genes, whole genomes) which are expected to differ significantly in codon usage and/or exhibit bias discrepancy; or MCB, when comparing sequences of varying lengths but relatively similar in codon preferences. We have also evaluated the methods' ability to estimate gene expression levels by comparing them to actual mRNA/protein abundance data from several species. Out of the tested predictors, GCB and MELP exhibit the most consistent behaviour. A reference set defined simply by including ribosomal protein genes appears to be a valid starting point for expression level predictions in examined prokaryotes and unicellular eukaryotes, although one should be cautious when interpreting the results of such estimations. The MILC and MELP methods have been implemented in the version 2 of the INCA software, available from the bioinfo-hr.org website [45].
Methods
Performance evaluation
The measures of codon usage ENC, B, MCB, ENC' and SCUO were computed as in [26,21,22,19] and [30] respectively. The test sets of randomly generated sequences follow the nucleotide compositions proposed in [20], and are reviewed in Table 1. The amino acid frequencies were kept proportionate to their degeneracy class (number of codons coding for it in the standard genetic code), i.e. a 4-fold amino acid is used twice as often as a 2-fold amino acid. As a consequence of the imposed restriction on amino acid composition, the nucleotide ratios in Table 1 reflect the nucleotide composition at silent sites only. For each combination of gene length (100, 150, 250, 500, 1000 and 2500 codons) and nucleotide composition used, 10000 sequences were generated; each sequence was compared, using all measures, to an expected frequency table (derived from data in Table 1) and the mean and standard deviation for all measurements were determined. Generated sequences did not contain stop codons.
Table 1 Nucleotide composition of the generated sequences at silent sites
None Low-1 Low-2 Med-1 Med-2 High-1 High-2
f(A) 0.250 0.200 0.200 0.125 0.125 0.050 0.050
f(G) 0.250 0.300 0.200 0.375 0.125 0.450 0.050
f(C) 0.250 0.300 0.400 0.375 0.125 0.450 0.850
f(T) 0.250 0.200 0.200 0.125 0.125 0.050 0.050
Values in Figures 1, 3 and 4 are expressed as percentages of the 'dynamic range' of a method, the largest difference between its high and low values under realistic conditions. This was assessed by comparing, using each method, first a set of 10000 'None' sequences (2500 codons long) to the 'None' frequency table, and then a set of 10000 'High-2' sequences (2500 codons long) to the 'None' frequencies, and finally by subtracting the numbers; this process is summarized in Table 2. Because of this normalization process, positive values of the mean always signify overestimation of bias, even though, for instance, a higher value of ENC' normally means less bias.
The codon frequency tables used to generate sequences, derived from the None, Low, Med and High nucleotide compositions, are available in the accompanying materials [see Additional file 1], as well as the frequency tables used to test for codon usage discrepancy effects.
Predictors of gene expression
The expression level predictors CAI, E, and GCB were computed as in [37,36] and [38], respectively. When calculating the 'frequency of optimal codons' Fop, a codon with a relative adaptiveness (codon frequency divided by the frequency of the most frequent codon) larger than 0.9 was considered optimal. Experimental datasets used to investigate the performance of the predictors are listed in Table 1. Datasets Sce-1, 2, 3, and Eco-4 were used 'as-is' from the respective sources. Eco-1 dataset was created by combining molar abundances (column "N-abd") from Tables a1, a2 and a3 in [46]; if a gene occurred in more than one table, its final abundance value was calculated as an average of the two/three measurements. Eco-2 dataset was created from the E. coli Gene-Protein Database [47] by multiplying values in the "AB" column (abundances) with values in the "RIC" column (rich media) and dividing by the "MWc" column to obtain molar abundances. Eco-3 dataset was created by averaging the "PHNppm", "PSppm" and "NSppm" (control groups for phosphorus and nitrogen starvation experiments), and by dividing by the "MWc" column. Ecj-6, Bsu-1, 2, 3, Syn-1 and Syn-2 datasets were downloaded from the KEGG expression data repository [48] and were processed in the following manner: the local background ("Control-bkg") was subtracted from the signal intensity ("Control-sig") for each microarray spot in the control groups, and the resulting values were normalised to the sum of 106 per experiment. Finally, for each spot/gene a median value over all experiments in a dataset was calculated. The Pfa-1 dataset was created by averaging the sequence coverage of a protein over all four life stages; if a protein was not detected in a P. falciparum stage, its sequence coverage was assumed to equal 0. To create the Pfa-2 dataset, the columns I, K, AB and AD were averaged to obtain an mRNA abundance for the trophozoite, Q and AJ for the merozoite; column AO provided values for the gametocyte, and column AQ for the sporozoite. The final abundance values were again obtained by averaging the four life stages. Files containing coding regions of genes were downloaded from the NCBI ftp site [49] for the Eco, Sce, Pfa and Syn datasets, and from the KEGG ftp site [50] for the Ecj and Bsu datasets.
Authors' contributions
FS devised, tested and implemented the MILC and MELP methods. KV supervised the project and contributed in biological expertise. Both authors read and approved the final manuscript.
Supplementary Material
Additional File 1
Rationale behind the length correction of the MILC method, codon frequencies used for testing of codon usage measures, and performance of the expression level predictors. Sheets 1a, 1b and 1c demonstrate, by example, how the chi-square and G scores for amino acids of different degeneracy classes behave when the observed codon counts are small. Sheet 2 contains the codon frequency tables used in testing of the codon usage measures. Sheet 3 describes the performance of the expression level predictors, expressed as Spearman (rank) and Pearson (linear) correlation coefficients of the predicted values and experimentally obtained mRNA/protein abundance data sets.
Click here for file
Acknowledgements
FS thanks Rainer Merkl for helpful discussion and data used in manuscript preparation.
==== Refs
Ikemura T Correlation between the abundance of Escherichia coli transfer RNAs and the occurrence of the respective codons in its protein genes: a proposal for a synonymous codon choice that is optimal for the E. coli translational system J Mol Biol 1981 151 389 409 6175758 10.1016/0022-2836(81)90003-6
Grantham R Gautier C Gouy M Jacobzone M Mercier R Codon catalog usage is a genome strategy modulated for gene expressivity Nucleic Acids Res 1981 9 r43 74 7208352
Gouy M Gautier C Codon usage in bacteria: correlation with gene expressivity Nucleic Acids Res 1982 10 7055 7074 6760125
Hooper SD Berg OG Gradients in nucleotide and codon usage along Escherichia coli genes Nucleic Acids Res 2000 28 3517 3523 10982871 10.1093/nar/28.18.3517
Ikemura T Codon usage and tRNA content in unicellular and multicellular organisms Mol Biol Evol 1985 2 13 34 3916708
Lawrence JG Ochman H Molecular archaeology of the Escherichia coli genome Proc Natl Acad Sci U S A 1998 95 9413 9417 9689094 10.1073/pnas.95.16.9413
Moriyama EN Powell JR Gene length and codon usage bias in Drosophila melanogaster, Saccharomyces cerevisiae and Escherichia coli Nucleic Acids Res 1998 26 3188 3193 9628917 10.1093/nar/26.13.3188
Daubin V Perriere G G+C3 structuring along the genome: a common feature in prokaryotes Mol Biol Evol 2003 20 471 483 12654929 10.1093/molbev/msg022
Lafay B Lloyd AT McLean MJ Devine KM Sharp PM Wolfe KH Proteome composition and codon usage in spirochaetes: species-specific and DNA strand-specific mutational biases Nucleic Acids Res 1999 27 1642 1649 10075995 10.1093/nar/27.7.1642
Seffens W Digby D mRNAs have greater negative folding free energies than shuffled or codon choice randomized sequences Nucleic Acids Res 1999 27 1578 1584 10075987 10.1093/nar/27.7.1578
D'Onofrio G Jabbari K Musto H Bernardi G The correlation of protein hydropathy with the base composition of coding sequences Gene 1999 238 3 14 10570978 10.1016/S0378-1119(99)00257-7
Oresic M Shalloway D Specific correlations between relative synonymous codon usage and protein secondary structure J Mol Biol 1998 281 31 48 9680473 10.1006/jmbi.1998.1921
Karlin S Mrazek J Campbell A Kaiser D Characterizations of highly expressed genes of four fast-growing bacteria J Bacteriol 2001 183 5025 5040 11489855 10.1128/JB.183.17.5025-5040.2001
Lafay B Atherton JC Sharp PM Absence of translationally selected synonymous codon usage bias in Helicobacter pylori Microbiology 2000 146 ( Pt 4) 851 860 10784043
Sharp PM Averof M Lloyd AT Matassi G Peden JF DNA sequence evolution: the sounds of silence Philos Trans R Soc Lond B Biol Sci 1995 349 241 247 8577834
Urrutia AO Hurst LD The signature of selection mediated by expression on human genes Genome Res 2003 13 2260 2264 12975314 10.1101/gr.641103
Moriyama EN Encyclopedia of the Human Genome: Codon Usage
Ermolaeva MD Synonymous codon usage in bacteria Curr Issues Mol Biol 2001 3 91 97 11719972
Novembre JA Accounting for background nucleotide composition when measuring codon usage bias Mol Biol Evol 2002 19 1390 1394 12140252
Comeron JM Aguade M An evaluation of measures of synonymous codon usage bias J Mol Evol 1998 47 268 274 9732453
Karlin S Mrazek J Campbell AM Codon usages in different gene classes of the Escherichia coli genome Mol Microbiol 1998 29 1341 1355 9781873 10.1046/j.1365-2958.1998.01008.x
Urrutia AO Hurst LD Codon usage bias covaries with expression breadth and the rate of synonymous evolution in humans, but this is not evidence for selection Genetics 2001 159 1191 1199 11729162
Rohlf FJ Sokal RR Biometry 1994 W. H. Freeman
Sharp PM Tuohy TM Mosurski KR Codon usage in yeast: cluster analysis clearly differentiates highly and lowly expressed genes Nucleic Acids Res 1986 14 5125 5143 3526280
Shields DC Sharp PM Synonymous codon usage in Bacillus subtilis reflects both translational selection and mutational biases Nucleic Acids Res 1987 15 8023 8040 3118331
Wright F The 'effective number of codons' used in a gene Gene 1990 87 23 29 2110097 10.1016/0378-1119(90)90491-9
Morton BR Codon use and the rate of divergence of land plant chloroplast genes Mol Biol Evol 1994 11 231 238 8170364
Freire-Picos MA Gonzalez-Siso MI Rodriguez-Belmonte E Rodriguez-Torres AM Ramil E Cerdan ME Codon usage in Kluyveromyces lactis and in yeast cytochrome c-encoding genes Gene 1994 139 43 49 8112587 10.1016/0378-1119(94)90521-5
Karlin S Mrazek J What drives codon choices in human genes? J Mol Biol 1996 262 459 472 8893856 10.1006/jmbi.1996.0528
Wan XF Xu D Kleinhofs A Zhou J Quantitative relationship between synonymous codon usage bias and GC composition across unicellular genomes BMC Evol Biol 2004 4 19 15222899 10.1186/1471-2148-4-19
Wang HC Badger J Kearney P Li M Analysis of codon usage patterns of bacterial genomes using the self-organizing map Mol Biol Evol 2001 18 792 800 11319263
Zeeberg B Shannon information theoretic computation of synonymous codon usage biases in coding regions of human and mouse genomes Genome Res 2002 12 944 955 12045147 10.1101/gr.213402
Supek F Vlahovicek K INCA: synonymous codon usage analysis and clustering by means of self-organizing map Bioinformatics 2004 20 2329 2330 15059815 10.1093/bioinformatics/bth238
Fuglsang A The effective number of codons for individual amino acids: some codons are more optimal than others Gene 2003 320 185 190 14597402 10.1016/S0378-1119(03)00829-1
Fuglsang A The 'effective number of codons' revisited Biochem Biophys Res Commun 2004 317 957 964 15081433 10.1016/j.bbrc.2004.03.138
Karlin S Mrazek J Predicted highly expressed genes of diverse prokaryotic genomes J Bacteriol 2000 182 5238 5250 10960111 10.1128/JB.182.18.5238-5250.2000
Sharp PM Li WH The codon Adaptation Index--a measure of directional synonymous codon usage bias, and its potential applications Nucleic Acids Res 1987 15 1281 1295 3547335
Merkl R A survey of codon and amino acid frequency bias in microbial genomes focusing on translational efficiency J Mol Evol 2003 57 453 466 14708578 10.1007/s00239-003-2499-1
Gygi SP Rochon Y Franza BR Aebersold R Correlation between protein and mRNA abundance in yeast Mol Cell Biol 1999 19 1720 1730 10022859
Gygi SP Corthals GL Zhang Y Rochon Y Aebersold R Evaluation of two-dimensional gel electrophoresis-based proteome analysis technology Proc Natl Acad Sci U S A 2000 97 9390 9395 10920198 10.1073/pnas.160270797
Schuchhardt J Beule D Malik A Wolski E Eickhoff H Lehrach H Herzel H Normalization strategies for cDNA microarrays Nucleic Acids Res 2000 28 E47 10773095 10.1093/nar/28.10.e47
Wagner A Inferring lifestyle from gene expression patterns Mol Biol Evol 2000 17 1985 1987 11110914
Jansen R Bussemaker HJ Gerstein M Revisiting the codon adaptation index from a whole-genome perspective: analyzing the relationship between gene expression and codon occurrence in yeast using a variety of models Nucleic Acids Res 2003 31 2242 2251 12682375 10.1093/nar/gkg306
Merkl R Personal communication 2004
Bioinfo-hr.org website
Link AJ Robison K Church GM Comparing the predicted and observed properties of proteins encoded in the genome of Escherichia coli K-12 Electrophoresis 1997 18 1259 1313 9298646 10.1002/elps.1150180807
VanBogelen RA Abshire KZ Moldover B Olson ER Neidhardt FC Escherichia coli proteome analysis using the gene-protein database Electrophoresis 1997 18 1243 1251 9298644 10.1002/elps.1150180805
Nakao M Bono H Kawashima S Kamiya T Sato K Goto S Kanehisa M Genome-scale Gene Expression Analysis and Pathway Reconstruction in KEGG Genome Inform Ser Workshop Genome Inform 1999 10 94 103 11072346
NCBI Genomes FTP site
KEGG Genomes FTP site
Greenbaum D Colangelo C Williams K Gerstein M Comparing protein abundance and mRNA expression levels on a genomic scale Genome Biol 2003 4 117 12952525 10.1186/gb-2003-4-9-117
Ghaemmaghami S Huh WK Bower K Howson RW Belle A Dephoure N O'Shea EK Weissman JS Global analysis of protein expression in yeast Nature 2003 425 737 741 14562106 10.1038/nature02046
Bernstein JA Khodursky AB Lin PH Lin-Chao S Cohen SN Global analysis of mRNA decay and abundance in Escherichia coli at single-gene resolution using two-color fluorescent DNA microarrays Proc Natl Acad Sci U S A 2002 99 9697 9702 12119387 10.1073/pnas.112318199
Allen TE Herrgard MJ Liu M Qiu Y Glasner JD Blattner FR Palsson BO Genome-scale analysis of the uses of the Escherichia coli genome: model-driven analysis of heterogeneous data sets J Bacteriol 2003 185 6392 6399 14563874 10.1128/JB.185.21.6392-6399.2003
Mori H Horiuchi T Isono K Wada C Kanaya S Kitagawa M Ara T Ohshima H [Post sequence genome analysis of Escherichia coli] Tanpakushitsu Kakusan Koso 2001 46 1977 1985 11593750
Asai K Yamaguchi H Kang CM Yoshida K Fujita Y Sadaie Y DNA microarray analysis of Bacillus subtilis sigma factors of extracytoplasmic function family FEMS Microbiol Lett 2003 220 155 160 12644242 10.1016/S0378-1097(03)00093-4
Kobayashi K Ogura M Yamaguchi H Yoshida K Ogasawara N Tanaka T Fujita Y Comprehensive DNA microarray analysis of Bacillus subtilis two-component regulatory systems J Bacteriol 2001 183 7365 7370 11717295 10.1128/JB.183.24.7365-7370.2001
Serizawa M Yamamoto H Yamaguchi H Fujita Y Kobayashi K Ogasawara N Sekiguchi J Systematic analysis of SigD-regulated genes in Bacillus subtilis by DNA microarray and Northern blotting analyses Gene 2004 329 125 136 15033535 10.1016/j.gene.2003.12.024
Hihara Y Sonoike K Kanehisa M Ikeuchi M DNA microarray analysis of redox-responsive genes in the genome of the cyanobacterium Synechocystis sp. strain PCC 6803 J Bacteriol 2003 185 1719 1725 12591891 10.1128/JB.185.5.1719-1725.2003
Yoshimura H Yanagisawa S Kanehisa M Ohmori M Screening for the target gene of cyanobacterial cAMP receptor protein SYCRP1 Mol Microbiol 2002 43 843 853 12085767 10.1046/j.1365-2958.2002.02790.x
Florens L Washburn MP Raine JD Anthony RM Grainger M Haynes JD Moch JK Muster N Sacci JB Tabb DL Witney AA Wolters D Wu Y Gardner MJ Holder AA Sinden RE Yates JR Carucci DJ A proteomic view of the Plasmodium falciparum life cycle Nature 2002 419 520 526 12368866 10.1038/nature01107
Le Roch KG Zhou Y Blair PL Grainger M Moch JK Haynes JD De La Vega P Holder AA Batalov S Carucci DJ Winzeler EA Discovery of gene function by expression profiling of the malaria parasite life cycle Science 2003 301 1503 1508 12893887 10.1126/science.1087025
|
16029499
|
PMC1199580
|
CC BY
|
2021-01-04 16:27:26
|
no
|
BMC Bioinformatics. 2005 Jul 19; 6:182
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-182
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-1861604276410.1186/1471-2105-6-186Methodology ArticleNonparametric tests for differential gene expression and interaction effects in multi-factorial microarray experiments Gao Xin [email protected] Peter XK [email protected] Department of Mathematics and Statistics, York University, 4700 Keele Street, Toronto, ON M3J 1P3, Canada2 Department of Statistics and Actuarial Science, University of Waterloo, 200 University Ave. W., Waterloo, ON N2L 3G1, Canada2005 21 7 2005 6 186 186 19 4 2005 21 7 2005 Copyright © 2005 Gao and Song; licensee BioMed Central Ltd.2005Gao and Song; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Numerous nonparametric approaches have been proposed in literature to detect differential gene expression in the setting of two user-defined groups. However, there is a lack of nonparametric procedures to analyze microarray data with multiple factors attributing to the gene expression. Furthermore, incorporating interaction effects in the analysis of microarray data has long been of great interest to biological scientists, little of which has been investigated in the nonparametric framework.
Results
In this paper, we propose a set of nonparametric tests to detect treatment effects, clinical covariate effects, and interaction effects for multifactorial microarray data. When the distribution of expression data is skewed or heavy-tailed, the rank tests are substantially more powerful than the competing parametric F tests. On the other hand, in the case of light or medium-tailed distributions, the rank tests appear to be marginally less powerful than the parametric competitors.
Conclusion
The proposed rank tests enable us to detect differential gene expression and establish interaction effects for microarray data with various non-normally distributed expression measurements across genome. In the presence of outliers, they are advantageous alternative approaches to the existing parametric F tests due to the robustness feature.
==== Body
Background
High density oligonucleotide microarray, spotted cDNA array, or other array technologies have presented not only daunting amount of expression data for biologists to explore the inherent biological mechanisms, but also challenging statistical analysis problems. A replicated microarray experiment involves multiple arrays to compare gene expression profile under different conditions. However, normality assumption justifying parametric testing is often untenable in microarray studies [1,2]. For instance, a set of 540 genes from a leukemia data set [3] were analyzed and various distributions for the different genes were found, in which only 13.3% genes have error distributions satisfying the normality assumption [4]. If the underlying distributions of expression measurements can be validated properly, model-based approaches such as likelihood or Bayesian inference can validly accept non-normally distributed data and gain satisfactory power to detect differentially expressed genes (e.g. [5-10]). For example, a hierarchical mixture-model has been proposed with parameterizations for Gamma or log-normally distributed measurements [10]. However when the distribution of the data is difficult to characterize, nonparametric inference makes less stringent distributional assumptions and thereby provide appropriate analysis.
Furthermore data contamination can arise in microarray setting due to different reasons. For instance, an image contamination can occur if a long scratch is present on the array image or a corner of the array is misaligned in the image processing stage. Sample contamination can occur if the mRNA sample is contaminated with other sources of RNA present in the laboratory environment. Such outliers are dramatically different from the majority of the observations and can greatly undermine the sensitivity of parametric approach. A method of assessing goodness of fit to a linear model has been used to automatically detect outliers that possess too large deviation from the overall pattern [11]. Alternatively, a quality index based on coefficient variation was adopted to filter out outlying values with poor quality [9]. Nevertheless the inspection process is time consuming for such large-scale expression data analysis [11]. In this context, nonparametric inference is advantageous as it is insensitive to the presence of outliers. Even without the step of outlier filtering, the validity and power of the nonparametric procedures would be minimally affected.
The development of both parametric and nonparametric methods to address the two condition problem in microarray setting has recently received much attention. Most of the parametric tests employed t or t-like statistics and differ primarily in the estimation of variance [12]. In contrast to these methods which treat the genes as separate fixed effects, the two-group Bayes method was proposed to treat the genes as arising from a certain population. Thus the dimensionality of the inference problem was reduced by sharing information across the array [5,9]. Nonparametric approaches have also been proposed for two-user defined groups [12,13]. The Wilcoxon rank sum test was considered in [14,15] to identify differentially expressed genes in comparison with the Fisher-Pitman permutation test, which is also referred as the nonparametric t test [15]. Recently, the Baumgartner-Weiß-Schindler test has been recommended to detect differentially expressed genes in two groups, which was shown to be less conservative and more powerful than the Wilcoxon rank sum test [16].
However, a microarray experiment often has more complicated design than that of two user-defined groups. Besides the treatment effects of interest, there may exist some clinical covariates such as age, gender and certain clinical symptoms, which also influence the gene expression level. For such experiments, a factorial design model is useful to account for the multiple sources of variation. Townsend and Hartl [6] derived a Bayesian model that has been widely used for the estimation of gene expression levels in multifactorial experiments [7,17]. This model has been extended [8] to accommodate not only additive error terms but also multiplicative error terms to resolve small yet statistically significant differences in gene expression. Alternatively, an overall ANOVA model has also been widely used that simultaneously considers all the genes on the arrays and incorporates array effect and dye effect [18]. A gene specific ANOVA model under the normality assumption was considered in [19]. A mixed linear model was proposed to assess gene significance in which both fixed treatment effects and random array effects were assumed [20]. Unfortunately, there has been no nonparametric procedure proposed up to date to analyze multifactorial microarray data. In addition, the establishment of interaction effect between the multiple attributing factors can help elucidate certain biological mechanisms related to the regulation of gene expression. Thus it is desirable to develop a set of nonparametric procedures to detect differential gene expression and establish interaction effects for multifactorial microarray data.
Results
Principle of the method
To account for the multiple sources of variation attributing to the gene expression, we consider the following model for each specific gene:
Xkijn = θk + Tij + Cj + εkijn, k = 1,..., K; i = 1,..., I; j = 1,..., J; n = 1,..., N (1)
with ∑j Cj = 0, and ∑i, j Tij = 0, where k indexes for the gene number, i indexes for the treatment group, j indexes for the covariate group, n indexes for the replicate number. In the equation, Xkijn represents the expression measurement, θk represents the kth gene specific mean, Tij represents the effect of the ith treatment group (for instance, drug treatments, tissue types, and strains of mice) through its main effect and interaction effect with the jth level of the clinical covariate, and Cj represents the effect of the jth level of the clinical covariate. The error terms εkijn are independently and identically distributed random noise from a continuous distribution function Fk.
To further discern the interaction effect, the treatment effect Tij, can be decomposed into Tij = Mi+ γij with ∑i Mi = 0 and ∑i, j γij = 0, where Mi denotes the main effect of the treatment group and γij denotes the interaction effect. Interaction effects are often of biological interest when the treatment effects are heterogeneous across the levels of the clinical covariate. For example, consider a data set with mouse strains as treatment groups and tissue types as covariate groups, the interaction effects arise when the effects of different mouse strains are disproportional over different tissue types. It is worth noting that model (1) is related to ANOVA models proposed by other researchers [18-20]. The difference between our factorial model (1) and the existing ANOVA models is two-fold: model (1) accommodates multifactor effects on each specific gene, and it does not make normality assumption on the error terms εijkn.
To develop nonparametric rank tests for multifactorial microarray experiments, it is natural to consider rank procedure which can be viewed as nonparametric analogue of the parametric analysis of variance approach. The popular rank transform (RT) method consists of replacing the observations by their ranks in the combined sample and performing one of the standard analysis of variance tests on these ranks [21]. However in the general multifactorial model, the RT method is not valid for most of the common hypotheses due to the nonlinear nature of the rank transformation. For example in the presence of interaction effect, the naive application of ranks into ANOVA formula cannot be used to detect for main effect nor for the interaction effect. Theoretical validations of these limitations of RT method have been thoroughly discussed by Brunner and Neumann, Akritas, and Wilcox, among many others [22-28]. Since the RT method can be easily accomplished by using standard computer packages, extra caution needs to be exerted to prevent the inappropriate extensions of RT method for microarray data analysis under the multifactorial model. In the following, we shall present rank procedures which are similar to RT methods in the sense that they also resemble the analysis of variance approach, however they incorporate more rigorous treatment on the data rather than just replacing the actual observations by the overall rankings.
Usually as the first step of the analysis, we wish to assess whether the genes are differentially expressed among the treatment groups. The testing of treatment effect under model (1) is equivalent to the testing of the hypotheses: H10 : Tij = 0, for all i, j versus H1a : Tij ≠ 0 for some i, j. To address this testing problem, we proposed to use the modified rank transform method (MRT) which consists of first standardizing the rank scores and then plugging them into the analysis of variance formula [25]. The resulting MRT statistic has proven to asymptotically follow a χ2 distribution with (I – 1)J degrees of freedom. In a replicated microarray analysis, the sample size N is often so small that the large-sample asymptotic chi-squared distribution is not accurate enough to obtain valid p-values. To assess the significance of the rank statistic, the permutation method will be invoked to provide p-values of the observed statistic. An alternative way to reduce the computational burden encountered by the permutation procedure is to assess the significance of the proposed rank tests by the limiting chi-squared distribution. Table 1 provides the type I error rates of MRT based on the chi-squared approximation as the sample size increases from 5 to 15 and 20. It is demonstrated that a cell sample size of 20 or more are required for the chi-squared approximation to maintain type I error rate close to the correct nominal level.
Table 1 Convergence of type I error rates of the MRT test based on chi-squared approximation. The type I error rates of the MRT test based on chi-squared approximation were evaluated under varying sample sizes. A 2 × 2 design and a 3 × 4 design were considered.
Dist Design N = 5 N = 15 N = 20
N 2 × 2 0.077 0.047 0.052
3 × 4 0.075 0.055 0.053
U 2 × 2 0.075 0.046 0.051
3 × 4 0.083 0.055 0.053
LN 2 × 2 0.074 0.060 0.055
3 × 4 0.078 0.061 0.056
CN 2 × 2 0.081 0.053 0.051
3 × 4 0.072 0.054 0.053
C 2 × 2 0.072 0.046 0.052
3 × 4 0.068 0.061 0.055
In practice, ties are commonly encountered in microarray data due to rounding and data modification [16]. In the presence of ties, we adopted the method of mid-ranks which assigns each tied individual the average of the tied ranks. There are other methods of dealing with ties such as the methods of randomization and the average statistics. However it has been shown that the randomization method is less powerful under the alternatives due to the supplementary random effects introduced by the randomization. In addition, the method of average statistics typically leads to a conservative test that has a lower significance level than the nominal one [29]. Thus the method of mid-ranks is most frequently used compared to other method to handle ties. As little is known about the small-sample performance of MRT using mid-ranks, it is of interest to conduct simulation studies to investigate this aspect. The related result is provided in the subsequent section.
An important aspect related to multifactorial design is to address treatment-covariate interaction effects. When the interaction is present, the gene expression level will be affected by the treatment disproportionally over different covariate levels. Based on the additive decomposition model for the treatment effects Tij = Mi + γij, the testing of interaction effect is equivalent to the testing of the hypotheses H20 : γij = 0, for all i, j versus H2a : γij≠ 0 for some i, j. As we have emphasized above, the RT method does not yield valid statistics for interaction effects (see [22-25]). Instead we employed the aligned rank transform test (ART) to test for the above hypotheses [30]. ART test consists of performing the analysis of variance test on the ranked residuals of the aligned observations. Although both utilize the ANOVA formula, the ART method differs from the RT method as it is based on residuals after the alignment. In contrast to RT, ART is a valid test for interaction regardless of the presence of main effects [31].
If there are no interaction effects, we can consider a simpler model: Xkijn = θk + Mi + Cj + εkijn, with Mi denoting the treatment main effect and Cj denoting the covariate effect. Testing for the treatment main effect corresponds to the hypotheses: H30 : Mi = 0 for all i against H3a : Mi≠ 0 for some i. We propose to employ the rank transform statistic suggested in [21]. It is worthy to point out that the testing of main effects in the absence of interaction is one of very few situations that naive application of the ANOVA formula on rank scores can yield valid statistic with satisfactory power properties.
In data analysis, the three testing procedures discussed above are connected. The following empirical rule regards how to proceed to choose the tests in a real data analysis. As the first step of analysis, the hypothesis of treatment effects (H10) is usually tested to see if the gene is differentially expressed across treatment groups. If H10 is accepted, no more actions will be taken as no differential expression is detected. If H10 is rejected, we may further perform the test for interaction effects (H20) to see if the differential expression is partly due to the interactions between treatment groups and covariate groups. The acceptance of H20 implies there exist no interaction effects. Then the testing for main effects (H30) can be pursued on the basis that the interaction effects are found insignificant.
Single gene analysis
Simulation studies were conducted to evaluate the performances of the proposed rank methods in comparison with the other two competing methods, the parametric F test (FT) and the permutation F test (PFT) that uses the F statistic but computes p-values through permutations. The criterion used in the comparison is the efficiency gain relative to the FT method, defined as
where T can be either the MRT, ART, RT, or PFT. Obviously, when the test T outperforms the FT, the EG will be positive; otherwise, the EG is negative.
The performances of these methods were evaluated under different noise distributions and different numbers of replications. We considered a replicated factorial array experiment involving two treatment groups, two levels of a clinical covariate and varying cell sample sizes. Average type I error rates and power were calculated from 1,000 simulation runs. From the literature it has been shown that normal, uniform, log-normal, Cauchy and normal mixture distributions, among others, are commonly seen for microarray expression data [4]. In our simulation, we considered normal N(0, 1), uniform U(-2, 2), log-normal LN(0,1), Cauchy C(0.5) and contaminated normal CN(0.75, 0.5, 2) = 0.75N(0, 0.5) + 0.25N(0, 2). To some extent, contaminated normal can be used to model data with sample contamination, with one normal component representing the true underlying mRNA population of interest and the other normal behaving as the mRNA population from the contamination source. It is recognized that this normal mixture model may not be able to describe more irregular and dramatic data contamination such as distorted array image or scratched array regions. Fortunately the proposed nonparametric method does not rely on the correct characterization of the underlying distribution. This set of distributions were selected mainly for comparison purpose and they represent a broad range of characteristics from light-tailed to heavy-tailed, and from symmetric to asymmetric distributions.
We first evaluated the performance of the proposed MRT statistic for the testing of the treatment effects. We set the clinical effects as C1 = -0.5, and C2 = 0.5. Under the alternative situation, we set the treatment effects as T11 = 0.7, and T12 = 0.7, T21 = -0.9, T22 = -0.5, which were induced by the main effects M1 = 0.6, M2 = -0.6, and the interaction effects γ11 = 0.1, γ12 = 0.1, γ21 = -0.3, γ22 = 0.1.
Table 2 provides the results of the type I error and power of the MRT as well as its two competitors FT and PFT. The type I error rates of the FT appear around 0.05 in the case of light or medium-tailed distribution (normal and uniform). However for heavy-tailed distributions, especially in the case of Cauchy distribution, even with the sample size N = 10, the FT seems to be very conservative. Thus, the performance of the FT under the null can become rather poor if the error distribution is very different from the normal. In contrast, the type I error rates of MRT and PFT are advantageous as they are close to the correct nominal levels regardless of the underlying distribution.
Table 2 Type I error rates and power for treatments effects. The type I error rates and power were evaluated under five different error distributions – normal, uniform, lognormal, contaminated normal and Cauchy. The values inside and outside parenthesis are type I error rates and power, respectively. The EG(PFT) and EG(MRT) denote the efficiency gain of PFT and MRT versus FT.
Dist N FT PFT MRT EG (PFT) EG (MRT)
N 5 0.732 (0.050) 0.732 (0.050) 0.691 (0.054) 0.000 -0.056
10 0.976 (0.050) 0.976 (0.050) 0.966 (0.050) 0.000 -0.010
U 5 0.575 (0.051) 0.565 (0.047) 0.486 (0.048) -0.017 -0.155
10 0.930 (0.054) 0.930 (0.054) 0.876 (0.050) 0.000 -0.058
LN 5 0.386 (0.029) 0.461 (0.047) 0.587 (0.048) 0.194 0.521
10 0.576 (0.036) 0.620 (0.052) 0.889 (0.052) 0.076 0.543
CN 5 0.565 (0.039) 0.598 (0.058) 0.736 (0.055) 0.058 0.303
10 0.829 (0.047) 0.838 (0.056) 0.958 (0.052) 0.010 0.156
C 5 0.230 (0.021) 0.366 (0.054) 0.564 (0.049) 0.591 1.452
10 0.259 (0.016) 0.402 (0.052) 0.848 (0.055) 0.552 2.274
With regard to the power, the results are distribution dependent. For the medium or light-tailed distributions (normal and uniform), FT and PFT have similar performances and both of them achieve higher power than MRT test. In contrast, for the other distributions with heavy-tails, skewedness and contamination, MRT appears superior to the two competing methods. When the sample size N is 10, the MRT's efficiency gain, EG(MRT), is 54.3%, 15.6%, and 227.4% under log-normal, contaminated normal and Cauchy respectively. On the other hand, with the same sample size, the efficiency loss of the MRT is approximately 1.0% and 5.8% under normal and uniform. Compared to the amount of efficiency gain for the MRT versus the FT, the amount of efficiency loss seems to be marginal. Similar conclusions can be drawn when the sample size is 5. That is, the MRT's efficiency loss is approximately 5.6% for normal and 15.5% for uniform; the MRT's efficiency gain is 52.1%, 30.3%, and 145.2% for log-normal, contaminated normal and Cauchy respectively.
One interesting variant of the MRT method is the involvement of mid-ranks to handle ties. We randomly introduced m ties in the simulated data set. Table 3 lists the results of type I error and power of the three tests in the presence of m = 2, 5 or 10 pairs of ties. Comparing to Table 2, it is clear that these ties incurred only marginal differences in both type I error and power, even for the extreme scenario of m = 10. The slight increase of power in the presence of ties could be due to the decrease of within-group variation caused by averaging the ranks for tied observations.
Table 3 Type I error rates and power in the presence of ties. The type I error rates and power were evaluated under different error distributions and varying number of ties. The values inside and outside parenthesis are type I error rates and power, respectively. The cell sample size N = 5.
Dist # of ties FT PFT MRT
N 2 0.729 (0.051) 0.726 (0.051) 0.684 (0.052)
5 0.730 (0.056) 0.730 (0.054) 0.682 (0.052)
10 0.724 (0.053) 0.715 (0.051) 0.687 (0.051)
U 2 0.589 (0.054) 0.577 (0.052) 0.495 (0.052)
5 0.580 (0.050) 0.568 (0.046) 0.493 (0.051)
10 0.554 (0.048) 0.539 (0.046) 0.482 (0.048)
LN 2 0.391 (0.032) 0.455 (0.041) 0.581 (0.048)
5 0.411 (0.039) 0.465 (0.054) 0.585 (0.056)
10 0.412 (0.031) 0.448 (0.043) 0.589 (0.045)
CN 2 0.563 (0.038) 0.594 (0.051) 0.730 (0.053)
5 0.575 (0.040) 0.599 (0.048) 0.735 (0.056)
10 0.597 (0.044) 0.613 (0.054) 0.741 (0.057)
C 2 0.244 (0.023) 0.397 (0.057) 0.533 (0.052)
5 0.262 (0.022) 0.370 (0.050) 0.562 (0.049)
10 0.292 (0.030) 0.368 (0.048) 0.558 (0.055)
Next we examined the performance of the ART in testing for interaction effects, as well as the comparison to the FT and PFT. The simulation was set up as follows: under the null situation, the main effects were assigned, respectively, as R1 = -0.8 and R2 = 0.8, and C1 = -0.5 and C2 = 0.5; under the alternative situation, the main effects remained the same and additionally the interaction effects were given by γ11 = 0.6, γ12 = -0.6, γ21 = -0.6, and γ22 = 0.6. Table 4 provides the results of type I error and power of the three testing methods for the significance of the interaction effects. All these tests echo similar performances as presented in the above simulation study for the treatment effects. Notably, when the distribution of noise is heavy-tailed, skewed or contaminated, the ART appears considerably more powerful than the FT and the PFT. Regarding the testing of main effects, the RT statistic exhibits rather similar performance as the above two rank tests. The details of the RT are provided in the authors's website [32].
Table 4 Type I error rates and power of different tests for interaction effects. The type I error rates and power of the three different tests for interaction effects were evaluated under different error distributions. The values inside and outside parenthesis are type I error rates and power, respectively.
Dist N FT PFT ART EG (PFT) EG (ART)
N 5 0.711 (0.055) 0.697 (0.053) 0.698 (0.048) -0.020 -0.018
10 0.952 (0.051) 0.951 (0.051) 0.947 (0.051) -0.001 -0.005
U 5 0.565 (0.049) 0.548 (0.045) 0.520 (0.050) -0.030 -0.080
10 0.891 (0.047) 0.891 (0.048) 0.833 (0.047) 0.000 -0.065
LN 5 0.405 (0.031) 0.430 (0.040) 0.579 (0.048) 0.062 0.430
10 0.576 (0.038) 0.593 (0.045) 0.893 (0.050) 0.030 0.550
CN 5 0.525 (0.035) 0.540 (0.044) 0.718 (0.045) 0.029 0.368
10 0.812 (0.043) 0.817 (0.050) 0.969 (0.051) 0.006 0.193
C 5 0.239 (0.014) 0.336 (0.029) 0.523 (0.044) 0.406 1.188
10 0.274 (0.022) 0.364 (0.056) 0.859 (0.052) 0.328 2.135
Global array analysis
The above discussion focuses on the single gene analysis. However, in microarray analysis the subsequent analysis step typically involves either adjusting the significance for multiple testing [33,34], or ranking genes according to the significance level such that the most relevant top k genes could be selected. Although discussing these global analysis approaches is beyond the scope of this paper, we are fully aware that the capability of a testing procedure to generate extreme p-values has a direct influence on the selection of the most relevant genes. When the Bonferroni procedure is employed to deal with the multiplicity, the Wilcoxon rank sum test is more conservative and less powerful than the Fisher-Pitman test or the parametric t-test [15]. It was further demonstrated that the discreteness of the exact permutation distribution of the Wilcoxon test is responsible for the conservatism [16]. Because of this, the Baumgartner-Weiß-Schindler test is recommended, as its exact permutation distribution has more non-zero mass probabilities and capable of generating richer small p-values than the Wilcoxon test. It is worthy pointing out that as the Bonferroni procedure is almost always more conservative than other multiple testing procedures, it will suffer most from the discreteness problem of the permutation distribution. Other multiple testing procedures impose less stringent p-value thresholds, therefore they are affected by the discreteness problem to a lesser extent.
Our rank methods face the same issue as they use the permutation distribution to obtain p-values. It is crucial to examine how the discreteness of permutation distribution affects the performance of the MRT. We plotted the p-values (in log scale) of the MRT versus those of the FT in the connection to the first simulation study of testing for the treatment effects with 1000 runs. Figures 1a – 1c, corresponding to N = 2, 5, 10, depict the agreement between the MRT and FT tests under the log-normal noise, in which the perfect agreement is indicated by the solid 45° division line. We comment that (i) the symmetry around the 45° division line decreases as the number of replicates N increases, this implies that the MRT becomes more capable of producing extreme p-values than the FT test. Thus when N is 5 or larger, the permutation approach works reasonably well for the MRT method; and (ii) when N is small, say 2, the p-value of the MRT is often bounded due to the limited number of distinct probability mass points. For the example of the 2 × 2 design, as the permutation is carried out within each covariate group, the number of different permutation configurations equals . With N = 2, the number of possible different permutations is limited to only 36, so there are at most 36 different probability mass points. When N increases to 5, the resulting number of permutations increases to 6.35 × 104, which considerably alleviates the problem of discreteness and improve the performance of the MRT. When N increases to 10, the corresponding number of permutations increases to 3.41 × 1010, and consequently further lessens the discreteness problem.
Figure 1 Comparison of the MRT test vs the parametric FT test. The P-values of the MRT test (X-axis) vs the FT test (Y-axis) under lognormal distribution were plotted under the logarithm scale. The replicate number ranges from 2, 5 to 10.
In order to fully understand the effect of the discreteness of the permutation distribution, it is of interest to compare the p-values from the permutation method to the p-values obtained from the true distribution. Under the null situation, it is known that the p-values of MRT obtained from the true distribution should follow a uniform distribution on (0, 1), and the corresponding cumulative distribution function (CDF) should be the straight line y = x, x ∈ (0, 1). Figure 2a to 2d provide the comparisons of the empirical CDFs of the permutation p-values versus the CDF of the true p-values for MRT test under the null situation when N = 2, N = 5, and N = 10. It is observed that with N = 2, the CDF of permutation p-values appears as a step function due to the discreteness at the limited number of probability mass points. The overall curve does not match very well with the CDF of the true p-values. When the sample size increases to 5 and 10, the agreement between the CDF of permutation p-values and the CDF of true p-values greatly improves and majority of the two curves overlap with each other. Therefore the plots suggest that the discreteness problem of permutation p-values is almost diminished with sample sizes greater than 5. In addition, the CDFs of the p-values from the chi-squared approximation are plotted in these figures. It is shown that the discrepancy between the CDFs of the chi-squared p-values and true p-values is generally larger than that between the permutation p-values and the true p-values. For instance, the Figure 2d is a zoomed image of the Figure 2c into the p-value range of (0, 0.25), it is shown that the CDF of the chi-squared p-values falls high above the y = x line indicating large inflation of type I error rate, while the CDF of the permutation p-values matches well with the y = x line. In conclusion, the permutation method provides a better control of type I error rate and therefore is more preferable compared to the chi-squared approximation in small sample size scenario.
Figure 2 The empirical CDFs of the P-values of the MRT test.The empirical cumulative distributions of the P-values of the MRT test obtained from the permutation method or the chi-squared approximation are compared with that of the true p-value under the null situation. The y = x line denotes the CDF of true p-values; solid curve denotes the CDF of permutation p-values; dashed curve denotes the CDF of chi-squared p-values. Figure 2d is the zoomed image of the Figure 2c into the p-value range of (0, 0, 25).
Biological data analysis
We now illustrate the proposed MRT and ART methods as well as their competitors, the FT and the PFT methods, to analyze the gene expression data collected from six brain tissue regions in two mouse strains [35]. The data is obtained from [36], which contains a subset of 1000 genes. The purpose of the study was to investigate the genetic components contributing to the neurobehavioral differences between two mouse strains. For each mouse strain, the samples were obtained from 6 tissue regions, which can be viewed as a clinical covariate with 6 levels. For each mouse strain and a specific tissue, the expression profiles of two biological replicates were assessed. The p-values of the MRT versus respective p-values of the FT and the PFT were plotted and the MRT appears to be less capable of producing extreme p-values than the FT due to the low replicate numbers. In fact, this discreteness phenomenon has been unveiled in the simulation study through Figure 1.1. It was further shown that for the majority of genes the MRT and the PFT agree with each other. Among the top 100 genes selected by the MRT, 61 genes were selected by the FT and 77 genes were selected by the PFT. We then selected 57 genes that were identified as differentially expressed in two mouse strains by all the three methods in their top 100 rankings. To verify if these selected biomarkers really play any biological roles in the neurological phenotypic differences in mouse strains, we explored the gene functions by NetAffx Analysis Center in Affymetrix website [37]. The complete list of the functions of these 57 genes are available from the authors' website. Among these 57 genes, 24 genes share similar functions related to protein binding, transfer activity, signal pathway, receptor activities and mitochondrial electron transport chain, which are known to be essential to the function of nervous system. Another 14 genes share similar functions related to muscle movement, catalytic activity, kinase activity, hydrolase activity, and two other genes are related to hormone regulations, which are all related to the proper function and the regulation of nervous system. In total, 40 genes out of our list of 57 genes exhibit biological functions attributing to the phenotypic difference in the two mouse strains. Figure 3 displays that the selected 57 genes yield a clear separation of the samples from the two mouse strains. Therefore, the common list of genes identified by these three methods provides a reliable list of biomarkers.
Figure 3 The common list of genes identified by all the three methods.The figure provides the common list of 57 genes identified by all the three methods in their top 100 rankings as differentially expressed in two mouse strains of the data of Sandberg et al. (2000).
If more exploratory research can be afforded to look for other genes, it is suggested to investigate the genes identified exclusively by the MRT (not by either the FT or the PFT). This extra list might provide a potential list of candidate genes that did not pass the two F-tests due to non-normal distributed noise in the data. To scrutinize this list, we also investigated the functions of 19 genes remaining in the list. The information regarding these 19 genes' functions is also available from the website as above. Among these 19 genes, 11 genes share similar functions as protein binding, transfer activity, signal pathway, receptor activities, mitochondrial electron transport chain, catalytic activities and kinase activities. It remains inconclusive if the other 8 genes can be supported as true positives due to the lack of known biological evidence.
As selecting the top listed genes only provides the set of most favorable candidates, no probabilistic statement can be attached to the findings. Alternatively, we can assess the significance of the findings under the multiple testing framework. Instead of using the stringent Bonferroni procedure, we applied the Benjamini and Hochberg's linear step-up procedure to control false discovery rate (FDR) [38]. As this procedure selected genes based on the ordered p-values, the significant genes were chosen consecutively down the top gene lists. By Controlling the FDR at level 0.05, the parametric F-test found 13 significant genes, 8 of which were found by either the permutation F-test or the nonparametric rank test. Given that there are only two replicates in the data set, it is not surprising that permutation-based methods identified a smaller number of significant genes, due to the discreteness of the permutation distribution discussed above.
The interaction effect can arise when the effect of changing mouse strain is disproportional over different brain regions. The ART, FT and PFT were applied to test for the interaction effects between the mouse strains and the tissue regions. Comparison of the p-values from the ART versus the respective p-values from the FT and the PFT demonstrate a good deal of agreement among the three methods. Since the permutation was carried out on the basis of 24 aligned observations, the number of distinct permutations is so large that the discreteness problem is alleviated. Among the top 100 genes selected by the ART, 80 and 80 genes appeared in the top 100 rankings by the FT and the PFT, respectively. To visualize the interaction effects, for each gene the two profile curves for the two mouse strains were plotted representing the average expression levels over the six brain regions. Figure 4 provides examples of the profile curves of genes which are identified as having interaction effect by all the three methods versus genes which are found to have no interaction effects. For genes with no interaction effect, the two curves have parallel trends and differ by a vertical shift corresponding to the strain effect. In contrast, for genes with interaction effect, the two curves exhibit rather different patterns and even intersect with each other. For instance, the level of probe AA209596 was higher by two-fold in the cerebellum of strain 129SVEv compared with the C57BL/6 cerebellum. By contrast, in the entorhinalcortex region the level of probe AA209596 was lower by a factor of 1.2-fold in 129SVEv. Thus the differential expression between the two mouse strains reverses direction in two different brain regions. Probe AA209596 corresponds to gene TIMM13 which is translocase of inner mitochondrial membrane and has prominent expression in the large neurons in the brain. The TIM family plays important role in neurological behaviors as mutation of TIM gene is linked to neurobehavioral disorders such as deafness. Our finding suggests that the strain effects and brain region effects interact to regulate the expression of TIMM13. This analysis exemplifies how certain interacting mechanism behind gene expression can be unveiled via the interaction test on multifactorial microarray data.
Figure 4 Comparison of genes with and without interaction effects.For each specific gene, two expression profiles are plotted for each of the two mouse strains across six brain regions-amygdala, cerebellum, cortex, entorhinalcortex, hippocampus and midbrain, which are denoted by 1 to 6 on x-axis. Figure 4a provides the expression profiles of four genes without interaction effects. Figure 4b provides the expression profiles of fours genes with interaction effects.
Discussion
Because there is a loss of information whenever the original data is collapsed to ranked data, the abandonment of parametric methods may not be cost-effective in all settings. In this article we have thoroughly investigated the positives and negatives of the proposed nonparametric rank tests versus the parametric ANOVA tests: (1) Due to the information loss, the rank tests are marginally less powerful than the ANOVA tests for normal, uniform or other light-tailed distributions. On the other hand, our simulation illustrated that the rank tests are substantially more powerful than the ANOVA tests if the data follow heavy-tailed, skewed or asymmetric distributions. (2) Our investigation also demonstrated that reasonable number of replicates (N ≥ 5 for 2 × 2 design) are required to lessen the discreteness of permutation distribution encountered by the rank tests to evaluate p-values. In contrast, when the normality assumption is validated, the p-value of the parametric ANOVA statistic can be evaluated from the exact F distribution. (3) In the presence of severe outliers, the robust rank tests is more favorable than the parameter ANOVA tests. (4) When it is difficult to characterize the distribution of the data, the proposed distribution-free rank tests are useful to conduct an appropriate and powerful analysis.
As the comparative properties of rank tests relative to ANOVA tests are distribution dependent, distribution diagnostics can help the practitioners to determine which test will yield better power for a specific data set. Graphic inspections such as box-plot and normal probability plot offer a convenient way to visualize the shape of the underlying distribution. To quantify the magnitude of the deviation from normality, the Shapiro-Wilk test can be performed [39]. Let x[1],...,x[N] be the ordered values of N independent and identically distributed observations. Let z[1],...,z[N] denote the vector of the associated quantiles of the standard normal distribution. The Shapiro-Wilk statistic is defined as the squared correlation between the ordered data values (sample quantiles) and the normal quantiles:
For data that are really generated from normal distribution, the W statistic would be close to one. A smaller value of W indicates more deviation from normality. To further discern the deviation due to heavy-tail from the deviation due to light-tail, another statistic W* similar to the above Shapiro-Wilk statistic can be formed. The W* is defined as the correlation between the sample quantiles and the quantiles from a uniform distribution. As a result, the relative sizes of W and W* indicate the tail property for a given distribution. For instance, data generated from a heavy-tailed distribution would yield W > W*. This is because the correlation between a heavy-tailed distribution with the medium-tailed normal distribution should be stronger than that with the light-tailed uniform distribution. A reasonable threshold value τ for the statistic W will be determined by the comparative property of the nonparametric test relative to the two parametric F tests. A simulation-based approach can be invoked to numerically calculate this cutoff value. We illustrate such a procedure in a design model with R = 2, C = 2 and N = 5. The noise εijn were simulated from a normal distribution. Let ε[1],..., ε[20] be the ordered noise. We gradually introduced heavy-tailedness into the data set by pulling the left and right end points of the ordered list of noise further away from the center. Each time a new data set was generated with = ε[i], for i = 4,..., 17, and = ε[i] * d, for i = 1, 2, 3, 18,19, 20, and d was chosen from the varying range of 1.1, 1.3,..., 3. The corresponding W and the p-values of the FT, PFT and MRT tests were recorded for the data set. The result was summarized based on 1000 replications. In Figure 5, the empirical power curves of the three competing methods were plotted against the varying level of W. Our simulation demonstrates that as heavy-tailedness is introduced into the data set, W level decreases correspondingly. When W value is above 0.92, the two parametric methods outperform the nonparametric method. When W value is below 0.92, the nonparametric method is superior to the two parametric competitors. Thus for the specific design setting that we simulated, we choose a threshold value of τ = 0.92. If W <τ and W > W*, we would recommend the use of the nonparametric method. Among many sources of the normality violation discussed above, if τ = 0.92 was used as the cutoff, we found about 10% of genes in the data set of Sandberg et al. [35] whose expression measurements are from heavy-tailed distributions.
Figure 5 Empirical power curves of three competing methods with respect to Shapiro-Wilk statistic. The empirical power curves of the three competing methods – MRT, FT and PFT, are plotted against the varying level of Shapiro-Wilk statistic. The Shapiro-Wilk statistic is employed to assess the magnitude of the heavy-tailedness in the distribution.
With regard to future extensions of the proposed methods, the tests discussed above can be applied to a high-way layout by collapsing these covariates into one. For example, a covariate with J levels and another covariate with K levels can be combined as a single factor of JK levels, so that the treatment effects can still be tested using the above two-way layout. When the data contains continuous covariates in certain applications, one can simply apply the proposed rank test method on the basis of residuals, the differences between the observations and the least squares fitted values calculated by using all the continuous covariates. Furthermore, it is possible to extend our methods to accommodate the dependence or heteroscedasticity which might occur in the microarray data sets. If the variances vary across different covariate groups, j = 1,..., J, the MRT statistic can still be employed to test for treatment effects using the standardized overall rank Zijn [25]. To deal with two-way models with repeated measures on one factor or on both factors, the rank statistic can be extended to a quadratic form incorporating an estimated covariance matrix reflecting the dependence structure in the data [40].
In this article, we have focused on the interaction effects between multiple attributing factors to the gene expression. Currently there has been an increasing interest in studying interactions between genes as opposed to clinical factors. To address this problem, we could select a number of genes and treat their expressions measurements as explanatory variables. The biological phenotype of interest can be chosen as the response variable. Then a linear model can be fitted linking the gene expressions and biological phenotype. The aforementioned interaction tests can be applied to this setting to investigate the possible interactions among the genes.
Conclusion
We have presented a set of nonparametric tests to detect treatment effects, clinical covariate effects, and interaction effects for multifactorial microarray data. These methods can be extended to accommodate high-way layouts, continuous covariates, dependent observations and heteroscedasticity which might occur in the microarray data sets. The proposed nonparametric procedures will prove to be of wide use in microarray data analysis as they can accommodate various noise distributions across genome.
Methods
Rank test for treatment effects
The first hypothesis H01 is formulated to test for treatment effects in two-way layout. Correspondingly, we have proposed a modified rank transform (MRT) test. This test standardizes the rank scores before plugging them into the analysis of variance formula. For simplicity in notation, we suppress the index k, as all the observations in the model are from a specific gene k. Let Rijn denote the rank of Xijn among all of the observations and define . Let denote the sample variance of ranks within the jth column. Define the standardized rank score Zijn = Rijn/sj. Denote the marginal and overall averages of the standardized rank scores by and . The proposed modified rank transform statistic takes the following form:
It has been shown that the standardization procedure is essential for the validity of the MRT method as the nonlinear rank transformation introduces the heteroscedasticity into the ranked data [25]. To assess the significance of the rank statistic, the permutation method will be invoked to provide p-values of the observed statistic. In implementation, we randomly relabel I treatment groups within each of J covariate levels. Namely, the set of observations X1j1,..., X1jN,..., XIj1,..., XIjN are shuffled within column j for 1 ≤ j ≤ J. For illustration purpose, consider a microarray data set with the covariate consisting of six different tissue regions and the treatment consisting of two distinct mouse strains. The six covariate levels correspond to the six tissue regions. To generate a permuted data set, for the 2N measurements obtained from the same tissue region, we randomly assign N of them to the first mouse strain and assign the remaining observations to the second mouse strain. Repeat this procedure until we have permuted for all the tissue regions to generate a new permuted data set. Then we calculate the proportion of the resulting statistic (3) being equal to or larger than the observed statistic over 10,000 permutations to obtain the permutation p-value.
Rank test for interaction effects
The second hypothesis H02 is formulated to test for interaction effects in two-way layout. To address this testing problem, the ART test is proposed to perform the analysis of variance test on the ranked residuals, of the aligned observations .
Here and are the Hodges-Lehmann estimates of the two main effects given by
Again, with low replicates, we propose to use the permutation method to compute p-values under the null H20. In implementation, we randomly relabel both indices i and j within all the aligned observations and obtain the empirical p-value over 10,000 permutations.
Rank test for main effects
The third hypothesis H03 is formulated to test for main effects in the absence of interaction effects in a two-way layout. We propose to employ the rank transform statistic suggested in [21] which is formulated as follows:
The resulting RT statistic asymptotically follows a χ2 distribution with I – 1 degrees of freedom. Likewise, we can test if there is a difference of gene expression among the clinical covariate levels for gene k, using a test statistic similar to (4), with only indexes I and J being swapped.
Authors' contributions
XG and PS developed the methods and wrote the manuscript.
Acknowledgements
This work is supported by Natural Sciences and Engineering Research Council of Canada grants. We thank Hong Xu and Rui Liu for their assistance to the implementation of the algorithms. We are grateful to the Editor and two referees for their helpful comments that improved the manuscript.
==== Refs
Hunter L Taylor RC Leach SM Simon R GEST: a gene expression search tool based on a novel Bayesian similarity metric Bioinformatics 2001 17 S115 S122 11473000
Zhao Y Pan W Modified nonparametric approaches to detecting differentially expressed genes in replicated microarray experiments Bioinformatics 2003 19 1046 1054 12801864 10.1093/bioinformatics/btf879
Golub TR Slonim DK Tamayo P Huard C Gaasenbeek M Mesirov CollerHJP Loh ML Downing JR Caligiuri MA Molecular classification of cancer: class discovery and class prediction by gene expression monitoring Science 1999 286 531 537 10521349 10.1126/science.286.5439.531
Carey, 2004 HowTo Use the Bioconductor edd package
Newton MA Kendziorski CM Richmond CS Blattne rFR Tsui KW On differential variability of expression ratios: improving statistical inference about gene expression changes from microarray data Journal of computational biology 2001 8 37 52 11339905 10.1089/106652701300099074
Townsend JP Hartl DL Bayesian analysis of gene expression levels: statistical quantification of relative mRNA level across multiple strains or treatments Genome Biology 2002 3 1 71 10.1186/gb-2002-3-12-research0071
Townsend JP Multifactorial experimental design and the transitivity of ratios with spotted DNA microarrays BMC Genomics 2003 4 41 14525623 10.1186/1471-2164-4-41
Townsend JP Resolution of large and small differences in gene expression using models for the Bayesian analysis of gene expression levels and spotted DNA microarrys BMC Bioinformatics 2004 5 54 15128431 10.1186/1471-2105-5-54
Tseng GC Oh MK Rohlin L Liao JC Wong WH Issues in cDNA microarray analysis: quality filtering, channel normalization, models of variations and assessment of gene effects Nucleic Acids Research 2001 29 2549 2557 11410663 10.1093/nar/29.12.2549
Kendziorski CM Newton MA Lan L Gould MN On parametric empirical Bayes methods for comparing multiple groups using replicated gene expression profiles Statistics in Medicine 2003 22 3899 3914 14673946 10.1002/sim.1548
Li C Wong WH Model-based analysis of oligonucleotide arrays: Expression index computation and outlier detection Proc Natl Acad Sci USA 2001 98 31 36 11134512 10.1073/pnas.011404098
Dudoit S Yang YH Speed TP Gallow MJ Statistical methods for identifying differentially expressed genes in replicated cDNA microarray experiments Statistica Sinica 2002 12 111 139
Park PJ Pagano M Bonetti M A nonparametric scoring algorithm for identifying informative genes from microarray data Pac Symp Biocomput 2001 52 63 11262969
Wu TD Analysis gene expression data from DNA microarrays to identify candidate genes Journal of Pathology 2001 195 53 65 11568891 10.1002/1096-9896(200109)195:1<53::AID-PATH891>3.0.CO;2-H
Troyanskaya OG Barber ME Brown PO Botstein D Altman RB Nonparametric methods for identifying differentially expressed genes in microarray data Bioinformatics 2002 18 1454 1461 12424116 10.1093/bioinformatics/18.11.1454
Neuhäuser M Senske R The Baumgartner-Weiß-Schindler test for the detection of differentially expressed genes in replicated microarray experiments Bioinformatics 2004 20 3553 3564 15284098
Ranz JM Castillo-Davis CI Meiklejohn CD Hartl DL Sex-dependent gene expression and evolution of the Drosophila transcription Science 2003 300 1742 1745 12805547 10.1126/science.1085881
Kerr MK Martin M Churchill GA Analysis of variance for gene expression microarray data J Comput Biol 2000 7 819 837 11382364 10.1089/10665270050514954
Pavlidis P Noble WS Analysis of strain and regional variation of gene expression in mouse brain Genome Biology 2001 2 0042.1 0042.15 10.1186/gb-2001-2-10-research0042
Wolfinger RD Gibson G Wolfinger ED Bennett L Hamadeh H Bushel P Afshari C Paules RS Assessing Gene Significance from cDNA Microarray Expression Data via Mixed Models Journal of Computational Biology 2001 8 625 637 11747616 10.1089/106652701753307520
Conover WJ Iman RL On some alternative procedures using ranks for the analysis of experimental designs Communications in Statistics 1976 A5 1349 1368
Brunner E Neumann N Rank tests for the 2 × 2 split plot design Metrika 1984 31 233 243 10.1007/BF01915206
Brunner E Neumann N Two-sample rank tests in general models Biometrical Journal 1986 28 395 402
Brunner E Neumann N Rank tests in 2 × 2 designs Statistica Neerlandica 1986 40 251 271
Akritas MG The rank transform method in some two-factor designs Journal of the American Statistical Association 1990 85 73 78
Akritas MG Limitations of the rank transform procedure: A study of repeated-measure designs, Part I Journal of the American Statistical Association 1991 86 457 460
Akritas MG Limitations of the rank transform procedure: A study of repeated-measure designs, Part II Statistics and Probability Letters 1993 17 149 156 10.1016/0167-7152(93)90009-8
Wilcox RR Applying Contemporary Statistical Techniques 2003 Academic press/Elsevier
Hájek J Sidák Z Theory of rank tests 1967 New York: Academic Press
Mansouri H Chang GH A comparative study of some rank tests for interaction Statistica Sinica 1995 19 85 96
Mansouri H Aligned rank transform tests in linear models Journal of Statistical Planning and Inference 1999 79 141 155 10.1016/S0378-3758(98)00229-8
Auxiliary Simulation Results
Reiner A Yekutieli D Benjamini Y Identifying differentially expressed genes using false discovery rate controlling procedures Bioinformatics 2003 19 368 375 12584122 10.1093/bioinformatics/btf877
Storey JD Tibshirani R Statistical significance for genomewide studies Proc Natl Acad Sci USA 2003 100 9440 9445 12883005 10.1073/pnas.1530509100
Sandberg R Yasuda R Pankratz DG Carter TA Del Rio JA Wodicka L Mayford M Lockhart DJ Barlow C Regional and strain-specific gene expression mapping in the adult mouse brain Proc Natl Acad Sci USA 2000 97 11038 11043 11005875 10.1073/pnas.97.20.11038
Pavlidis P Using ANOVA for gene selection from microarray studies of the nervous system Methods 2003 31 282 289 14597312 10.1016/S1046-2023(03)00157-9
Liu G Loraine AE Shigeta R Cline M Cheng J Valmeekam V Sun S Kulp D Siani-Rose MA NetAffx: Affymetrix probesets and annotations Nucleic Acids Res 2003 31 82 86 12519953 10.1093/nar/gkg121
Benjamini Y Hochberg Y Controlling the false discovery rate: a practical and powerful approach to multiple testing Journal of the Royal Statistical Society, Series B 1995 57 289 300
Shapiro SS Wilk MB An analysis of variance test for normality (complete samples Biometrika 1965 52 591 611
Thompson GL A unified approach to rank tests for multivariate and repeated measures designs Journal of the American Statistical Association 1991 86 410 419
|
16042764
|
PMC1199581
|
CC BY
|
2021-01-04 16:27:26
|
no
|
BMC Bioinformatics. 2005 Jul 21; 6:186
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-186
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-1961607899010.1186/1471-2105-6-196DatabaseCOPASAAR – A database for proteomic analysis of single amino acid repeats Depledge Daniel P [email protected] Andrew R [email protected] Schools of Biological and Chemical Sciences and Engineering, Computer Science and Mathematics, Washington Singer Laboratories, University of Exeter, Prince of Wales Road, Exeter, EX4 4PS UK2005 3 8 2005 6 196 196 18 2 2005 3 8 2005 Copyright © 2005 Depledge and Dalby; licensee BioMed Central Ltd.2005Depledge and Dalby; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Single amino acid repeats make up a significant proportion in all of the proteomes that have currently been determined. They have been shown to be functionally and medically significant, and are associated with cancers and neuro-degenerative diseases such as Huntington's Chorea, where a poly-glutamine repeat is responsible for causing the disease. The COPASAAR database is a new tool to facilitate the rapid analysis of single amino acid repeats at a proteome level. The database aims to simplify the comparison of repeat distributions between proteomes in order to provide a better understanding of their function and evolution.
Results
A comparative analysis of all proteomes in the database (currently 244) shows that single amino acid repeats account for about 12–14% of the proteome of any given species. They are more common in eukaryotes (14%) than in either archaea or bacteria (both 13%). Individual analyses of proteomes show that long single amino acid repeats (6+ residues) are much more common in the Eukaryotes and that longer repeats are usually made up of hydrophilic amino acids such as glutamine, glutamic acid, asparagine, aspartic acid and serine.
Conclusion
COPASAAR is a useful tool for comparative proteomics that provides rapid access to amino acid repeat data that can be readily data-mined. The COPASAAR database can be queried at the kingdom, proteome or individual protein level. As the amount of available proteome data increases this will be increasingly important in order to automate proteome comparison. The insights gained from these studies will give a better insight into the evolution of protein sequence and function.
==== Body
Background
Single amino acid repeats (SAARs) are uninterrupted runs of identical amino acids that exist in many proteins and are currently a major focus of research. These are an example of a simple sequence repeat (SSR), which occurs when a simple sequence motif is repeated in the DNA sequence. These repeats are found in the proteome and can eventually dictate the structure and function of proteins. Repeats within the amino acid sequence are usually dependent on repetitive elements in the genome. They originate from unequal crossing-over or replication errors resulting from the formation of unusual DNA secondary structures such as hairpins or slipped strands [1-3]. Amongst the various DNA duplication events, SSRs are abundant in eukaryotic genomes and may be a major source of quantitative genetic variation [4-6]. SSRs in the codingregions of proteins can give rise to a variety of repeats including SAARs, short tandem repeats, and the repetition of homologous domains of 100 or more residues. However the focus of this work is solely on SAARs.
There has been some suggestion that these repeated sequence patterns may be a mechanism that provides regular arrays of spatial and functional groups, useful for structural packing or for one to one interactions with target molecules [7]. This suggests that error-prone SAAR expansion allows the rapid evolution of proteins with repetitive structure, which can lead to rapidly changing phenotypes [8].
Marcotte et al., suggested that eukaryotic proteomes have a significantly higher incidence of SAARs than either bacterial or archaeal proteomes [9]. They showed that most SAARs occur in protein classes associated only with eukaryotes so protein classes associated with both eukaryotes and prokaryotes are much less likely to contain repeats. This would imply that the formation of SAARs is a relatively recent evolutionary event.
What is interesting is that SAARs can be either functionally significant or extremely pathogenic depending on the proteins involved. Several human inherited neurodegenerative diseases are triplet-repeat diseases associated with proteins containing long runs of glutamine (long CAG codon iterations which result from mutations of SAARs) as shown in Table 1[10]. The severity of these diseases seems can be correlated with the extent of iterations of the CAG codon above a certain threshold [11]. Also notable is that most of these proteins contain two or more additional long runs of amino acids other than glutamine [12]. Pathogenicity is due to inflammatory brain responses, oxidative damage and protein aggregations that clog the proteosome [13].
Table 1 Dominantly inherited neurodegenerative diseases are associated with abnormally expanded tracts of glutamine residues.
Disease protein Gln repeats Other notable repeats/comments
Huntington's disease protein 1 SAAR (23-residues) 2x Pro repeats (11-and 10-residues)
2x Glu repeats (6-and 5-residues)
Spinocerebellar ataxin type 1 2 SAARs (15-and 12-residues) The two Gln repeats are separated by 4 residues
Androgen receptor (Kennedy's disease) 3 SAARs (21-, 6-and 5-residues) 1x Pro repeat (8-residues)
1x Ala repeat (5-residues)
1x Gly repeat (24-residues)
Legend: These SAARs are often accompanied by at least 2 other long SAARs of different amino acids. Gln – Glutamine, Pro – Proline, Ala – Alanine, Gly – Glycine, Glu – Glutamic acid.
Examples of functional SAARs can be seen in proteins which are associated with development and transcriptional regulatory capacities, with the majority of them active in central or peripheral nervous system function and development [14]. This has been extensively studied in Drosophila melanogaster but there are also examples in other eukaryotes, for example, the case of the transcription factor II (TFII) in humans [15] which contains a 34 residue glutamine run. This SAAR is absent from all related proteins, and yet appears to be functionally important. Extended runs can also provide substrates for caspase cleavage, yielding tangles, plaques, dead neurons and triggering apoptosis [16]. They also provide binding sites for protein-protein interactions [14].
SAARs are generally less than 20 residues long and are primarily composed of the residues of the amino acids glutamine, asparagine, serine, threonine, proline, histidine, glycine, alanine, aspartic acid and glutamic acid [17,18]. It is curious that glutamine followed by asparagine and serine are the most common SAARs found, especially when considering that the occurrence of leucine, isoleucine, alanine and valine in proteins is much greater. This is particularly interesting when considering that long SAARs of these 4 amino acids are rarely found.
The greatest challenge facing scientists who wish to study SAARs is the lack of tools for analysing SAARs and mining the data collected. While some software exists [19] for detecting and analysing SAARs, it is limited in its application in that it is only designed for analysing single proteins rather than whole proteomes. The aim of this paper is to describe a new web application dedicated to the analysis of SAARs in whole proteomes.
Construction and content
The COPASAAR (COmparative Proteome Analysis of Single Amino Acid Repeats) database was developed in MySQL 4.0.18 running on Mandrake Linux version 10.0. Access to the database is through a web interface written in Perl:CGI and uses the Perl ChartDirector [20] and Descriptive::Statistics modules to generate histograms and statistical analysis of the data. Currently the database contains 244 proteomes, which are made up of 862,886 proteins with a storage requirement of 1.2 Gbytes.
Repeat analysis software
Proteome data files were obtained from the integr8 database at the EBI [21] in Fasta format. These files were analysed for repeats using a series of scripts written in Perl. The database itself was written in SQL and the data was imported into the database as tab delimited text files using the mysqlimport client. This process was automated by the use of shell scripts.
The algorithm used for detecting and measuring a repeat compares each residue with the next one. If it finds two identical residues side-by-side then it continues the comparison to the next residue until it encounters a different amino acid. If a different residue is detected the programme records the repeat in an array of amino acid type and repeat length.
Expected repeat lengths
As a reference to the actual occurrence of SAARs a statistical model was created where the amino acids are assumed to be distributed randomly based on their occurrence in a specific protein [22]. The probability of a SAAR of length n occurring will then be;
P(SAAR of length n) = fn (1 - f)2
Where f is the frequency of the particular amino acid in the protein. The (1-f)2. term accounts for there being a different amino acid at each end of the SAAR.
To find the expected number of repeats of a given amino acid within a protein this probability is multiplied by the number of potential starting points for the repeat. This will be equal to the sequence length minus the length of the repeat plus one.
Expected number of repeats of length n = fn (1 - f)2 (l - (n - 1))
Where l is the length of the protein.
Running times for the software
For all of the currently available proteomes the running time to extract all of the repeats and to generate the expected repeat tables at the protein and proteome level is about 3 hours on a Pentium 4 2.0 GHz. Import of the tables into MySQL is very rapid and takes less than 30 minutes. All of the scripts used to create the database can be downloaded from the COPASAAR website.
The database schema
The database contains 83 tables, most of which contain amino acid specific data. The database schema and table structure are shown in Figure 1. The data is stored at three different levels. At the individual protein, proteome and kingdom levels. While this means there is some redundancy of data this is required to speed up searches so that the amount of analysis that needs to be performed by a query is reduced. For example the expected frequencies of repeats could be calculated during a query from the amino acid occurrences in the proteins, but if this query is at the proteome level this would have to be done for all of the proteins within that proteome and then these would have to be summed to give a proteome level expectation. This would be less efficient and would result in slow querying of the database.
Figure 1 Database schema for COPASAAR. Note that each of the species_repeats, species_expected, protein_repeats and protein_expected tables will be repeated 20 times once for each amino acid.
COPASAAR website
The COPASAAR website houses the user interface to the main programmes, a documentation page featuring software documentation, and a download section so that users can download the database for use on local machines. The user-interface provides menu driven query access to the database. The user simply selects the species they wish to analyse and uses the 'post' method to send the request. Results are displayed either in tabulated or graphical form as bar charts. The website is hosted on an Apache (version 2.0.44) webserver.
COPASAAR proteome data
The current database consists of 244 proteomes; 19 eukaryotic species, 205 prokaryotic species and 20 archaeal species. A full list of the species can be found in additional file 1.
Utility
There have been previous systematic studies of simple amino acid repeat distributions in proteomes [7,14,23,24] but what COPASAAR aims to do is to provide a comprehensive and simple to update resource that means that makes comparative studies much easier to carry out and which also increases the number of biological questions that can be asked.
Adding new proteomes to the database is simple using the repeat analysis scripts and this procedure will be made even easier by the new naming convention for proteome data files that will use the organism name rather than using taxonomic identifiers that can change for the same proteome between database releases.
Access to the database can be either through the web interface or for more experienced users the database can be queried directly using SQL. Accessing the MySQL database directly using SQL allows almost any query to be performed. Figure 2 shows the query to find all proteins with a repeat of 6 alanine residues from the human proteome. The problem with accessing the database this way is that it requires a working knowledge of SQL and also the structure of the database. For this reason the web interface will remain the preferred mode of interaction for novice or infrequent users. The web interface currently contains a set of simple queries that can be rapidly expanded depending on requirements. It is expected that users who download the database will want to implement queries specific to their own research which can be done by customising existing template scripts that generate SQL queries and that format the output either as tables or graphically to be displayed as webpages.
Figure 2 Example SQL script used to query the database for all proteins in humans with an alanine repeat of 6 amino acids.
To illustrate some of the capabilities of the database using the current web interface functionality we have made a high level comparison of occurrence of SAARs across the three super kingdoms.
Discussion
Comparison of archaeal, eukaryotic and bacterial kingdoms
The mean number of amino acids within SAARs (as a percentage of the proteome) within the three super kingdoms is greatest in the eukaryotes at 14.34%. The archaea mean is 13.34% while bacterial proteomes are the lowest with a mean of 13.05% (Table 2a). The overall mean is 13.18%. Of the 19 eukaryotic species, 18 of these proteomes (95%) contain a greater percentage of SAARs than the overall mean compared to 8 out of 20 of the archaea (42%) and 65 out of 205 of the bacteria (32%) (Table 2). If however you look at the maximum length of the repeats between the three kingdoms the distinction between the eukaryotes and the archaea and bacteria is much clearer. In eukaryotes repeats over 20 amino acids occur in most species so far sequenced, although they are less common in the yeasts, whereas for bacteria repeats over 20 residues in length only occur exceptionally in certain Vibrio species (the seafood associated pathogens) and in Lactobacillus plantarum. In archaea a glycine repeat over 20 residues only occurs in Haloarcula marismortui. These results supports the finding of Marcotte [9] and also suggest that the differences between the kingdoms can be specified in terms of repeats for a few amino acids. Glutamine repeats are particularly characteristic of eukaryotes where they have a long tailed distribution, which in the mammals and plants extends beyond 20 residues. It is of particular clinical interest that glutamine repeats play a significant role in eukayotic proteomes because they are associated with amyloid plaque formation in diseases such as Huntington's chorea and spinocerebellar ataxia [25-27]. A functional explanation for the occurrence of glutamine repeats in transcription factor genes has been suggested by Fondon et al. and this could be the main contributing factor to the occurrence of these repeats in eukaryotes [8]. The only eukaryotes where glutamine does not form the longest repeat are Plasmodium falciparum, Arabidopsis thaliana and Caenorhabditis elegans. A. thaliana contains a very characteristic long lysine repeat (over 100 amino acids), while in C. elegans the longest repeat is serine.
Table 2 The proportion of a proteome composed of SAARs and the percentage of proteomes in each kingdom with a greater number of SAARs than the mean. *The overall mean is 13.18%
Kingdom SAARs (as a percentage of the whole proteome) Proteomes (with a greater % of repeats than the overall mean*)
Eukaryotes 14.5% 95%
Archaea 13.3% 45%
Bacteria 13.1% 32%
P. falciparum, has a very unusual repeat distribution that is different to all other proteomes, prokaryotes and archaea included. Nearly 20% of the P. falciparum proteome is made up of repeats. The distribution of asparagine repeats is particularly significant. There are 137 repeats of over 20 asparagines in length which is highly unusual as long asparagine repeats are associated with prion domains and fibril formation [28,29]
The amino acid compositions of SAARs across the kingdoms are shown in Table 3. The eukaryotes, feature leucine, serine and glutamic acid as the top three constituents. Archaea features leucine and glutamic acid as its top two constituents, while bacteria feature leucine and alanine as the top two constituents. These results agree with the overall distributions of amino acids in the three kingdoms, but although leucine appears in many short repeats and so makes a large contribution to the number of amino acids in SAARs it very rarely has long repeats in any proteome and it is the longest repeat in only a few bacterial species.
Table 3 SAARs composition by amino acid.
Amino Acid Eukaryotes Archaea Bacteria
Arginine 0.75% 0.75% 0.74%
Lysine 1.0% 1.05% 0.72%
Glutamic Acid 1.28% 1.45% 0.89%
Aspartic Acid 0.67% 0.65% 0.55%
Glutamine 0.57% 0.18% 0.42%
Asparagine 0.63% 0.37% 0.35%
Histidine 0.17% 0.09% 0.13%
Proline 0.83% 0.42% 0.43%
Tyrosine 0.23% 0.37% 0.21%
Tryptophan 0.04% 0.04% 0.04%
Serine 1.76% 0.90% 0.85%
Threonine 0.68% 0.60% 0.59%
Glycine 0.90% 1.10% 1.09%
Alanine 1.14% 1.38% 1.92%
Methionine 0.10% 0.11% 0.12%
Cysteine 0.10% 0.03% 0.04%
Phenylalanine 0.38% 0.38% 0.37%
Leucine 1.87% 1.95% 2.02%
Valine 0.80% 1.31% 1.03%
Isoleucine 0.61% 1.23% 0.74%
Legend: The dominant amino acids are leucine and serine (Eukaryotes), leucine and glutamic acid (Archaea), and leucine and alanine (Bacteria). Amino acids considered highly abundant are highlighted in bold.
Prediction model
The prediction model shows a close correlation to the actual repeat distribution in many cases and in particular for short SAARs although there is a consistent slight under-estimation of the number of expected repeats. This would suggest that shorter repeats are mostly randomly distributed and that few of them are likely to be functionally significant. Short repeats are therefore likely to form part of the neutral drift of protein sequence evolution.
Conclusion
COPASAAR provides an essential tool for the study of repeats in comparative proteomics. The ability to quickly analyse proteomes (and individual proteins) and to map the distribution and size of SAARs will hopefully benefit scientists from many different fields. COPASAAR will provide a useful resource for finding new protein families that can be used as species specific markers. Data on the evolution of repeats between species will also allow us to develop models of adaptive traits in proteomes. This will be particularly important in understanding the evolution of amyloid associated diseases.
Availability and requirements
Online access to COPASAAR can be found at;
All of the source code for the project and the database files are also available from this site and are available under the GPL. Software requirements have been described above and non-academics should be aware of licensing restrictions regarding the use of the commercial software Perl ChartDirector.
Authors' contributions
Both authors contributed to the design of COPASAAR and the underlying algorithms. The implementation of the system was carried out by DPD. Both DPD and ARD contributed to the final draft of the paper.
Supplementary Material
Additional File 1
List of the proteomes in the database. Gives the species name and the FASTA identification number for the proteome
Click here for file
Acknowledgements
DPD would like to acknowledge an EPSRC studentship that funded this work. ARD would like to acknowledge Simon Lofting for his comments on the statistical analysis.
==== Refs
Pearson CE Sinden RR Trinucleotide repeat DNA structures: dynamic mutations from dynamic DNA Curr Opin Struct Biol 1998 8 321 330 9666328 10.1016/S0959-440X(98)80065-1
Kruglyak S Durrett R Schug MD Aquadro CF Distribution and abundance of microsatellites in the yeast genome can be explained by a balance between slippage events and point mutations Mol Biol Evol 2000 17 1210 1219 10908641
LeProust EM Pearso CE Sinden RR Gao XL Unexpected formation of parallel duplex in GAA and TTC trinucleotide repeats of Friedreich's ataxia J Mol Biol 2000 302 1063 1080 11183775 10.1006/jmbi.2000.4073
Kashi Y King D Soller M Simple sequence repeats as a source of quantitative genetic variation Trends Genet 1997 13 74 78 9055609 10.1016/S0168-9525(97)01008-1
Alba MM Santibanez-Koref MF Hancock JM Conservation of polyglutamine tract size between mice and humans depends on codon interruption Mol Biol Evol 1999 16 1641 1644 10555295
Alba MM Guigo R Comparative analysis of amino acid repeats in rodents and humans Genome Res 2004 14 549 554 15059995 10.1101/gr.1925704
Katti MV Sami-Subbu R Ranjekar PK Gupta VS Amino acid repeat patterns in protein sequences: Their diversity and structural-functional implications Protein Sci 2000 9 1203 1209 10892812
Fondon JW Garner HR Molecular origins of rapid and continuous morphological evolution Proc Natl Acad Sci U S A 2004 101 18058 18063 15596718 10.1073/pnas.0408118101
Marcotte EM Pellegrini M Yeates TO Eisenberg D A census of protein repeats J Mol Biol 1999 293 151 160 10512723 10.1006/jmbi.1999.3136
Djian P Evolution of simple repeats in DNA and their relation to human disease Cell 1998 94 155 160 9695944 10.1016/S0092-8674(00)81415-4
Sutherland GR Richards RI The Molecular-Basis of Fragile Sites in Human-Chromosomes Curr Opin Genet Dev 1995 5 323 327 7549426 10.1016/0959-437X(95)80046-8
Karlin S Burge C Trinucleotide repeats and long homopeptides in genes and proteins associated with nervous system disease and development Proc Natl Acad Sci U S A 1996 93 1560 1565 8643671 10.1073/pnas.93.4.1560
Bence NF Sampat RM Kopito RR Impairment of the ubiquitin-proteasome system by protein aggregation Science 2001 292 1552 1555 11375494 10.1126/science.292.5521.1552
Karlin S Brocchieri L Bergman A Mrazek J Gentles AJ Amino acid runs in eukaryotic proteomes and disease associations Proc Natl Acad Sci U S A 2002 99 333 338 11782551 10.1073/pnas.012608599
Hoffmann A Sinn E Yamamoto T Wang J Roy A Horikoshi M Roeder RG Highly Conserved Core Domain and Unique N-Terminus with Presumptive Regulatory Motifs in a Human Tata Factor (Tfiid) Nature 1990 346 387 390 2374612 10.1038/346387a0
Sun B Fan W Balciunas A Cooper JK Bitan G Steavenson S Denis PE Young Y Adler B Daugherty L Manoukian R Elliott G Shen WY Talvenheimo J Teplow DB Haniu M Haldankar R Wypych J Ross CA Citron M Richards WG Polyglutamine repeat length-dependent proteolysis of huntingtin Neurobiol Dis 2002 11 111 122 12460551 10.1006/nbdi.2002.0539
Huntley MA Golding GB Simple sequences are rare in the protein data bank Proteins 2002 48 134 140 12012345 10.1002/prot.10150
Nance MA Clinical aspects of CAG repeat diseases Brain Pathol 1997 7 881 900 9217974
Pellegrini M Marcotte EM Yeates TO A fast algorithm for genome-wide analysis of proteins with repeated sequences Proteins 1999 35 440 446 10382671 10.1002/(SICI)1097-0134(19990601)35:4<440::AID-PROT7>3.0.CO;2-Y
Advanced Software Engineering
integr8
Brendel V Bucher P Nourbakhsh IR Blaisdell BE Karlin S Methods and Algorithms for Statistical-Analysis of Protein Sequences Proc Natl Acad Sci U S A 1992 89 2002 2006 1549558
Sim KL Creamer TP Abundance and distributions of eukaryote protein simple sequences Mol Cell Proteomics 2002 1 983 995 12543934 10.1074/mcp.M200032-MCP200
Sim KL Creamer TP Protein simple sequence conservation Proteins 2004 54 629 638 14997559 10.1002/prot.10623
Ross CA Margolis RL Huntington's disease Clin Neurosci Res 2001 1 142 152 10.1016/S1566-2772(00)00014-1
Cervantes-Kardasch VH Garcia-Martinez E Molecular physiopathology of the spinocerebellar ataxia type 6 (SCA6) Rev Invest Clin 2004 56 368 374 15612520
Poirier MA Jiang H Ross CA A structure-based analysis of huntingtin mutant polyglutamine aggregation and toxicity: evidence for a compact beta-sheet structure Hum Mol Genet 2005 14 765 774 15689354 10.1093/hmg/ddi071
Singh GP Chandra BR Bhattacharya A Akhouri RR Singh SK Sharma A Hyper-expansion of asparagines correlates with an abundance of proteins with prion-like domains in Plasmodium falciparum Mol Biochem Parasitol 2004 137 307 319 15383301 10.1016/j.molbiopara.2004.05.016
Kreil DP Kreil G Asparagine repeats are rare in mammalian proteins Trends Biochem Sci 2000 25 270 1 10838564 10.1016/S0968-0004(00)01594-2
|
16078990
|
PMC1199582
|
CC BY
|
2021-01-04 16:27:26
|
no
|
BMC Bioinformatics. 2005 Aug 3; 6:196
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-196
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-1991608683110.1186/1471-2105-6-199Methodology ArticleA comparative review of estimates of the proportion unchanged genes and the false discovery rate Broberg Per [email protected] Biological Sciences, AstraZeneca R&D Lund, S-221 87 Lund, Sweden2005 8 8 2005 6 199 199 22 12 2004 8 8 2005 Copyright © 2005 Broberg; licensee BioMed Central Ltd.2005Broberg; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
In the analysis of microarray data one generally produces a vector of p-values that for each gene give the likelihood of obtaining equally strong evidence of change by pure chance. The distribution of these p-values is a mixture of two components corresponding to the changed genes and the unchanged ones. The focus of this article is how to estimate the proportion unchanged and the false discovery rate (FDR) and how to make inferences based on these concepts. Six published methods for estimating the proportion unchanged genes are reviewed, two alternatives are presented, and all are tested on both simulated and real data. All estimates but one make do without any parametric assumptions concerning the distributions of the p-values. Furthermore, the estimation and use of the FDR and the closely related q-value is illustrated with examples. Five published estimates of the FDR and one new are presented and tested. Implementations in R code are available.
Results
A simulation model based on the distribution of real microarray data plus two real data sets were used to assess the methods. The proposed alternative methods for estimating the proportion unchanged fared very well, and gave evidence of low bias and very low variance. Different methods perform well depending upon whether there are few or many regulated genes. Furthermore, the methods for estimating FDR showed a varying performance, and were sometimes misleading. The new method had a very low error.
Conclusion
The concept of the q-value or false discovery rate is useful in practical research, despite some theoretical and practical shortcomings. However, it seems possible to challenge the performance of the published methods, and there is likely scope for further developing the estimates of the FDR. The new methods provide the scientist with more options to choose a suitable method for any particular experiment. The article advocates the use of the conjoint information regarding false positive and negative rates as well as the proportion unchanged when identifying changed genes.
==== Body
Background
The microarray technology permits the simultaneous measurement of the transcription of thousands of genes. The analysis of such data has however turned out to be quite a challenge. In drug discovery, one would like to know what genes are involved in certain pathological processes, or what genes are affected by the intervention of a particular compound. A more basic question is 'How many genes are affected or changed?' It turns out that the answer to this basic question has a bearing on the other questions.
The proportion unchanged
In the two-component model for the distribution of the test statistic the mixing parameter p0, which represents the proportion unchanged genes, is not estimable without strong distributional assumptions, see [1]. Assuming this model the probability density function (pdf) ft of a test statistic t may be written as the weighted sum of the null distribution pdf and the alternative distribution pdf
If, on the other hand, we know the value of p0 we can estimate e.g. through a bootstrap procedure as described in [1], and thus obtain also .
The mixing parameter p0 has attracted a lot of interest lately. Indeed it is interesting for a number of applications. Here follow four examples.
1) Knowing the proportion changed genes in a microarray experiment is of interest in its own right. It gives an important global measure of the extent of the changes studied.
2) The next example concerns FDR. Suppose we reject null hypothesis j, and call gene j significantly regulated, when the corresponding p-value pj falls below some cutpoint α. The question that motivates the FDR concept, which originates from [2], is: "What proportion of false positives is expected among the selected genes?" A goal would then be to quantify this proportion, and one possible estimate is
where '^' above a quantity (here and henceforth) means that it is a parameter estimate, P(L) is the largest p-value not exceeding α and p(α) is the proportion significantly regulated genes which equals the proportion of the p-values not exceeding α, see also [3]. In practice P(L) will be very close to α, and may be replaced by the latter. We thus obtain an estimate of p0 × α/p(α), which verbally translates into "the number of true null cases (Np0) multiplied by their probability of ending up on the top list (α) divided by the number of selected cases (Np(α))". Putting p0 ≡ 1 above and rejecting hypotheses j with the estimated FDR(pj) less than or equal to β, will give a test procedure controlling the FDR at the level β, i.e. we may expect that the FDR is no more than β [2-4]. By finding good estimates of p0 (and FDR) we may increase the power to detect more true positives at a given FDR bound.
3) Knowing p0, we may calculate the posterior probability of a gene being a Differentially Expressed Gene (a DEG) as
see [1]. Also, (3) equals one minus the local false discovery rate: 1 - p1(x) = LFDR(x) = p0 ft0(x) / ft(x) [1].
4) Knowing p0, it is also possible to estimate the number of false positives and false negatives at a given cutpoint α as a proportion of the total number of genes. Call these proportions the false-positive and false-negative rates, and denote them by FP(α) and FN(α), respectively. In the samroc methodology [5] one calculates estimates of these quantities as
and
One may choose a p-value threshold αmin, which minimises the amount of errors FP(α) + FN(α). Alternatively, one may want to fine-tune the test statistic such that it will minimise the errors at a given threshold. Or, one may try to do both, as suggested in [5], see also [6].
Earlier research providing estimates of p0 include [1,3,7-13]. Articles that compare methods for estimating p0 and FDR include [12,14]. I will focus on the FDR as the main use of p0.
In this article, the formulation of the theory is in terms of p-values rather than in terms of test statistics. Two basic assumptions are made concerning their distribution. First, it is assumed that test statistics corresponding to true null hypotheses will generate p-values that follow a uniform distribution on the unit interval, e.g. [15]. Thus, under the null distribution, the probability that a p-value falls below some cutpoint α equals α. Second, p-values are, unless stated otherwise, assumed to be independent. Empirical investigations will assess the effects of deviations from the second assumption.
The use of p-values means lumping up- and downregulated genes together. However, one may look separately at the two tails of the distribution of the test statistic to assess differential expression corresponding to up- and downregulation.
This article will not concern how p-values are calculated, but rather how they are used to calculate estimates of p0 and FDR, and draw conclusions based on this evidence. It is assumed that p-values capture the essence of the research problem. Neither does the article treat the choice of an optimal test statistic. For illustration the t-test will be used repeatedly without regard to whether there are better methods or not. By the t-test we mean the unequal variance t-test: for sample means mean1 and mean2, sample variances , , and sample sizes n1 and n2. We apply the t-test to simulated normally distributed data and a permutation t-test to real data, where normality may be uncertain. All calculations were performed in R. The methods presented are available within packages for the free statistical software R [16,17] and take a vector of p-values as input and output an estimate of p0 and of FDR. Emphasis lies on methods available within R packages downloadable from CRAN [18] or Bioconductor [19]. Inevitably any review will exclude interesting work, but time and space limitations will not permit an all comprehensive review. The new and highly interesting concept of a local false discovery rate (LFDR) [1] only receives a cursory treatment.
This article builds on and finds motivation from the experience of the analysis of microarrays, which typically assay the expression of 10,000 or more genes. However, the methods presented apply equally well to other high dimensional technologies, such as fMRI or Mass Spectrometry.
False discovery rate
In the analysis of microarray experiments, the traditional multiple test procedures are often considered too stringent, e.g. [20] and [3]. In the last decade alternatives based on the concept of an FDR have emerged. For more details consult e.g. [2,3,9,11,21,22]. There are different definitions proposed, but loosely speaking one would want to measure the proportion of false positive genes among those selected or significant. Loosely put the FDR may be interpreted as the proportion of false positives among those genes judged significantly regulated. Equation (2) is the FDR estimate presented in [3].
Denote by E[X] the expectation (or true mean) of any random variable X. With V the number of false positives given a certain cut-off and R the number of rejected null hypotheses, one may define the FDR as the expectation of the ratio of these quantities, or
,
where care is taken to avoid division by zero.
In [10] and [11] the FDR is estimated as the ratio of the expected proportion of false positives given the cut-off to the expected proportion selected. Viewed as a function of a cut-off α, such that genes gi with pi less than α are judged significant in terms of p-values, following the continuous cumulative distribution function (cdf) F, the FDR estimate is
which is nearly equal to (2) with the exception that the P(L) has been replaced by its upper bound, and the step-wise empirical distribution by a smooth version, either a parametric model or a smoothed version of the empirical distribution. Thus, the FDR is now a continuous function instead of piece-wise continuous with jumps at each observed p-value. This ratio of expected values tries to estimate the expectation of a ratio: In general such an approach will give an overestimation, but in practice this will have little effect, see the Additional file.
The related concept of the positive FDR, pFDR = E[V/R|R > 0], the expectation conditional on at least one rejection, appears in [2]. Other forms of FDR have been proposed such as the conditional FDR [2], cFDR, defined as the expected proportion of false positives conditional on the event that R = r rejected have been observed : cFDR(r) = E[V|R = r]/r. This would answer to the question "What proportion of false positives may I expect in my top list of r genes?". Under independence and identical distribution in a Bayesian setting it is proved in [23], that pFDR, cFDR and the marginal FDR, mFDR = E[V]/E[R] [2], all coincide with p0α/F(α) at the cutpoint α, cf. (6).
Instead of p-values it has been suggested in to calculate q-values that verbally have the following meaning for an individual gene [2,9]:
The q-value for a particular gene is the minimum false discovery rate that can be attained when calling all genes up through that one significant [9].
These q-values can be used to determine a cut-off similar to the classic 5% cut-off for univariate tests developed in statistics long ago. But in many applications one should not be too rigid about any particular value, since the emphasis often is on discovery rather than hypothesis testing: we generate hypotheses worthy of further investigation. Thus the balance between false positives and false negatives will be crucial: Rather than keeping the risk of erroneously selecting one individual gene at a fixed level, it is the decision involving thousands of genes given the amount of genes we can follow up on that is the focus, and where both types of error must be considered. The q-value does not fully address this problem, but nevertheless represents an improvement over the classical multiple test procedures in these applications.
More mathematically the q-value can be expressed as
Taking minimum in (7) enforces monotonicity in pi, so that the q-value will be increasing (non-decreasing) in the observed p-value. If the FDR is non-increasing, as it should, then q (pi) = FDR (pi).
Additionally, the FDR offers a framework for power and sample size calculations, see [24] and the new developments in [25].
Results
Eight estimates of p0 and six of FDR (based on six of the former) were tested on both simulated data and real data. The differing numbers are motivated below.
Next follows a list of six p0 estimation methods and the corresponding R functions. The six were
1. the beta-uniform model (BUM) [10], which fits a mixture of a uniform and a beta distribution to the observed p-values; function ext.pi.
2. spacing LOESS histogram (SPLOSH) [11], which fits a non-parametric spline that estimates the logarithm of the pdf;function splosh.
3. the Lowest SLope estimator (LSL) [12,26] ;function fdr.estimate.eta0.
4. the smoother [9], which fits a spline to a function of a cut-off value, namely the proportion of p-values greater than that cut-off divided by the expected proportion under the uniform distribution;function qvalue.
5. the bootstrap least squares estimate (bootstrap LSE) [3], which is related to the previous estimate;function qvalue or estimatep0.
6. the Successive Elimination Procedure (SEP) [13];selects a subset which represents the null distribution by behaving like a uniform;function twilight.
7. a new method based on a moment generating function approach (mgf);function p0.mom.
8. a Poisson regression approach (PRE); an adaptation of [27,28];function p0.mom.
The bootstrap estimate and mgf did not participate in the calculation of FDR. The smoother gives the basically same value as the bootstrap estimate, and mgf is unnecessarily conservative for lower values of p0, compared to PRE.
The six were
1. BUM FDR (based on BUM;function bum.FDR)
2. BH FDR (based on LSL and function fdr.control).
3. SPLOSH FDR (based SPLOSH;function splosh)
4. smoother FDR or R function qvalue [9] (based on the smoother)
5. SEP fdr (based on SEP;function twilight)
6. the new method pava FDR (based on PRE;function pava.fdr)
For brevity mgf, PRE and pava FDR will all be referred to as new methods. It would be more exhaustive to say that PRE is a minor modification of an existing method [25] applied to p-values rather than test statistics and provided as a new implementation in R; and that pava fdr is based on [29] with local splines replaced by isotonic regression and provided as a new R function. On the other hand mgf seems quite new. More details follow in Methods.
For reference some graphs include an estimate of the SEP local FDR, defined as LFDR(p) = p0/f(p), estimating the probability that a gene whose p-value equals p is a false positive. Furthermore, the ouput from R function locfdr applied to the real life data (with nulltype = 0, i.e. a standard normal distribution which has cdf Φ, see Methods) and the transformed t-test statistics : Z = Φ-1(F(t)), and F the t-test distribution (details below) gives perspective on the other methods and opens up an alternative route to making inferences. This function produces an estimate of the local FDR as a function of the transformed test statistic Z [25]. In that same reference the author argues in favour of the cutpoint LFDR ≤ 0.2, which implies quite high posterior odds in favour of the non-null case : (1-p0)f1/p0f0 ≥ 4.
Simulated data
Two simulation models were used: one generating values independent between genes and the other generating observations displaying clumpy dependence [14,30].
Simulated independent data corresponding to two groups of four samples each were generated 10,000 genes 400 times for each of the true p0 values ranging from 0.5 to 0.99 using the R script from [5]. Normal distributions were chosen randomly from the ones in Table 1. The cases of DEGs correspond to the power of 41%, 54% and 97% given a t-test with a significance level of 5% [12]. Briefly, the script generated a mixture of normal distributions and for each run calculated t-tests to obtain p-values.
Table 1 Simulation of independent data. Denote by N(μ, σ) a normal distribution with mean μ and standard deviation σ. For each DEG one of the above three scenarios was chosen with equal probabilities. For the rest both groups follow the same distribution chosen randomly from the 'Group 1' column. The scenarios are such that the power to detect the regulation with a 5% two-sided t-test ranges from small to large given two groups of size four.
Scenario Group 1 Group 2 Power
1 N(6, 0.1) N(6.1, 0.1) 0.19
2 N(8, 0.2) N(8.5, 0.2) 0.79
3 N(10, 0.4) N(11, 0.7) 0.47
To generate dependent data the protocol from [14] was followed. This generates data following clumpy dependence in the sense of [30] such that blocks of genes have dependent expression. First a logarithmic normal distribution with mean 1.5 and standard deviation 0.3 generated a profile for each gene. Denoting by N(μ, σ) a normal distribution with mean μ and standard deviation σ, random errors following a standard normal distribution N(0,1) were added. To create dependencies genes were partitioned into sets of 50 and for each sample the same term from a N(0, 1) distribution was added to the expression of each gene in the set. Finally, genes were randomly assigned to become DEGs with probability 1-p0 and for each gene a regulation term following either N(0.5, 0.2) or N(0.7, 0.2), with equal probabilities, was added to the expression of one of the groups of samples of size 30. The power to detect either of these two alternatives with a t-test at the 5% significance level equals 31% and 50%, respectively. The procedure generated for each of 400 iterations a set of observations of 10,000 genes. The protocol gives rise to high correlation within the blocks of 50 (on average on the order of 0.5). Results for smaller or weakly dependent datasets appear in the Additional file. Here weakly dependent means that the clumpy dependence term follows a N(0,1/20) distribution (correlation within blocks slumps to 0.003 on an average).
With the simulated independent data all methods for estimating p0 perform rather well, see Table 2 and 3, and Figure 1.
Table 2 Over-all results of simulations of independent data. Data sets with p0 ranging from 0.6 to 0.99 were simulated. The summary statistics of the absolute difference between target value and its estimate show a rather varying performance for all methods, with PRE having the smallest bias and variation.
BUM SPLOSH smoother bootstrap SEP LSL mgf PRE
Mean 0.039 0.061 0.038 0.036 0.045 0.18 0.072 0.022
Sd 0.048 0.078 0.032 0.034 0.032 0.12 0.048 0.016
Table 3 Detailed statistics on the estimates of p0 based on simulations of independent data. The displays the mean bias (true – estimated) and standard deviation of estimates for each level of true p0.
True p0 0.5 0.6 0.7 0.8 0.9 0.95 0.99
BUM mean bias 0.013 0.043 0.038 0.028 -0.090 -0.050 -0.010
Sd 0.0044 0.0074 0.0072 0.0076 0.033 0.0021 0.0010
SPLOSH mean bias -0.083 -0.068 -0.043 -0.012 0.018 0.026 0.14
Sd 0.014 0.020 0.028 0.034 0.040 0.041 0.063
QVALUE mean bias -0.066 -0.057 -0.046 -0.034 -0.019 -0.015 0.0012
Sd 0.022 0.022 0.023 0.027 0.027 0.024 0.0160
Bootstrap LSE mean bias -0.065 -0.054 -0.040 -0.025 -0.0067 -0.0037 0.0057
Sd 0.023 0.023 0.021 0.024 0.026 0.024 0.023
SEP mean bias -0.084 -0.072 -0.059 -0.043 -0.028 -0.020 0.0068
Sd 0.014 0.013 0.014 0.013 0.013 0.012 0.014
LSL Mean bias -0.36 -0.31 -0.25 -0.18 -0.096 -0.049 -0.010
Sd 0.025 0.019 0.014 0.0086 0.0036 0.0022 0.0010
mgf mean bias -0.15 -0.12 -0.095 -0.067 -0.040 -0.027 -0.0095
Sd 0.0052 0.0052 0.0051 0.0056 0.0056 0.01098 0.0018
PRE mean bias -0.036 -0.033 -0.028 -0.018 -0.0078 -0.018 -0.0018
Sd 0.0098 0.010 0.0097 0.012 0.016 0.0080 0.0067
Figure 1 Boxplots of the results from eight methods for estimating p0 using simulated independent data. A simulation model of real-life microarray data was used to give data where the expected proportion of unchanged genes was set at 50, 60, 70, 80, 90, 95 or 99%. The central box of the boxplot represents the data between the 25% and 75% percentiles with the median represented by a line. Whiskers go out to the extremes of the data, and very extreme data appear as points. The abscissa shows the expected proportion unchanged and ordinata the estimate. Legends : A – BUM;B – SPLOSH; C – smoother; D – bootstrap; E – SEP; F – LSL; G – mgf; H – PRE. A and B underestimate in the low range and in the high range, respectively; C and D appear very similar with a sound overestimation and some high variance; E appears stable and reliable; F gives a stable and quite conservative estimate; G overestimates and varies very little; G is stable and gives a small degree of overestimation.
The new methods mgf and PRE were very competitive on these data, and had both low bias and variation, excluding mgf at the 0.5 and 0.6 level. Since mgf tends to overestimate p0 rather much in the lower range, one may prefer PRE. For practical purposes though overestimation is desirable and enables control of the error rate (exact control in the terminology from [31]).
The smoother and the bootstrap had good and quite similar performance. They give more or less the same variation and bias over the whole range. This variation can be a bit high though, especially when comparing to PRE.
In the higher range BUM gives a crude estimate of the true p0. In a certain lower range however it underestimates. As we can see in Figure 1, however, the method considerably overestimates p0 in the higher range, which brings down the power to detect DEGs.
SPLOSH has the advantage of fitting the observed distribution quite well, judging from some tests (data not shown), compare also [11]. This enables a Bayesian analysis as in (3). However, it does the fitting of the p-values close to 0 sometimes at the expense of the accuracy concerning the values at the other end, thus misses the plateau and the minimisation in (10) will give a misleading result. In particular, this tends to happen when there are few DEGs. As we can see in Figure 1 the method underestimates p0 at the higher range (p0 ≥ 0.9), which is worrisome and may lead to underestimation of the error rate, which is undesirable for a method of statistical inference.
From Tables 2 and 3 we can see that PRE has the best over-all performance, followed by the smoother and the bootstrap. This does not however imply that the other methods could not be considered. The results vary quite a lot depending on the value of p0: LSL is quite competitive for p0 = 0.99, but too conservative for p0 = 0.5.
With simulated data it is possible to calculate the actual false discovery rate. Figure 3 shows boxplots based on the simulated independent data. Here qvalue and pava FDR stand out as most reliable. Especially BUM FDR has severe problems, often over the whole range of cut-offs, while SPLOSH FDR does rather well.
Figure 2 Boxplots of the results from eight methods for estimating p0 using simulated dependent data. Legend as in Figure 1. There is a general trend towards greater overestimation compared to the independent case.
The dependent data gave a slightly different picture. Most importantly, the variation increased considerably. The results concerning p0 appear in Figure 2. Interestingly underestimation becomes less of a problem for SPLOSH this time. Relatively speaking BUM did better this time, and BH comes out worst. As far as estimation of p0 goes BUM does quite well, see Tables 4 and 5. The method manages to overestimate at high p0 by consistently outputting 1. At p0 = 0.95 the estimate was nearly always equal to 1. Both qvalue and bootstrap underestimated by a small amount at p0 = 0.99. SEP emerges as a sharp contender, but suffers from a small underestimation at p0 = 0.99. In fact, all methods except BUM give underestimation at p0 = 0.99, and for SPLOSH it amounts to almost 7%.
Table 4 Over-all results of simulations of dependent data. Data sets with p0 ranging from 0.6 to 0.99 were simulated. The summary statistics of the absolute difference between target value and its estimate show a rather varying performance for all methods, with BUM now having the smallest bias and variation with PRE in second place.
BUM SPLOSH Smoother Bootstrap SEP LSL Mgf PRE
Mean 0.054 0.075 0.07.3 0.085 0.071 0.125 0.091 0.064
Sd 0.035 0.086 0.062 0.076 0.062 0.124 0.069 0.060
Table 5 Detailed statistics on the estimates of p0based on simulations of dependent data. The table displays the mean bias (true – estimated) and standard deviation of estimates for each level of true p0.
True p0 0.5 0.6 0.7 0.8 0.9 0.95 0.99
BUM mean bias -0.09885 -0.0744 -0.0524 -0.0308 -0.0525 -0.0493 -0.0100
Sd 0.01555 0.0167 0.0169 0.0194 0.0520 0.0088 0.0010
SPLOSH mean bias -0.15811 -0.1179 -0.0710 -0.02570 0.0180 0.0318 0.06761
Sd 0.02702 0.0309 0.0338 0.0398 0.0423 0.0379 0.0560
QVALUE mean bias -0.15089 -0.1217 -0.0895 -0.0593 -0.0307 -0.0142 0.0093
Sd 0.02661 0.0306 0.0302 0.0345 0.0368 0.0309 0.0255
Bootstrap LSE mean bias -0.18009 -0.1499 -0.1080 -0.0691 -0.0313 -0.0126 0.0084
Sd 0.04242 0.0382 0.0351 0.0326 0.0299 0.0269 0.0232
SEP mean bias -0.15525 -0.1238 -0.0915 -0.0584 -0.0277 -0.0134 0.0138
Sd 0.02419 0.0239 0.0241 0.0265 0.0269 0.0203 0.0201
LSL Mean Bias -0.2710 -0.2203 -0.1667 -0.1110 -0.0554 -0.0292 0.0017
Sd 0.13333 0.1119 0.0902 0.0671 0.0410 0.0221 0.0165
Mgf mean bias -0.18322 -0.1468 -0.1349 -0.0898 -0.0460 -0.0235 0.0007
Sd 0.04633 0.0401 0.0140 0.0153 0.0160 0.0153 0.0128
PRE mean bias 0.14088 -0.1091 -0.0770 -0.0442 -0.0152 -0.0021 0.0194
Sd 0.02300 0.0239 0.0239 0.0291 0.0310 0.0269 0.0248
The results concerning FDR appear in Figure 4. This time the BUM method captures FDR in a competitive way, with the lowest median error but at the same time the second worst mean error. Again the results for pava FDR appear quite stable and accurate, giving the second lowest median error and the lowest mean error.
Figure 4 Simulations of FDR based on dependent data. All methods perform worse on dependent data. All except BH perform reasonably well. In the case of BH the problem lies mainly in the lower range of p0.
Results for 300, 5,000 and 10,000 simulated weakly dependent genes appear in the Additional file. Briefly, they resemble those of the independent case. For 300 genes however the variation is such that the value of using these methods seems doubtful.
Real data
Data from Golub et al
These data represent a case where there are many DEGs. The data set concerning two types of leukaemia, ALL and AML, appeared in [32,33]. Samples of both types were hybridised to 38 Affymetrix HG6800 arrays, representing 27 ALL and 11 AML. In reference [32] 50 genes were identified as DEGs using statistical analysis. The data consisted of Average Difference values and Absolute Calls, giving for each gene (probe set), respectively, the abundance and a categorical assessment of whether the gene was deemed Absent, Marginal or Present. The Average Difference values were pre-processed as in [5], and the proportion of samples where each probe set was scored Present was calculated, giving a present rate. A permutation t-test was used to compare ALL and AML. Figure 5 shows a histogram of t-test statistics revealing a bell-shaped region around the origin and large tails, indicating that a substantial part of the genes are DEGs.
Figure 5 The empirical distribution of the t-test statistic based on the data from Golub et al. The test compares the sample types AML and ALL. A comparison with the theoretical null distribution, Student's t-distribution, hints that there are too many extreme values to be accounted for by pure chance. In fact, evidence is that a substantial part of the genes differ.
Visual inspection of Figure 6 showing the p-value distribution also suggests that many genes are altered. The methods tested vindicate this : BUM, SPLOSH, bootstrap, qvalue, mgf, PRE, SEP, LSL give the estimates 0.57, 0.62, 0.65, 0.65, 0.69, 0.60, 0.64 and 0.86, respectively. Thus, roughly 35% of the genes are regarded as changed. The function qvalue finds 873 probe sets with a q-value less than 5%. The estimated of FDR appear in Figures 7 and 8, and show a great deal of concordance between methods. The curves rise to the respective estimate of p0 and and in doing so slowly diverge. SEP approaches roughly 0.07 in a vicinity of zero, while the other methods continue the decline.
Figure 6 Golub et al. data: p-values. The blue and red lines represent the minimum and maximum estimate of p0 obtained from the six methods under investigation. The plateau to the right resembles a uniform distribution, and most likely to a large degree represents a population of unchanged genes.
Figure 7 FDR for the Golub et al. [32] data. The dotted line represents a cut-off that minimises the number of false positives and false negatives, summing (4) and (5) using the smoother estimate. It would make sense to consider choosing a threshold in this region. The line corresponds to the 2920th ordered p-value and a q-value of just above 0.2. All methods agree rather well. For reference the local FDR of [29] has been added. As expected it exceeds all others, and is caught up at zero by its FDR counterpart SEP FDR. However, they meet at a level well above other methods.
Figure 8 A blow-up of the region to the left of the dotted line in Figure 7. The SEP estimates of FDR and LFDR join at roughly 0.07.
The function locfdr applied to Z = Φ-1 (F36(t)) with F36 the cdf of the t-distribution with 36 degrees of freedom gives the estimate of p0 0.66 and outputs Figure 9 which bears witness of the skewness and of the different local false discovery rates in the two tails of the distribution.
Figure 9 The local fdr from locfdr applied to data from [32]. Using Z statistics based on the transformed test statistics (in R code Z = qnorm(pt(tt, df = 36)), pct0 = 0.5 and a standard normal density as null type), the function outputs the estimate p0 = 0.66 and a plot of the local fdr which reflects the skewness of the test statistic distribution. The requirement that LFDR be less than 0.2 would identify 1198 genes as significant.
Removing probe sets with less than 20% present rate will leave us with 2999 probe sets, and qvalue indicates that there are 977 of them that are significant with a q-value less than 5% (p0 = 0.47). In general it is wise to remove probe sets with low presence rate prior to analysis, since this will make the inference more reliable, compare [20]. Doing so will most likely produce more true positives.
Data from Spira et al
Next let us turn to a case where there are rather few DEGs. In [34] the results from a microarray experiment where bronchial epithelial brush biopsies have been hybridised to Affymetrix U133A arrays are presented. The biopsies come from three different subject categories: Current smokers, Never smokers and Former smokers. The cel intensity files were downloaded from the NCBI Gene Expression Omnibus (accession no. GSE994) [35]. The Bioconductor package affy [19] was used to normalise intensities with the quantile method, and to calculate the RMA measure of abundance. The function mas5calls in affy output absolute calls.
Here we will take a brief look at the comparison between Former smokers and Never smokers. The comparison may help identify genes that remain changed after smoke cessation. A fuller analysis would include more analyses, such as the Current smokers vs. Former smokers comparison, and possibly also adjust for the fact that Former smokers tend to be older than Never smokers (Mean Age 45 and 53 years, respectively).
Graphical representations of the results appear in Figures 10 through 14.
Figure 10 Spira et al. data: Former smokers vs. Never smokers t-test statistic. The shape of the histogram indicates that there are more genes that are more highly expressed in Former smokers than the other way round, but on the whole there are rather few DEGs. Indeed, the estimates of p0 vindicate this and suggest that less than 10% of the genes are DEGs. Compare Figure 11.
Figure 14 Output from locfdr applied to the Spira et al. data. An estimated proportion of p0 = 0.96 were not changed. The input variable was a transformed t-test statistic Z (in R code Z = qnorm(pt(tt, df = 39))).
Figure 10 suggests that rather few genes are changed. The methods give the following estimates of p0 : bootstrap 0.92, mgf 0.97, PRE 0.97, LSL 1.00, SEP 0.98, SPLOSH 0.88, BUM 1.00 and smoother 0.92, compare Figure 11. The estimates of FDR appear in Figures 12 and 13, and show some separation between methods. The trio BH, pava and qvalue largely agree. SEP LFDR and FDR stabilise at a value above 0.8 when the cut-off approaches zero. This is consistent: If the local FDR levels off, then the Averaging Theorem, see Methods, implies that so will the FDR. However, this level may seem incompatible with the assessment of SEP that 2% of the genes are truly changed. A rough calculation, replacing f by a histogram, would yield the estimate of LFDR(p) = p0/f(p) = 0.22 in a vicinity of p = 0.00025. Neither does it agree with locfdr, see Figure 14. Most likely the spline function employed by SEP is playing tricks here. SPLOSH approaches 0.4 in a vicinity of zero. BUM returns the estimate FDR ≡ 1, which is consistent with its estimate of p0, but does not seem to agree with Figure 11. The method may display numerical problems with these data. The function qvalue finds one gene significant with a q-value less than 5%. However, if we restrict attention to the probe sets with at least 20% present rate (9841 genes) there are fifteen genes fulfilling the criterion.
Figure 11 Spira et al data: Former smokers vs. Never smokers p-values. The binned densities in the histogram coupled with the inequality (9) points to a value of p0 of more than 0.9. The red and blue lines represent the maximum and the minimum estimate obtained from the methods under investigation.
Figure 12 FDR calculated from the Spira et al. data. The dotted line indicates where the optimal cut-off is estimated to be in the sense of minimising the sum of false positives and false negatives. At this point the FDR of BH, pava and qvalue is close to 0.4, and corresponds to the 298th smallest p-value. This high proportion of false positives may be acceptable given some high throughput validation procedures, but may be unacceptable in other contexts. Note that BUM produces the overly pessimistic estimate FDR ≡ 1, and hence appear at the upper limit of the graph. For comparison the local FDR [13] has been added to the graph.
Figure 13 Blow-up of region to the left of dotted line in Figure 12. Legend as in Figure 11A. SEP Local FDR and SEP FDR behave similarly in a vicinity of zero. These two methods indicate that the relative frequency of false positives stabilises to a value above 80% when the cut-off approaches 0. SPLOSH, on the other hand, approaches 0.4. Note that BUM does not appear in the graph.
Figure 14 displays the graphical output from locfdr, where the LFDR seems to approach zero in the tails, contradicting SEP, SPLOSH and BUM, but essentially agreeing with the concordant trio BH, pava and qvalue.
Discussion
Over-all results will lump together performance under different conditions and may thus be less relevant for a particular application. For instance, in practice the performance for high p0 will probably matter more than that for lower values. When many genes are changed the cutpoint will likely be chosen based on other criteria than FDR, and hence the difference between methods becomes less relevant. However, the detailed results presented here should give the practitioner some guidance as to what methods could be considered. Looking at the p-value histogram one can find some decision support in the choice of method. Comparing the output from several methods provides further clues.
All the methods performed worse on the dependent data; both the estimate of p0 and FDR suffered. To some extent that may be due to the lower mean power of the alternatives in that simulation model. However, the methods were derived under the assumption of independence and the small difference in mean power of 0.08 does not explain the great deterioration in most methods. Indeed, for simulated datasets with weak dependencies the results came close to the independent case, see the Additional file.
Through all tests PRE and pava FDR proved quite successful. Of the methods for estimating FDR, qvalue has the advantage of a well-documented and good track record, and behaves well here. BUM displays a varying performance, but does handle dependent data well. In practice it will be difficult to know where on the scale from independence to strong clumpy dependence a particular dataset will rate, if indeed it follows clumpy dependence at all. LSL, and BH, have some problems, but on the other hand they arise mainly at low p0, where they probably matter the least. As noted above regarding the Spira et al. data, SEP LFDR and FDR stabilised at a value above 0.8 when the cut-off approaches zero. In other tests SEP performed well, particularly with independent data. BUM in this case produced the estimate FDR ≡ 1 which can hardly reflect the truth.
The locfdr method offers the possibility to choose between three different null type distributions. The choice of the null type N(0,1) produced p0 estimates similar to those of the other methods. The need to specify the transform m may seem like an obstacle. But in many situations a parametric test statistic with a known null distribution exists. Alternatively, m could be identified by modelling a bootstrap distribution [25].
All the described methods assume the p-values were obtained in a reliable fashion, e.g. by a warranted normal approximation, a bootstrap or a permutation method. Reference [10] describes a case when a two-way ANOVA F-distribution was used when the distributional assumptions were not met. The estimate of p0 gave an unrealistic answer. When permutation p-values were used instead their method gave a more realistic result. One always has to bear this caveat in mind. To further complicate matters, the permutation of sample labels approach is no panacea if the independence between samples assumption does not hold true, as detailed in [27]. (Let us follow the usual convention that genes come in rows of the data matrix, and samples in columns.) Permuting within columns provided some remedy there. Misspecifying the null distribution will jeopardize any simultaneous inference, whether based on FDR or not. It may pay off to consider the correlation structure in data, both in view of this finding and in view of the different performance of methods depending on the strength of correlations.
The q-value q(pi) has been criticised for being too optimistic in that it weighs in also genes that are more extreme than i when calculating the measure. Note that a similar criticism could be levied against the classical p-value: the p-value gives the probability under the null hypothesis of observing a test statistic at least as extreme as the one observed. Also, there is no clear stable, reliable and tested alternative. This is not to say that the q-value is unproblematic, but it still has been studied and used much more than e.g. the local FDR, which may suffer from high random variation, see examples in [29]. Other examples from Results section give evidence of stability. Contrary to what one may anticipate the FDR is not always more stable than LFDR [25]. The concept of a local FDR seems quite interesting and may lead the way towards improved inference, and it begs a thorough investigation of the various recently published options.
To avoid pit-falls in the inference one must use the total information obtained from p0 and the FDR or q-value curve, see also Storey in the discussion of [31]. There is not one cut-off in terms of q-value that will suit all problems. Take the case of Figure 12, where one will have to accept a high FDR in order to find any DEGs. At the other end of the spectre, in Figure 7, the cut-off can be much more restrictive. The choice of cut-off must be made with a view to one's belief regarding p0, and calculating the sum of (4) and (5) to assess to total of false positives and false negatives gives further guidance in this choice. In general it makes sense to choose a cut-off in the region [0, αmin], where αmin is the value which minimises the total relative frequency of errors committed FP(α)+FN(α), see (4) and (5). However, since false positives and false negatives have different consequences with possibly different losses, it is difficult to state an algorithm that would cover all scenarios.
Conclusion
This article deals in the main with a simple frequentist framework for the analysis of microarray experiments. The conclusion is that the concept of the proportion of unchanged genes and the related concept of a q-value or false discovery rate are practical for such analysis. Furthermore, there exists open source code that implements methods that address the needs of the practitioner in this field. New methods gave evidence of improved performance, allowing better control of the error rate and thus enabling a more careful identification of DEGs. Issues still remain and improvements will probably appear over the next couple of years, but as a provisional solution these methods have much to offer.
Methods
The current article focuses on the two-component model. Other points of view exist. In reference [25] the two-component model is reshaped into a conceptually attractive one-group model allowing a continuum of effects.
Denote the pdf of p-values by f, the proportion of unchanged by p0 and the distribution of the p-values for the changed genes by f1. Then the pdf of p-values may be written as
f(x) = p0 × 1 + (1 - p0)f1(x) (8)
using the fact that p-values for the unchanged genes follow a uniform distribution over the interval [0,1]. This model is unidentifiable without further assumptions, e.g. that p-values in a vicinity of 1 only represent unchanged genes. From the non-negativity of pdf's, clearly
f(x) ≥ p0 (9)
This leads to the estimate based on the minimum of the estimated pdf [1]
see also Figure 6. In most cases the minimum in (10) will occur for some x close to or at 1. Hence (10) will in these cases agree well with an estimate of f(1). If one has reason to believe that p0 is close to 1, it may pay off to replace (10) by the 25% percentile or simply put the estimate equal to 1, in order to make overestimation more likely.
LSL
Let R(α) = # {i : pi ≤ α}, the number of rejected given the cut-off α. In [36] the approximation
N - R(α) ≈ E[N - R(α)] ≈ N0(1 - α)
for small α and N0 = Np0 the number of true null hypotheses appears. Consequently, (N - R(p(i))/(1 - p(i)) = (N - i)/(1 - p(i)) will approximate N0, which lead the pioneering authors to consider plotting 1 - p(i) against N - i, thus giving them an estimate of N0. In [26] the Lowest SLope estimator (LSL) of N0 based on the slopes Si = (1 - p(i))/(N - i + 1) is presented. Starting from i = 1, the procedure stops at the first occurrence of Si0 <Si0-1, and outputs the estimate
In [12] the two above estimates are presented, derived and compared together with a method called Mean of Differences Method (MD). MD and LSL are motivated by assuming independence and approximating the gaps d(i) = p(i) - p(i-1) (define p(0) = 0 and p(N+1)) ≡ 1) with a Beta(1, N0) distribution, which has expectation 1/(N0 + 1). This expectation may be estimated by the inverse of a mean of the form
MD proceeds downward and chooses i0 equal to the first j satisfying
Of these three methods LSL and MD give very similar results, and outperform their predecessor [12].
LSL is available as function fdr.estimate.eta0 in package GeneTS [18] with the option method= "adaptive".
The smoother
A method here referred to as the smoother appeared in [9]. This method, like all presented, is based on a comparison of the empirical p-value distribution to that of the uniform distribution. There will likely be fewer p-values close to 1 in the empirical than in the null distribution, which is a uniform. The ratio of the proportion of p-values greater than some η to the expected proportion under the uniform distribution, 1-η, will give a measure of the thinning of observed p-values compared to the null distribution. Thus, with Fe denoting the empirical distribution, the ratio {1-Fe(η)}/{1-η} will often be a good estimate of p0 for an astutely chosen threshold η. A spline is fitted to the function p0(η) = {1-Fe(η)}/{1-η}, and the resulting function is evaluated at η = 1, yielding the estimate
(tilde '~' above p0 denoting the spline-smoothed version of p0(η)) which goes into the calculation of the FDR in (2), see Figure 15. Note the relationship with LSL, p0(p(i)) = (N-i)/{(N-i+1)SiN}. It can be shown that for fixed η this method offers a conservative point estimate of the FDR:
Figure 15 The smoother estimate of p0. This is based on the ratio p0(η) = (1 - Fe(η))/(1 - η), the observed ratio proportion of p-values greater than η to the proportion expected from the uniform distribution. The data were simulated with p0 = 0.6. The red line represents the true value, while the blue gives a smooth curve representation of p0(η) approaching a limit close to the true value.
The q-value is estimated by combining (2), (7) and (11), and an implementation is provided as the function qvalue in package qvalue available on CRAN [18].
In [30] the authors go to great lengths to prove that for fixed η, as above, the conservativeness remains under various forms of dependency, such as clumpy dependence.
The Bootstrap LSE
Another approach pioneered by Storey in [3] is to use a bootstrap least squares estimate (LSE), which chooses a value of η in p0(η), that minimises the variation of the estimate for random samples of the original p-values. The bootstrap standard reference [37] provides more theoretical background. Generate B new samples p-values p*b (b = 1,..., B) by sampling with replacement from the observed ones, calculate a measure of the Mean Squared Error (MSE) of the corresponding estimates p*b0(η) for a lattice of values of η and choose the value minimising the MSE. More formally, the optimal η is obtained through
The version of the bootstrap used in this article uses more samples B than the version available in qvalue (B = 500 instead of B = 100), and seems to perform better (data not shown).
Available in functions qvalue [18] and p0.mom (in package SAGx) [18,38].
SPLOSH
In [11] a spline function estimates the log-transformed pdf log[f(x)] using a complex algorithm involving splines called spacings LOESS histogram (SPLOSH). To obtain a stable estimate of FDR near zero a technique from mathematical analysis called l'Hospital's rule is used to approximate the ratio in (5) and to yield
,
where the numerator has been estimated as in (4). An R package with the same name is available [39].
The FDR estimate (6) is used with F obtained by the non-parametric estimate of the pdf.
Note that we can now calculate the posterior probability given its p-value that a gene is a DEG as p1(x) = 1 - p0/f(x), compare (3).
The method is available in R function splosh [40].
BUM
In [10] the authors assume a beta-uniform (BUM) distribution, i.e. in (1) they replace f1 by a beta distribution,
f(x) = λ + (1 - λ)axa-1 (12)
where in addition to λ which corresponds to p0 the shape parameter a has to be estimated. Thanks to the simple form of the distribution it is possible to estimate parameters through the maximum likelihood principle, i.e. by choosing values that maximise the likelihood of observing the p-values that were actually observed. However, due to problem in identifying p0 with λ, the authors instead use an upper bound
which corresponds to f(1).
The FDR estimate (5) is used with F the cdf corresponding to (12).
The authors provide R code for the application of their method [39].
A more intricate Hierarchical Bayes model based on the beta-uniform concept allowing for different parameter values in different intervals appears in [41]. The R function localFDR provides an implementation of the method [42].
Poisson regression
In [28] it is suggested to estimate any empirical distribution by dividing the real axis into intervals and regarding the number of hits in each interval as the result of an inhomogeneous Poisson process, much like counting the number of cars arriving at a crossing during different time intervals. This method was used in [27] to model the distribution of a transformed test statistic, it also appears in function locfdr which estimates a local FDR as a function of a test statistic. In our case, of course, the support of the distribution is the unit interval [0,1]. Then the expected number of hits in each subinterval of [0,1] can be modelled as a polynomial in the midpoints of the subintervals by a technique called Poisson regression (PRE). The approach taken here is to choose a polynomial of low degree so that the plateau representing the uniform distribution is well captured. In doing so the ability the capture the distribution at low p-values is sacrificed.
A more mathematical description now follows. The PRE method assumes that the counts Sk follow a Poisson distribution whose intensity is determined by the midpoint tk of the interval Ik, see [28]. To be specific: in the current application it is assumed that the expected frequency of observations in an interval is given by
where μok are the smoothed observed frequencies in each interval Ik. In statistical jargon this is a Poisson regression model with μok as offset. This assumes independence between counts in different intervals. Although this does not hold true the model succeeds to capture the essential features of distributions. Standard functions in e.g. R can fit this model. Normalising the curve by the total number of p-values we get an estimate of the pdf. Finally, smooth the pdf f(x) with a spline to obtain a more stable result, and use the estimate (10). An implementation of PRE is provided through R function p0.mom in package SAGx [18,38].
SEP
The Successive Elimination Procedure (SEP) excludes and includes pi successively such that the similarity of the distribution of the included tends to behave increasingly like a uniform [13]. Finally, an index set Jfinal will map to a set of p-values that represent the true null hypotheses. This yields the point estimate
with NJ = # J, the cardinality of the set J. The identification of Jfinal proceeds by an intricate optimisation algorithm where the objective function consists of two terms : one Kolmogorov-Smirnov score
for the empirical cdf FJ (based on J), to measure the distance to a uniform, and one penalty term to guard against overfitting
for some tuning parameter λ.
A local FDR is obtained from smoothed histogram estimates based on equidistant bins
,
where the function h0 refers to the Jfinal set and h to the total set of p-values.
The function twilight in package twilight provides an implementation of SEP [19].
Moment generating function approach
The next approach is based on the moment generating function (mgf), which is a transform of a random distribution, which yields a function M(s) characteristic of the distribution, cf. Fourier or Laplace transforms, e.g. [43]. Knowing the transform means knowing the distribution. It is defined as the expectation (or the true mean) of the antilog transform of s times a random variable X, i.e. the expectation of esXor in mathematical notation:
M(s) = ∫esx f(x)dx.
To calculate the mgf for p-values, we use the fact that the pdf is a mixture of pdf's (8). This yields the weighted sum of two transformed distributions:
,
where we have used the fact that the mgf of a uniform distribution over [0,1] equals g(s) = (es - 1)/s. Denoting the second transform by M1(s) we finally have
M(s) = p0g(s) + (1 - p0)M1(s). (13)
Now, the idea is to estimate these mgf's and to solve for p0. In the above equation M(s) can be estimated based on an observed vector of p-values and g(s) can be calculated exactly, respectively, while p0 and M1(s) cannot be estimated independently. The estimable transform is, given the observed p-values p = p1,..., pn, estimated by
Then, one can solve (13) for p0:
Let us do so for sn > sn-1, equate the two ratios defined by the right hand side in (14) and solve for M1(sn). This gives the recursion
If we can find a suitable start for this recursion we should be in a position to approximate the increasing function M1(s) for s = s1 <s2 < ... <sm in (0, 1]. Now, note that 1 ≤ M(s), for any mgf, with close to equality for small values of s. It makes sense to start the recursion with some M1(s1) in I = [1, M(s1)]. In general, it will hold true that 1 ≤ M1(s) <M(s) <g(s), since f1 puts weight to the lower range of the p-values at the expense of the higher range, the uniform puts equal weight, and f being a mixture lies somewhere in between. We can calculate g, M and M1 for an increasing series of values in [0,1], e.g. for s = (0.01, 0.0101, 0.0102, ..., 1). The output from one data set appears in Figure 16. Since all ratios (14) should be equal, a good choice of M1(s1) will be one that minimises the variation of the ratios. Standard one-dimensional optimisation will find the value in I that minimises the coefficient of variation (CV, standard deviation divided by mean)
Figure 16 Estimated moment generating functions (mgf's). Given an observed vector of p-values it is possible to calculate mgf's for the observed distribution f (M) and the unobserved distribution f1 (M1), and without any observations we can calculate the mgf for the uniform (g).
where s = (s1, ..., sn). The CV will in contrast to the variance put both small and high values of the ratios on an equal footing and enable comparison.
Finally, these essentially equal ratios provide an estimate of p0.
A heuristic convexity argument suggests that mgf over-estimates p0, see the Additional file. Furthermore, the bias seems to decrease as p0 grows.
An implementation of mgf appears as function p0.mom in package SAGx [18,38].
Local FDR and FDR
The concept of a local false discovery rate originates from [1]. Let the (true) local FDR at t be defined as the probability that gene i is unchanged conditional upon that its p-value equals t, or in formulas : LFDR(t) = Pr(gene i unchanged | pi = t) = p0/f(t). The Averaging Theorem of [4] states that integrating the local FDR over the rejection region R, such as R = [0, 0.01], yields the FDR : FDR(R) = E[LFDR(y) | y ∈ R]. In [29] it is noted that the estimated q-value equals the mean of a local FDR
where the local FDR at the ith ordered p-value p(i) equals
,
where N denotes the total number of genes. This rephrases the theorem in terms of estimators. The local FDR is meant as an approximation of the probability that an individual gene i is a DEG. As remarked in [29] the q-value does not estimate the probability that a gene is a false positive. Indeed, the theorem shows that it is the mean of that probability for all genes at least as extreme as i. Thus the q-value will tend to give a lower value than LFDR(i).
Under a wide range of models, where f(x) is non-increasing, e.g. the BUM model, the expected local LFDR(i) will be non-increasing, and hence the differences above should tend to increase, see the Additional file. Hence there is a need for enforcing monotonicity as in (7). One tool for enforcing monotonicity is the Pooling of Adjacent Violators (PAVA) algorithm [44]. This algorithm has an intuitive appeal, is less ad-hoc than the local spline approach presented in [29], and is the non-parametric maximum likelihood estimate under the assumption of monotonicity. As an example of how it works, consider the series (1,2,4,3,5), which PAVA turns into the non-decreasing series (1, 2, 3.5, 3.5, 5) by pooling the violators of monotonicity (4, 3) and replacing them by their mean. Though not equivalent to the q-value from (2) and (7), the results from applying PAVA to the terms in (15) agreed rather well with the values obtained from function qvalue. In the Results section this approach combined with the PRE estimate of p0 is referred to as pava FDR. We could have used mgf for calculating FDR, but it was excluded due to the better over-all performance of PRE.
The bootstrap LSE gives a very similar result to the smoother and thus was excluded in comparison of FDR estimates.
The R function pava.fdr in package SAGx provides an implementation of pava FDR, and returns a list of estimates of FDR, LFDR and p0 [18,38].
The reference [25] presents the theory behind the estimation of local false discovery rates provided by R package locfdr [18,25]. The method procedes by transforming the test statistic t into Z = Φ-1(m(t)), where Φ is the cdf corresponding to N(0,1) and m is a transform that for the uninteresting/null class renders the distribution of Z into a N(0,1). Typically, m could equal the cdf of the t-distribution. Assuming the model (1) the method estimates the mixture density ft using Poisson regression, and fits either a theoretical null sub density (from now on suppressing superscript t and denoting the pdf corresponding to Φ by φ) f+0(z) = p0 φ(z) around z = 0, or fits an empirical null distribution. Then the procedure estimates the local false discovery rate fdr(z) = f+0(z)/f(z).
Figure 3 Mean squared difference between estimated and true FDR for the simulated independent data. Note that qvalue and pava FDR stand out for having the smallest deviation from the true FDR.
Supplementary Material
Additional File 1
contains further results concerning weakly dependent data and smaller datasets, as well as some mathematical details supporting points made in the article.
Click here for file
Acknowledgements
Thanks are due to Niclas Sjögren at AstraZeneca R&D Södertälje for valuable comments. Furthermore, suggestions made by the three reviewers improved both content and presentation.
==== Refs
Efron B Tibshirani R Storey JD Tusher VG Empirical Bayes analysis of a microarray experiment Journal of the American Statistical Association 2001 96 1151 1160 10.1198/016214501753382129
Benjamini Y Hochberg Y Controlling the false discovery rate : a practical and powerful approach to multiple testing Journal of the Royal Statistical Society B 1995 57 289 300
Storey JD A Direct Approach to False Discovery Rates Journal of the Royal Statistical Society B 2001 64 479 498 10.1111/1467-9868.00346
Efron B Storey JD Tibshirani R Microarrays, Empirical Bayes Methods, and False Discovery Rates Technical report 2001 217 Stanford, Department of Statistics, Stanford University
Broberg P Statistical methods for ranking differentially expressed genes Genome Biology 2003 4 R41 12801415 10.1186/gb-2003-4-6-r41
Delongchamp RR Bowyer JF Chen JJ Kodell RL Multiple-testing strategy for analyzing cDNA array data on gene expression Biometrics 2004 60 774 782 15339301 10.1111/j.0006-341X.2004.00228.x
Tusher VG Tibshirani R Chu G Significance analysis of microarrays applied to the ionizing radiation response PNAS 2001 98 5116 5121 11309499 10.1073/pnas.091062498
Allison DB Gadbury GL Moonseong H Fernandez JR Cheol-Koo L Prolla TA Weindruch RA A mixture model approach for the analysis of microarray gene expression data Computational Statistics and Data Analysis 2002 39 1 20 10.1016/S0167-9473(01)00046-9
Storey JD Tibshirani R Statistical significance for genomewide studies PNAS 2003 100 9440 9445 12883005 10.1073/pnas.1530509100
Pounds S Morris SW Estimating the occurrence of false positives and false negatives in microarray studies by approximating and partitioning the empirical distribution of p-values Bioinformatics 2003 19 1236 1242 12835267 10.1093/bioinformatics/btg148
Pounds S Cheng C Improving false discovery rate estimation Bioinformatics 2004 20 1737 1745 14988112 10.1093/bioinformatics/bth160
Hsueh H Chen JJ Kodell RL Comparison of Methods for Estimating the Number of True Null Hypotheses in Multiplicity Testing Journal of Biopharmaceutical Statistics 2003 13 675 689 14584715 10.1081/BIP-120024202
Scheid S Spang R A Stochastic Downhill Search Algorithm for Estimating the Local False Discovery Rate IEEE Transactions on Computational Biology and Bioinformatics 2004 1 98 108 17048385 10.1109/TCBB.2004.24
Scheid S Spang R Hornik K, Leisch F and Zeilis A A False Discovery Approach to Separate the Score Distributions of Induced and Non-induced Genes Proceedings of the 3rd International Workshop on Distributed Statistical Computing (DSC 2003) 2003 Vienna, Austria, Technische Universität Wien. Institut für Statistik
Lehmann EL Testing Statistical Hypotheses 1953 New York, Wiley
R Foundation for Statistical Computing R Development Core Team. R: A language and environment for statistical computing
Ihaka R Gentleman R R: A language for data analysis and graphics Journal of Computational and Graphical Statistics 1996 5 299 314
The Comprehensive R Archive Network
Bioconductor
Yang YH Speed T Speed T Design and analysis of comparative microarray experiments Statistical Analysis of Gene Expression Microarray Data 2003 London, Chapman & Hall/CRC
Simes R An improved Bonferroni Procedure for Multiple Tests of Significance Biometrika 1986 73 751 754
Reiner A Yekutieli D Benjamini Y Identifying differentially expressed genes using false discovery rate controlling procedures Bioinformatics 2003 19 368 375 12584122 10.1093/bioinformatics/btf877
Tsai CA Hsueh H Chen JJ Estimation of False Discovery Rates in Multiple Testing: Application to Gene Microarray Data Biometrics 2003 59 1071 1081 14969487 10.1111/j.0006-341X.2003.00123.x
Yang MCK Yang JJ McIndoe RA She JX Microarray experimental design: power and sample size considerations Physiol Genomics 2003 16 24 28 14532333 10.1152/physiolgenomics.00037.2003
Efron B Selection and Estimation for Large-Scale Simultaneous Inference Technical report 2004 , Department of Statistics, Stanford University
Benjamini Y Hochberg Y On the adaptive control of the false discovery rate in multiple testing with independent statistics Journal of Educational and Behavioural Statistics 2000 25 60 83
Efron B Large-scale simultaneous hypothesis testing : the choice of a null hypothesis Journal of the American Statistical Association 2004 99 96 104
Efron B Tibshirani R Using specially designed exponential families for density estimation The Annals of Statistics 1996 24 2431 2461 10.1214/aos/1032181161
Aubert J Bar-Hen A Daudin JJ Robin S Determination of the differentially expressed genes in microarray experiments using local FDR BMC Bioinformatics 2004 5 125 15350197 10.1186/1471-2105-5-125
Storey JD Tibshirani R Estimating false discovery rates under dependence, with applications to DNA microarrays Technical Report 2001-28, Department of Statistics, Stanford University 2001
Ge YC Dudoit S Speed TP Resampling based multiple testing for microarray data analysis Test 2003 12 1 77
Golub TR Slonim DK Tamayo P Huard C Gaasenbeek M Mesirov JP Coller H Loh ML Downing JR Caligiuri MA Bloomfield CD Lander ES Molecular Classification of Cancer: Class Discovery and Class Prediction by Gene Expression Monitoring Science 1999 286 531 537 10521349 10.1126/science.286.5439.531
Affymetrix web site
Spira A Beane J Shah V Liu G Schembri F Yang X Palma J Brody JS Effects of cigarette smoke on the human airway epithelial cell transcriptome PNAS 2004 101 10143 10148 15210990 10.1073/pnas.0401422101
Gene Expression Omnibus
Schweder T Spjotvoll E Plots of p-values to evaluate many tests simultaneously Biometrika 1982 69 493 502
Efron B Tibshirani R An Introduction to the Bootstrap 1993 London, Chapman & Hall
The SAG homepage
BUM software
SPLOSH web page
Liao JG Lin Y Selvanayagam ZE Shih WJ A mixture model for estimating the local false discovery rate in DNA microarray analysis Bioinformatics 2004 20 2694 2701 15145810 10.1093/bioinformatics/bth310
localFDR web site
Feller W An Introduction to Probability Theory and Its Applications 1971 2 Second New York, John Wiley
Ayer M Brunk HD Ewing GM Reid WT Silverman E An empirical distribution function for sampling with incomplete information Annals of Mathamatical Statistics 1955 26 641 647
|
16086831
|
PMC1199583
|
CC BY
|
2021-01-04 16:27:26
|
no
|
BMC Bioinformatics. 2005 Aug 8; 6:199
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-199
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2011609114710.1186/1471-2105-6-201SoftwareDynGO: a tool for visualizing and mining of Gene Ontology and its associations Liu Hongfang [email protected] Zhang-Zhi [email protected] Cathy H [email protected] Department of Information Systems, University of Maryland, Baltimore County, 1000 Hilltop Circle, MD 21050, USA2 Department of Biochemistry and Molecular Biology, Georgetown University Medical Center2, 3900 Reservoir Road, NW, Washington, DC 20057, USA2005 9 8 2005 6 201 201 15 12 2004 9 8 2005 Copyright © 2005 Liu et al; licensee BioMed Central Ltd.2005Liu et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
A large volume of data and information about genes and gene products has been stored in various molecular biology databases. A major challenge for knowledge discovery using these databases is to identify related genes and gene products in disparate databases. The development of Gene Ontology (GO) as a common vocabulary for annotation allows integrated queries across multiple databases and identification of semantically related genes and gene products (i.e., genes and gene products that have similar GO annotations). Meanwhile, dozens of tools have been developed for browsing, mining or editing GO terms, their hierarchical relationships, or their "associated" genes and gene products (i.e., genes and gene products annotated with GO terms). Tools that allow users to directly search and inspect relations among all GO terms and their associated genes and gene products from multiple databases are needed.
Results
We present a standalone package called DynGO, which provides several advanced functionalities in addition to the standard browsing capability of the official GO browsing tool (AmiGO). DynGO allows users to conduct batch retrieval of GO annotations for a list of genes and gene products, and semantic retrieval of genes and gene products sharing similar GO annotations. The result are shown in an association tree organized according to GO hierarchies and supported with many dynamic display options such as sorting tree nodes or changing orientation of the tree. For GO curators and frequent GO users, DynGO provides fast and convenient access to GO annotation data. DynGO is generally applicable to any data set where the records are annotated with GO terms, as illustrated by two examples.
Conclusion
We have presented a standalone package DynGO that provides functionalities to search and browse GO and its association databases as well as several additional functions such as batch retrieval and semantic retrieval. The complete documentation and software are freely available for download from the website .
==== Body
Background
A large amount of data and information about genes and gene products has been stored in various molecular biology databases. To assist with the representation and integration of knowledge about genes and gene products, biological ontologies and tools have been developed. One such ontology is the Gene Ontology (GO) [1], which has become a major vocabulary for annotating genes and gene products in various databases, such as MGI (Mouse Genetic Informatics) [2], SGD (Saccharomyces Genome Database) [3], and RGD (Rat Genome Database) [4], as well as in protein databases such as UniProt (Universal Protein Resource) [5], and InterPro [6]. Genes and gene products from these databases are annotated with GO terms, and GO annotations of genes or gene products are distributed through the GO website as "association files." The databases are referred to as "annotation databases" hereafter.
Annotating genes and gene products is laborious and requires biological expertise – curators need to understand the precise definition and usage of terms as well as their hierarchies in GO for accurate and consistent curation. A browsing tool with the functionality of retrieving descendants and ancestors of a specific GO term will help curators to select the most appropriate GO terms for annotating genes and gene products. Additionally, a tool that shows GO annotations for a list of specified genes and gene products may help the identification of common molecular functions, cellular components or biological processes among these genes and gene products.
Furthermore, the main purpose of having GO is to allow integrated queries across multiple databases. The retrieval of genes and gene products across multiple databases that are semantically related to a query gene or gene product (i.e., genes and gene products that share similar GO annotations with the query gene or gene product) allows users to investigate functional relationships among them. For example, the yeast YAP1 protein (YAP1_YEAST, 650aa) and the human Jun-B protein (JUNB_HUMAN, 347aa) share little sequence similarity (30% sequence identity in 53aa overlapped region) (Figure 1a). However, their functional relationship is revealed by their associated GO annotations, which indicate that both are involved in transcription (GO:0006350), DNA binding (GO:0003677) and transcription regulator activity (GO:0030528). Note that these GO annotations are based on literature citations with the TAS (Traceable Author Statement) evidence code. Such investigation may help biologists to generate new hypotheses.
Figure 1 Detection of functional homologs using GO annotations. (a) Sequence alignment based on Smith-Waterman algorithm for YAP1 and JUNB proteins; (b) GO annotations of these two proteins, where the lowest common GO terms in the hierarchies are highlighted based on TAS (Traceable Author Statement) leaf nodes.
At the same time, with more genes and gene products having been annotated, many applications have been developed to utilize GO annotations for knowledge discovery. Already over a dozen tools can be found from the GO website for browsing GO terms and GO annotations, and for microarray data analysis. While many tools are available for GO browsing, few allow users to dynamically interact with the retrieved results without losing previous pages; in most web-based tools, activating a link may replace or hide previous pages. A tool that can display GO browsing results in one single window would be helpful to investigate relevant information.
We have developed a tool, DynGO (Dynamic GO), that allows users to dynamically interact with GO annotation data, and to browse all GO terms and annotations. Besides the standard search capability of AmiGO – the "official" GO interface, DynGO provides advanced functionalities such as retrieving and presenting GO annotations based on a list of genes and gene products, or a list of genes and gene products sharing similar GO annotations.
In the following, we describe the graphical user interface (GUI) of DynGO that supports dynamic functionalities of DynGO. Also, we present the semantic similarity of GO, which is used by DynGO for semantic retrieval.
DynTreeViewer
The GUI of DynGO was adapted from an existing GUI package, DynTreeViewer [7,8]. The viewer, written in JAVA, provides various dynamic tree views for displaying hierarchical data. The program provides functionalities that allow data to be filtered, sorted, deleted, and lifted.
The tree displayed by the DynTreeViewer provides frequency information (or other critical statistical information) for each tree node by combining the frequency counts of all of its child nodes. Various applications using the DynTreeViewer have been developed including a vocabulary development tool based on medical reports [7,8] and a summarization tool for multiple medical reports [9]. Major functionalities of DynTreeViewer that were incorporated into DynGO include tree lifting and tree sorting.
Tree lifting allows the user to transform the tree and view the data in different orientations dynamically. An example of tree lifting of GO association data in XML format is shown in Figure 2. The data can be displayed based on RGD with GO annotations after lifting the RGD genes and gene products (RGD-centric; Figure 2a) or based on GO hierarchy with the annotation of RGD genes and gene products (GO-centric; Figure 2b). Tree lifting enables users to view hierarchical data through different orientations so that people can switch dynamically from viewing genes and gene products annotated with the same GO term to viewing all GO terms that are associated with the same gene and gene product.
Figure 2 An example of tree lifting. (a) RGD-centric view; (b) GO-centric view.
Tree sorting allows the user to view a sub-tree sorted according to the alphabetical (or reversed alphabetical) order, or sorted numerically according to frequency or other statistical information. The sorting functionality allows users to locate nodes easily in a tree.
GO and semantic similarity of GO terms
GO comprises three hierarchies that hold terms defining the basic concepts of molecular function, biological process, and cellular component, respectively. The terms, organized using a Directed Acyclic Graph (DAG), may have one or more parents with a relation type of either "is-a" or "part-of." GO terms are used to annotate entities in dozens of genomic and protein databases. The development of GO and its wide adoption for genome annotation makes it feasible to perform semantic retrieval from multiple databases (i.e., to retrieve entities with similar GO annotations as the query entity).
As with any other similarity search task, semantic retrieval requires the definition of a similarity measure for entities sharing similar GO annotations. In DynGO, we adapted one of the semantic similarity measures reported by Lord et al. [10], which is based on the notion of "information content" – i.e., a term in an ontology is more informative if the term and its descendants have fewer annotated genes or proteins. The GO annotation measure starts with a probability measure of each term t. Let Dt be the collection of GO terms that are either t or its descendants Let A(t, c) be the occurrence of t annotations given a collection c. The probability of t in c, or p(t, c), is defined as:
Let CA(t1, t2) be the lowest common ancestor set for terms t1 and t2, since GO allows multiple parents for each term. The semantic similarity of two GO terms is defined as:
The similarity of two genes or gene products is then defined as the average similarity between GO annotations for them.
Implementation
DynGO was designed as a server-client application. As depicted in Figure 3, DynGO contains three functional components: Preprocessor and GOEngine running on the server side [see Additional file 1], and GOGUI on the client side [see Additional file 2]. The Preprocessor uses the GO distribution file as input and generates several database tables, while the GOEngine accepts queries from the client and generates trees for displaying in GOGUI.
Figure 3 The overall server-client architecture of DynGO.
Server – preprocessor and GOEngine
Both the preprocessor and GOEngine were coded using PERL, an open source programming language. The Preprocessor dynamically downloads the GO distribution file go_YYYYMM-assocdb.rdf-xml.gz (YYYY stands for year and MM stands for month) from the GO ftp site and generates tables that are stored in BerkeleyDB, an open source database management system. Information stored in the database includes hierarchical relations, GO terms and attributes (e.g., names, synonyms, references, and definitions), and GO annotations and attributes (e.g., names and references) of genes and gene products. Some intermediate tables for semantic retrieval are also generated and stored, including the probability table of terms in collections shown in Formula 1 and the semantic similarity table storing the similarity of any pair of GO terms computed according to Formula 2. Note that the collection used to derive the probability of GO terms consists of all entries from annotation databases in the GO distribution file.
The GOEngine processes queries and generates trees. Two types of trees can be generated depending on the nature of the queries: a GO tree or an association tree. The GO tree arranges GO terms according to the GO hierarchies. The association tree arranges genes and gene products as leaf nodes of associated GO terms according to the GO hierarchies. In the current implementation, the GOEngine supports seven types of queries as listed below, with the function name followed by its argument(s) in parentheses.
• Generate_GO () generates a GO tree consisting of all GO terms where a term may appear in multiple branches of the tree.
• Generate_Assoc (AssocDB) creates an association tree for all genes and gene products in a given association database, AssocDB (e.g., MGI). GO terms not associated with genes and gene products in AssocDB are not shown in the tree. A gene entity may appear in multiple branches of the tree if it is annotated by multiple GO terms.
• Retrieve_Genes (GeneIDs) creates an association tree for all genes and gene products from a list of GeneIDs in one or more annotation databases, where each GeneID is the unique identifier used in the corresponding association database (e.g., MGI:108111).
• Retrieve_Relatives (GeneID, AssocDBs, Parameters) retrieves genes and gene products that are "relatives" of the GeneID based on similar GO annotations from a list of annotation databases (AssocDBs), and displays the related genes and gene products in an association tree. Several parameters are needed for the function: parameters that assign weight to each hierarchy (the default values are 0.4 for molecular function, 0.4 for biological process, and 0.2 for cellular components) and the similarity threshold value (the default value is 0.5).
• Retrieve_Genes (GoTerm, AssocDBs) returns genes and gene products from a list of annotation databases (AssocDBs) for a query GO term (GoTerm), and displays the entities in an association tree.
• Retrieve_Descendants (GoTerm) generates a GO tree consisting of all ancestors and descendants of a given GoTerm.
• Retrieve_Search (QueryString, AssocDBs) searches IDs, terms, and attributes of all GO terms or genes and gene products in AssocDBs using a query string that can be a word or any identifier (such as GO identifiers or gene entity references). All matches found from GO or AssocDBs are displayed in a GO tree or an association tree.
Client – GOGUI
The JAVA-based client interacts with users using menus, mouse clicks, or user input dialogs. The primary interface of GOGUI is a four-panel window in which the user can inspect the GO hierarchies and GO annotations. Figure 4a shows a screenshot after loading the GO tree generated by the function Generate_GO. The left-top panel of the window (InputPanel) handles user queries to GO and GO annotation databases. The right-top panel (TreeHolder) displays trees. The tree shown in Figure 4a has been sorted using probability information. The left-bottom panel (ReferencePanel) lists all database cross-references for the selected tree node. The right-bottom panel (WebPanel) displays the website of the selected tree node or the selected references. The example in Figure 4a shows that, after choosing the GO node "GO:0015075 ion transporter activity," the reference list for GO:0015075 was displayed in the ReferencePanel; and that the selection of "InterPro IPR004749" from the reference list returned the website of the InterPro reference in the WebPanel.
Figure 4 Screen shots of DynGO. (a) GO tree obtained from all GO terms; (b) Association tree obtained by retrieving products for GO:0015075.
The TreeHolder can hold multiple trees, where each tree is displayed using DynTreeViewer. Figure 4b displays two trees, the GO tree and an association tree. The latter is obtained by the function Retrieve_Genes, where genes and gene products for GO term ion transporter activity were retrieved from three associated databases MGI, UniProt, and SGD. Other panels in GOGUI are also dynamically changed to indicate the current tree. The MGI website of a selected gene "MGI:108111" is shown at the WebPanel. Links shown in the website can also be activated.
Figures 5 and 6 illustrate the flexibility of various tree display options in GOGUI and the functionality of semantic retrieval. Figure 5 shows three orientations of an association tree for RGD generated by the function Generate_Assoc (RGD): the default association tree which arranges RGD genes and gene products as leaf nodes of GO terms, where users can easily identify entities associated with a specific GO term (Figure 5a); the "lifted" tree arranged based on RGD entities, where users can view GO annotations for a specific gene or gene product (Figure 5b); and the tree that arranges annotations according to their evidence codes (Figure 5c). Figure 6 shows the functionality of semantic retrieval using three association trees: the default association tree that displays the GO annotations for SGD generated by the function Generate_Assoc (SGD) (Figure 6a); the tree that shows the annotations for one gene, SDS24 (Figure 6b); and the tree that displays the semantically related genes and gene products for gene SDS24 sorted according to the semantic similarity (Figure 6c). The latter was obtained dynamically using the function Retrieve_Relatives with default parameters to identify entities with similar GO annotations to those of SDS24 from RGD. As indicated in Figure 6c, the closest related gene in RGD to yeast gene SDS24 in SGD is the rat gene Itsn. By overlaying Figures 6b and 6c, which show the detailed annotations for Itsn and SDS24, respectively, users can compare their GO annotations – for example, both genes are associated with GO:0006810 transport.
Figure 5 Tree lifting function to dynamically obtain trees with different orientations. (a) Association tree for RGD; (b) View obtained by lifting reference identifiers; (c) View obtained by lifting evidence codes.
Figure 6 Screen shot for exploring association trees. (a) Association tree for all genes in SGD; (b) Relatives of gene SDS24; (c) Annotations for SDS24.
System testing and example applications
We conducted an integrated system testing and analysis using the November 1, 2004 GO distribution file go_200411-assocdb.xml.gz [11]. There were 18,017 GO terms in this distribution, which were used to annotate 1,004,671 genes and gene products from 17 different annotation databases. It took about 5 hours for the Preprocessor to prepare all database tables on a PC laptop running Window XP with a 1.6 GHz Intel processor and 512 MB of RAM. One of the most computing-intensive tasks is the construction of the similarity table, where the similarity of every pair of GO terms is computed and stored. Note that semantic retrieval operation is a kind of time-consuming when the number of records in the association database is over ten thousands. For users who do not have plans to use the semantic retrieval function of DynGO, a light version of the Preprocessor is available that can prepare tables much faster (i.e., in less than 10 minutes using the same laptop).
The basic searching and retrieving functionality of DynGO was tested by four users (bioinformatics researchers and PhD students). They were pleased that DynGO could instantly retrieve the GO hierarchy for a specific GO term and retrieve a GO annotation tree for a specific gene or gene product. It was commented that DynGO is easy to move around to gather information about GO terms as well as GO annotations for a specific gene or gene product since multiple views were held in one window. Some indicated that semantic retrieval was slow (longer than 5 minutes) when they used annotation databases with more than thousands of records.
Besides searching and browsing GO terms and GO annotation databases, the advanced functionalities of DynGO allow it to be used for applications where records are annotated using GO. Here, we illustrate through two applications: one is to visualize GO annotations of informative probe sets for microarray data analysis [12] and the other is in the study of complementing GO with PIRSF classification-based protein ontology [13].
The example data set we used in visualizing GO annotations of informative probe sets is the Head and Neck Squamous Cell Carcinoma (HNSCC) data set, in which differentially expressed genes in head and neck cancer were examined by Kuriakose et al [12]. In the study, RNA extracted from 22 paired samples of HNSCC and normal tissue from the same donors was hybridized to the Affymetrix U95A chip. Forty-two differentially expressed probe sets were selected for further validation by hierarchical clustering, multiple probe-set concordance, target-subunit agreement, and real-time PCR analysis. We downloaded the GO annotations for the probe sets of the U95A chip from the TIGR Resource website [14] and treated them as an association database. In DynGO, we used the function Retrieve_Genes (GeneIDs), where elements in GeneIDs are annotated probe sets. Users could then investigate the GO annotations in multiple orientations, inspect different statistical measures such as probability of a GO term in the set, retrieve genes that hold similar annotations from other annotation databases, among other functions (Figure 7).
Figure 7 DynGO for the visualization of annotated probe sets for microarray data analysis.
The study of GO and PIRSF [15] mapping shows that PIRSF classification-based protein ontology can complement GO concepts by identifying missing GO branches/nodes and linking GO terms among the three sub-ontologies (i.e. molecular function, biological process, and cellular component) (Figure 8). In the study, we mapped PIRSF protein families to the GO hierarchy and used DynGO to superimposing the two classification hierarchies with a bidirectional display showing either a GO-centric (Figure 8a) or a PIRSF-centric (Figure 8b) view to explore GO and PIRSF relationships.
Figure 8 DynGO for the study of complementing GO with the PIRSF protein classification system. (a) Identification of missing GO nodes according to PIRSF families; (b) Linkage of GO sub-ontologies based on PIRSF GO associations.
We found that the majority of curated PIRSF families map to GO leaf nodes, many of which also share common GO leaf nodes. The PIRSF associations to GO nodes allow us to examine whether certain GO subtrees might need expansion if GO concepts are too broad and to identify missing GO nodes when entire groups of superfamilies cannot be mapped to existing GO terms. As an example, Figure 8a shows that nested under receptor activity (GO:0004872) is a leaf node estrogen receptor activity (GO:0030284), under which we can identify two distinctly different concepts estrogen receptor α and β (SF500101, SF500102) that are not represented in GO. Also identified are activities of several groups of gene products missed in GO: progesterone receptor (SF002528) of steroid hormone receptor activity (GO:0003707), and retinoid acid receptor RAR (SF036753) of retinoid acid receptor activity (GO:0003708). Furthermore, directly under receptor activity there is no concept for orphan nuclear receptor (ligand-independent nuclear receptor) activity, which encompasses a large group of gene products such as TR2 (SF002526) and SF-1 (SF002530).
PIRSF classification can also provide links between the three GO sub-ontologies, each of which presently has its own hierarchical organization with no relationships inter-connecting them. Figure 8b shows the PIRSF-centric view where estrogen receptor α (PIRSF500101) is annotated with GO terms in both molecular function and biological process, which connects the receptor binding (GO:0005102) of signal transducer activity (GO:0004871) with the steroid hormone receptor signaling pathway (GO:0030518) of signal transduction (GO:0007165).
Discussion
In this paper, we adapted one of the similarity measures used by Lord et al [10] for semantic retrieval. We plan to investigate other similarity measures for GO terms (e.g., the tree similarity measure used by Wang and Zhao [16]) and incorporate them into DynGO. Note that we include the evidence code when displaying the trees. Researchers that wish to filter out the annotations whose evidence codes are IEA can lift the tree according to the evidence code so that IEA annotations can be deleted from the tree.
Many software tools have been developed for GO, most of which are web-based browsing systems. Some of these web-based systems do not provide tree navigation where all GO terms are loaded; instead, the browsing is limited only to queries of a single GO term as in GenNav [17]. Other systems provide navigation function for all GO terms but do not accept queries, such as the GO Browser developed at NCI's Cancer Genome Anatomy Project. A few tools do accept queries and have full tree navigation, such as the MGI GO Browser [18], but the results are shown in flat-tables so that users cannot directly inspect relations among multiple GO terms or genes and gene products. Additionally, all such web-based systems depend on Internet connection and may experience slow response time.
There is one standalone tool that is similar to DynGO, the Berkeley Drosophila Genome Project (BDGP) DAG-EDIT [1]. Both tools are JAVA-based and display search results in tree views. DAG-EDIT, however, is designed as an editor for GO term editing; therefore, it does not provide the functionality for browsing genes and gene products in association trees and for dynamically linking to the websites. DynGO, on the other hand, is for querying, visualizing, and mining GO terms as well as GO annotations from annotation databases. DynGO enables GO curators and users to load multiple trees, and to dynamically obtain views in different orientations for the resulting trees.
Note that few of the existing tools support semantic retrieval (i.e., to retrieve genes and gene products sharing similar annotations), batch retrieval (i.e., to retrieve GO annotations for a list of entities), or allow users to dynamically display trees with different orientations at the same time. These features distinguish DynGO from other tools.
Several GO mining tools are available. These include GoMiner [19], MAPPFinder [20], FatiGO [21], and GoSurfer [22]. Zeeberg et al. [19] did an extensive comparison of GoMiner with some of these mining tools. Here, we concentrate on comparing GoMiner and DynGO. GoMiner is a program package that organizes a list of interesting genes for biological interpretation using GO. It includes statistical analysis and two visualizations: a hierarchical tree view and a DAG view. While DynGO does not perform advanced statistical analysis, it can display the GO annotation trees for a gene list identified and exported from other software. GO terms for clusters of differentially expressed genes can be easily detected using the statistical information provided in the tree, either from the leaf descendant statistics or from the microarray data analysis software. Instead of providing two views as in GOMiner, we present the result in one tree that can be manipulated dynamically to change the orientation or to display information such as definitions or associated websites of a node. DynGO is flexible in the sense that it can be coupled with any microarray data analysis software to display the GO annotation for informative genes. Meanwhile, it combines well with browsing functionality and allows users to view the overall GO tree structure.
In the future, we will extend DynGO to incorporate microarray data analysis software on the server side to allow advanced microarray data analysis. Additionally, we plan to improve DynGO by collecting comments and feedback from the GO community. A formal usability evaluation is also planned.
Conclusion
We have presented a standalone package DynGO that provides functionalities to search and browse GO and its association databases as well as several additional functions such as batch retrieval and semantic retrieval. It enables users to browse the whole GO as well as GO annotations for many genomics and protein databases. It also includes statistical measures such as the number of leaf nodes and the probability of a specific GO term being assigned for a given collection. The results are displayed as trees that can be sorted; or users can dynamically obtain views with different orientations. For GO curators, curators of genomic and protein annotation databases, and users who use GO frequently, DynGO provides fast and convenient access to the GO and gene entity data since the data are located locally (the same computer or a computer on a local intranet) and the DynGO interface holds all results inside one browsing window. DynGO can also be extended to view any other data sets where the GO annotations are available.
Availability and requirements
• Project name: DynGO
• Project homepage:
• Operating systems: platform independent, tested on Window XP system and Red Hat Linux System
• Programming language: Perl, Java
• Other requirements: Java version 1.4, Perl version 5.8, Berkeley DB, Perl model BerkeleyDB
• License: GNU GPL
• Any restriction to use by non-academics: license needed.
Documentation, the source code of the server, and the executable of the client are available in the project website [also see Additional files]. Other used software components are available at the according sources.
Supplementary Material
Additional file 1
client_download.zip The executable jar file for client
Click here for file
Additional file 2
server_download.zip The source code for server
Click here for file
Additional file 3
dyngo.html The documentation for DynGO
Click here for file
Additional file 4
dingo-download.html The tutorial for downloading DynGO
Click here for file
Acknowledgements
The project is supported by grant IIS-0430743 from the National Science Foundation.
==== Refs
Harris MA Clark J Ireland A Lomax J Ashburner M Foulger R Eilbeck K Lewis S Marshall B Mungall C Richter J Rubin GM Blake JA Bult C Dolan M Drabkin H Eppig JT Hill DP Ni L Ringwald M Balakrishnan R Cherry JM Christie KR Costanzo MC Dwight SS Engel S Fisk DG Hirschman JE Hong EL Nash RS Sethuraman A Theesfeld CL Botstein D Dolinski K Feierbach B Berardini T Mundodi S Rhee SY Apweiler R Barrell D Camon E Dimmer E Lee V Chisholm R Gaudet P Kibbe W Kishore R Schwarz EM Sternberg P Gwinn M Hannick L Wortman J Berriman M Wood V de la Cruz N Tonellato P Jaiswal P Seigfried T White R The Gene Ontology (GO) database and informatics resource Nucleic Acids Res 2004 32 D258 61 14681407
Bult CJ Blake JA Richardson JE Kadin JA Eppig JT Baldarelli RM Barsanti K Baya M Beal JS Boddy WJ Bradt DW Burkart DL Butler NE Campbell J Corey R Corbani LE Cousins S Dene H Drabkin HJ Frazer K Garippa DM Glass LH Goldsmith CW Grant PL King BL Lennon-Pierce M Lewis J Lu I Lutz CM Maltais LJ McKenzie LM Miers D Modrusan D Ni L Ormsby JE Qi D Ramachandran S Reddy TB Reed DJ Sinclair R Shaw DR Smith CL Szauter P Taylor B Vanden Borre P Walker M Washburn L Witham I Winslow J Zhu Y The Mouse Genome Database (MGD): integrating biology with the genome Nucleic Acids Res 2004 32 D476 81 14681461 10.1093/nar/gkh125
Cherry JM Adler C Ball C Chervitz SA Dwight SS Hester ET Jia Y Juvik G Roe T Schroeder M Weng S Botstein D SGD: Saccharomyces Genome Database Nucleic Acids Res 1998 26 73 79 9399804 10.1093/nar/26.1.73
Twigger S Lu J Shimoyama M Chen D Pasko D Long H Ginster J Chen CF Nigam R Kwitek A Eppig J Maltais L Maglott D Schuler G Jacob H Tonellato PJ Rat Genome Database (RGD): mapping disease onto the genome Nucleic Acids Res 2002 30 125 128 11752273 10.1093/nar/30.1.125
Apweiler R Bairoch A Wu CH Barker WC Boeckmann B Ferro S Gasteiger E Huang H Lopez R Magrane M Martin MJ Natale DA O'Donovan C Redaschi N Yeh LS UniProt: the Universal Protein knowledgebase Nucleic Acids Res 2004 32 D115 9 14681372 10.1093/nar/gkh131
Apweiler R Attwood TK Bairoch A Bateman A Birney E Biswas M Bucher P Cerutti L Corpet F Croning MD Durbin R Falquet L Fleischmann W Gouzy J Hermjakob H Hulo N Jonassen I Kahn D Kanapin A Karavidopoulou Y Lopez R Marx B Mulder NJ Oinn TM Pagni M Servant F Sigrist CJ Zdobnov EM The InterPro database, an integrated documentation resource for protein families, domains and functional sites Nucleic Acids Res 2001 29 37 40 11125043 10.1093/nar/29.1.37
Friedman C Liu H Shagina L A vocabulary development and visualization tool based on natural language processing and the mining of textual patient reports J Biomed Inform 2003 36 189 201 14615228 10.1016/j.jbi.2003.08.005
Liu H Friedman C A method for vocabulary development and visualization based on medical language processing and XML Proc AMIA Symp 2000 502 506 11079934
Liu H Friedman C CliniViewer: a tool for viewing electronic medical records based on natural language processing and XML Medinfo 2004 11 639 643 15360891
Lord PW Stevens RD Brass A Goble CA Investigating semantic similarity measures across the Gene Ontology: the relationship between sequence and annotation Bioinformatics 2003 19 1275 1283 12835272 10.1093/bioinformatics/btg153
Go archive distribution on November 01, 2004
Kuriakose MA Chen WT He ZM Sikora AG Zhang P Zhang ZY Qiu WL Hsu DF McMunn-Coffran C Brown SM Elango EM Delacure MD Chen FA Selection and validation of differentially expressed genes in head and neck cancer Cell Mol Life Sci 2004 61 1372 1383 15170515 10.1007/s00018-004-4069-0
Nikolskaya AN Complementing Gene Ontology With PIRSF Classification-based Protein Ontology 2004
TIGR Resource website 2004
Wu CH Nikolskaya A Huang H Yeh LS Natale DA Vinayaka CR Hu ZZ Mazumder R Kumar S Kourtesis P Ledley RS Suzek BE Arminski L Chen Y Zhang J Cardenas JL Chung S Castro-Alvear J Dinkov G Barker WC PIRSF: family classification system at the Protein Information Resource Nucleic Acids Res 2004 32 D112 4 14681371 10.1093/nar/gkh097
Wang L Zhao J Parametric alignment of ordered trees Bioinformatics 2003 19 2237 2245 14630652 10.1093/bioinformatics/btg305
GenNav
MGI GO Browser
Zeeberg BR Feng W Wang G Wang MD Fojo AT Sunshine M Narasimhan S Kane DW Reinhold WC Lababidi S Bussey KJ Riss J Barrett JC Weinstein JN GoMiner: a resource for biological interpretation of genomic and proteomic data Genome Biol 2003 4 R28 12702209 10.1186/gb-2003-4-4-r28
Doniger SW Salomonis N Dahlquist KD Vranizan K Lawlor SC Conklin BR MAPPFinder: using Gene Ontology and GenMAPP to create a global gene-expression profile from microarray data Genome Biol 2003 4 R7 12540299 10.1186/gb-2003-4-1-r7
Al-Shahrour F Diaz-Uriarte R Dopazo J FatiGO: a web tool for finding significant associations of Gene Ontology terms with groups of genes Bioinformatics 2004 20 578 580 14990455 10.1093/bioinformatics/btg455
Zhong S Li C Wong WH ChipInfo: Software for extracting gene annotation and gene ontology information for microarray analysis Nucleic Acids Res 2003 31 3483 3486 12824349 10.1093/nar/gkg598
|
16091147
|
PMC1199584
|
CC BY
|
2021-01-04 16:27:26
|
no
|
BMC Bioinformatics. 2005 Aug 9; 6:201
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-201
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2031610217510.1186/1471-2105-6-203Research ArticleCalibration of mass spectrometric peptide mass fingerprint data without specific external or internal calibrants Wolski Witold E [email protected] Maciej [email protected] Peter [email protected] Knut [email protected] Max Planck Institute for Molecular Genetics, Ihnestraße 63–73, D-14195 Berlin, Germany2 Institute for Computer Science, Free University Berlin, Takustr. 9, 14195 Berlin, Germany3 Max Delbrück Center for Molecular Medicine, Robert-Roessle-Str. 10, D-13125 Berlin-Buch, Germany4 Max Planck Institute for Infection Biology, Schumannstr. 21–22, D-10117 Berlin, Germany5 School of Mathematics and Statistics, Merz Court, University of Newcastle upon Tyne, NE1 7RU, UK2005 15 8 2005 6 203 203 26 4 2005 15 8 2005 Copyright © 2005 Wolski et al; licensee BioMed Central Ltd.2005Wolski et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Peptide Mass Fingerprinting (PMF) is a widely used mass spectrometry (MS) method of analysis of proteins and peptides. It relies on the comparison between experimentally determined and theoretical mass spectra. The PMF process requires calibration, usually performed with external or internal calibrants of known molecular masses.
Results
We have introduced two novel MS calibration methods. The first method utilises the local similarity of peptide maps generated after separation of complex protein samples by two-dimensional gel electrophoresis. It computes a multiple peak-list alignment of the data set using a modified Minimum Spanning Tree (MST) algorithm. The second method exploits the idea that hundreds of MS samples are measured in parallel on one sample support. It improves the calibration coefficients by applying a two-dimensional Thin Plate Splines (TPS) smoothing algorithm. We studied the novel calibration methods utilising data generated by three different MALDI-TOF-MS instruments. We demonstrate that a PMF data set can be calibrated without resorting to external or relying on widely occurring internal calibrants. The methods developed here were implemented in R and are part of the BioConductor package mscalib available from .
Conclusion
The MST calibration algorithm is well suited to calibrate MS spectra of protein samples resulting from two-dimensional gel electrophoretic separation. The TPS based calibration algorithm might be used to correct systematic mass measurement errors observed for large MS sample supports. As compared to other methods, our combined MS spectra calibration strategy increases the peptide/protein identification rate by an additional 5 – 15%.
==== Body
Background
Proteomics inter-alia focuses on the identification of peptides/proteins in complex biological samples [1]. Before the identification of the complex constituents, several separation steps are required to reduce the sample complexity. The classical separation method is the two-dimensional gel electrophoresis [2-5], followed by excision of the detected spots from the gel, digestion with sequence specific proteases and extraction of the cleaved proteins [6,7]. Mass Spectrometric (MS) analysis [8-13] of the resulting mixture of peptides yields a peptide mass fingerprint (PMF): a set of measured molecular masses of the proteolytic peptides derived from the analysed protein [14-16].
PMF commonly requires matrix assisted laser desorption/ionisation (MALDI) time of flight (TOF) instruments, capable of high throughput analysis of complex samples with minimal pre-cleanup, high femtomolar range sensitivity and accuracy of peptide molecular mass determination up to 5 – 10 parts per million (ppm) [17-20]. Due to the high ion transmission of the TOF mass analyzer, this technique is more sensitive compared with other MS techniques. In relation to Electrospray ionisation (ESI) MS [21], MALDI-MS is more tolerant to sample contamination resulting from salts and detergents often present in protein samples due to the separation method. MALDI-MS and ESI-MS have become the standard high throughput proteome analysis techniques in many research laboratories.
The experimental peptide mass lists are generated by the analysis of TOF spectra [22]. Ideally, the TOF is proportional to the square root of mass over charge . Thus, in order to transform the spectrum from TOF into m/z, two calibration constants A and B are necessary. These can be derived by measuring the flight times t of at least two different ions with known masses and fitting them such that . After the transformation from time into m/z, the mono-isotopic peptide signals in the spectrum are identified and their intensity is determined by computational methods [23-26]. The lists of the first mono-isotopic peptide peaks – further called peak-lists – are used to identify the protein of interest. In order to assign the PMF to a protein in a sequence database, database search algorithms use the match (within a given measurement accuracy) of theoretical peptide masses computed from protein sequence databases [27] with observed MS masses [15,16].
Usually the scoring schemes model the mass frequencies of the proteins and peptides in the sequence databases [24,28-30]. Other properties to be considered include the different sensitivity of detection for individual peptides, known protein modifications, and/or possible mutations [23,31-33], although generally, all popular search scores depend on the precise assignment of experimental to theoretical peptide masses.
Two novel calibration methods
In a high throughput setting [34,35], where the samples are placed on a moving sample support, the calibration coefficients for transforming the TOF into m/z differ depending on sample position. This is due to deviations in plate flatness, sample topography changing the size of the acceleration region [34,36], and alterations in the strength of the electric field on the sample support borders which influences the drift velocity of the ions [22]. Thus, when calibration constants determined from one position on the sample support are used to calibrate TOF spectra acquired on other positions (a procedure known as external calibration), the determined m/z values have errors of up to 500 ppm.
Calibration is usually performed using external [36-38] or internal calibrants [39,40], which rely on known masses to calibrate the spectra to common co-ordinates. It must be stressed, that in some cases the signal of a reference compounds might be suppressed by the analyte molecules, thus precluding internal calibration. In other cases, the reference signal may partially overlap with an analyte signal, resulting in an erroneous assignment. A third category of calibration methods is based on the peptide mass rule [23,24]. A major advantage of the latter method is that no internal calibrants are required to calibrate the peak-lists. The limitation of this method is it's sensitivity to the presence of non-peptide peaks in the spectra, and that it completely fails if the number of peptide peaks in peak-lists are small [23,24,39]. Therefore, in practice this method usually is used only to pre-calibrate [24] or to support the results of internal calibration [26,39].
We have developed two novel calibration methods for PMF data. Both calibration methods exploit similarities of peak-lists due to closeness in the origin of the analysed samples. The first method combines the computation of dissimilarities [41] between peak-lists with internal calibration. The second method employs spatial statistical methods [42] to model systematic changes of the calibration-model over the MALDI sample support. The major advantage of the presented methods originates from the fact that the MS calibration derives from samples without internal standards or external calibrants positioned on each sample support.
Evaluating the methods
To demonstrate the accuracy of our methods, we studied one sample set of 380 mass spectra, consisting of a part of the Arabidopsis thaliana proteome study [43]. For this purpose, a MALDI MS sample support in pre-structured [35] (384-well) microtitre plate format was used. The measurements were performed using the Autoflex MALDI-TOF MS [44] instrument.
To compare the performance of calibration methods described here with those already published [26,39], we used two different data sets. The first set consisted of 1193 spectra deposited on four pre-structured sample supports and measured on a Reflex MALDI-TOF MS [44] instrument (Reflex data set). Spectra were generated via mass spectrometric analysis of the Rhodopirellula baltica proteome (unpublished data). The second set was generated in connection with a proteome study of Mus musculus and consisted of 1882 spectra deposited on five pre-structured sample supports and measured on an Ultraflex MALDI-TOF MS [44] instrument (Ultraflex data set).
During MS sample preparation of the Ultraflex data set, standard peptides of known masses (human Angiotensin I – 1, 296.6853Da, human ACTH (18–39) 2, 465.1989Da) were added before the measurement to the MS matrix. This was done because the data sets were optimised for the calibration methods, which required the internal calibrants. We examined if the standard peaks could be observed in more than 33% of spectra and if so, we removed the peaks matching these masses from the data set. This procedure was applied in order to simulate a data set not optimised for internal calibration.
The Rhodopirellula peptide peak-lists were searched against a Pirelulla database [45] with 13, 331 predicted Open Reading Frames (ORFs). The Mus musculus samples underwent searches against the Mus musculus entries (69, 343 -sequences) of the NCBI non-redundant protein database [46].
Results and discussion
Internal calibration using a pre-calibrated list of calibration masses
Internal calibration is a widely used method in mass spectrometry. This method fails however, either if no peaks matching known masses are present or if MS peak assignment is false. A detailed description of the application of internal calibration in a high throughput-MS setting, addressing the two points is given by e.g. Chamrad et al. [39], Levander et al. [40] and Samuelson et al. [26]. In order to avoid the lack of MS peaks matching the known calibration masses the authors used a pre-compiled list, e.g. trypsin autolysis peaks and unidentified, frequently observed masses [47].
Chamrad et al. [39] initiated the calibration procedure with searches for matching masses using a relatively large search window and iterated it with an increased accuracy. In this scheme, a large search window allows false assignments for calibration masses to occur more frequently. If a false assignment occurs in the first iteration, then the determined calibration constants are false and the entire calibration would be wrong. In the next round of calibration, where a search for matching masses is performed with a higher mass accuracy, the calibration would also fail. To prevent this, the authors [26,39] checked the obtained calibration coefficients against the peptide mass rule (PM-rule) [24,48] and stopped further calibration attempts where they disagreed substantially.
Levander et al. [40] introduced an adaptive method to eliminate low-sensitivity auto-proteolysis trypsin peaks from the calibration mass list if no high-sensitivity trypsin peaks e.g. (842.5099Da, 1045.5642Da, 2211.1046Da) were found to decrease the chance of false matches. Unfortunately, this method could only be applied for "tryptic" calibration peaks.
Figures 1A &1B demonstrate the limitations of a calibration list compiled from ubiquitous masses of the whole data set. One can recognise that out of three abundant masses (in red, Figure 1A), only two can be practically used for calibration. Specifically, the first and the third abundant mass in the list of ubiquitous masses (Figure 1A) match simultaneously two peaks in peak-list 3, 4 and 5 (Figure 1B). Thus, out of five peak-lists only three could be calibrated. The second calibration mass is also of no use, since it is the only calibration mass in the peak-lists 1 and 2 (although these peak-lists do contain other shared masses). This illustrates that the usage of a global calibration list may fail to calibrate a set of peak-lists.
It is therefore feasible to address the following questions: How can one obtain a short calibration list to avoid spurious matches while at the same time it matching a sufficient number of peaks in every peak-list of the set? In addition, how can one minimise the initial search window to avoid false matches?
Figure 1 A: Histogram of masses present in the stick spectra in B. In red, marked masses recognised as ubiquitous. B: Stick spectra of five hypothetical peak-lists. Red vertical lines mark the position of ubiquitous masses determined using the histogram in A. C: Single linkage-clustering dendrogram of the peak-lists in B. As dissimilarity the mass measurement range (1500 Da) minus the range enclosed by matching peaks was used. D: Minimum spanning tree.
Finding the optimal multiple peak-list alignment using a modified Minimum Spanning Tree (MST) algorithm
In order to bypass the limitations imposed by global calibration we used an observation made by Schmidt et al. [49]. They noticed that protein samples excised from high-resolution 2D-gels are usually not ideally separated and therefore exhibit local similarities. Compiling a calibration list of abundant masses from a whole data set obtained from a 2D-gel does not differentiate local spectra similarities. For example peak-lists 1, 2 and 3 (Figure 1B) share peaks, which were not recognised as ubiquitous masses and hence not used further for calibration using a global calibration list. The peak-list pairs (2,3) and (1,3) shared more than one peak, thus allowing an easy calibration.
We explored the property of local pairwise peak-list similarities for calibration of data sets. To achieve it, we used a modified minimum spanning tree MST [50] algorithm on the complete weighted graph G(V, E, d), where the vertex set V corresponds to the individual peak-lists and the edges E are weighted by a dissimilarity measure d. We denned the measure between two peak-lists p1 and p2 as d(p1, p2) = -s(p1, p2), where s represented a similarity measure denned in Equation 10. This measure not only counts the number of matching peaks, but also weights the mass range enclosed by them. Hence, it also considers that if the matching masses lie very close to each other, the calibration model describes a small mass range only, and can result in a large error when aligning masses that are out of this range. Using the dissimilarities one can compute a MST (Figure 1D). The algorithm to compute the MST of the peak-list data set starts by choosing a peak-list (named s), which belongs to the peak-list pair of smallest dissimilarity, for example peak-list 2 or 3 in Figure 1. This peak-list is the root of the growing tree T (Figure 8 line 1). Next, a peak-list v was chosen, which easily could be aligned to peak-list u where v is a part of the growing tree i.e. u ∊ T (Figure 8 line 5), for example peak-list v = 2 can easily be aligned to peak-list u = 3. Using linear regression, we computed the coefficients c(v, u) = (c0, c1) of the affine function, modelling the absolute mass differences of the peaks matching in the peak-list pair (v, u). Having these coefficients one can compute the calibration coefficients c(v, s) using the update rule in Equation 11, which described the mass measurement error (MME) between the peak-list v and the starting peak-list s. The calibration is not terminated until the whole tree is built. We then added peak-list v to the tree T and have iterated the procedure until all peak-lists were appended to the tree, for example by adding peak-list 4, then 5 and finally 1 to T (Figure 1D).
Figure 8 Modified Dijkstra-Prim MST algorithm. The algorithm starts with vertex s (peak-list) belonging to the peak-list pair with smallest distance (line 1) (the standard algorithm starts with an arbitrary pair). In addition to computing the MST T, the algorithm computes the calibration constants C(v, s) (line 8) and the connection weight W(u) (line 9).
In the MST algorithm, the vertices are joined by edges of smallest dissimilarity. Consequently, the MST algorithm connects all peak-lists in the data set in the way that the length of the path from the peak-list of origin (root of the tree: peak-list 3 in Figure 1D) to any peak-list in the data set is minimal. The algorithm for computing the agglomerative clustering using the single linkage method [51,52] works similarly like the MST algorithm and therefore the dendrogram (Figure 1C) provides (as read from bottom to top) the order, by which the peak-list pairs were chosen. The horizontal lines joining two dendrogram tree branches were drawn at the height of the value of the minimal dissimilarity of two peak-lists in either branch.
Finally, the algorithm returns a list of coefficients and a measure of confidence for all peak-lists equalling the smallest similarity in the path from s to v.
Figure 2A demonstrates how the samples on the target are connected by the edges. Green dots (brighter) represent leaves, while blue dots (darker) denote interior vertices. The peak-list of origin s is marked with a red cross-hairs (sample position D15). Note that long peak-lists (brighter squares) are interior vertices of the MST.
Figure 2 A: Colour scheme coded peak-list lengths in dependence of the sample support position. Blue dots – interior vertex, Green dots – end vertex, white arrows – connecting edges of the MST. The red hair-cross indicates the peak-list of origin s. B: Colour scheme coded slope coefficient of the mass- dependent calibration function in relation to sample support position. C1, C2: Strip chart of the data set for a mass range of 2210 – 2212Da (top) and 842 – 843Da (bottom), including the tryptic autolysis peaks 842.508 and 2211.100Da. Black hair-crosses – masses of peaks before calibration, red circles – masses after calibration. Vertical blue line – the exact position of trypsin autolysis masses 842.508 and 2211.100Da.
The strip-charts of mass ranges including peaks of the trypsin autolysis products 842.508 and 2, 211.100 are presented in Figure 2C1 and C2. One can observe that the MST-method works robustly on raw data with a mass measurement error of up to ± 0.7Da (black crosses), even if the search for matching peaks when computing the similarities and calibration coefficients was performed within a much smaller window of ± 0.45Da. Notably, if the maximal error among two peak-lists is much larger than the search window, it is still possible to find a path, thus allowing alignment of two extreme peak-lists.
Due to the fact that all peak-lists were aligned to the peak-list of origin s, which did not necessarily match to the theoretical trypsin autolysis masses, a final correction was required to calibrate the whole tree to the theoretical co-ordinate system before database searches (not shown).
Determining the calibration model of the sample support using Thin-Plate Spline interpolation (TPS)
Because a large part of the MME is of systematic origin and depends on the sample support position, the mapping of the calibration coefficients across the entire MALDI plate was introduced by Gobom et al. [36] and Moskovets et al. [38]. The calibration coefficients were determined using a standard mixture of peptides with known masses. Subsequently, the calibration coefficients were used during MS analysis in order to correct for the masses measured afterwards on the same plate.
We introduced here a method that derives the calibration model from calibration coefficients acquired from samples, which do not necessarily contain internal standards. Instead of refining the MST calibration model, we chose the peptide mass rule based approach, namely Linear Regression on Peptide Rule (cf. Methods), to obtain the calibration coefficients. The methods based on the peptide mass rule do not rely on the specification of an initial search window or on internal calibrant masses. The peptide rule based calibration method calibrates the peak-lists into the theoretical co-ordinate system and increases the mass accuracy to approximately 0.1Da, but fails if the peak-list is too short, which indeed could be observed for several samples (Figure 3A and 3C). Figure 3A provides the color scheme coded slope coefficient c1 as determined by the peptide rule based calibration method in dependence of the target location. One can observe that some erroneous predictions occur (Figure 3C; black crosses marked by magenta triangles).
Figure 3 A: Colour scheme coded slope coefficients c1 of the MME determined by the peptide rule based calibration method. B: The slope coefficient as predicted from the refined samples determined by TPS with λ = 0.001. C: Strip chart of the data set for a mass range of 2210 – 2212Da (C1) and 842 – 843Da (C2), including the tryptic autolysis peaks 842.508 and 2211.100Da. Black crosses – masses of peaks predicted by the peptide rule based calibration method, red circles – masses predicted by the TPS calibration method. Vertical blue line – exact position of trypsin autolysis masses 842.508 and 2211.100Da. Dashed red vertical line – mass of the extreme peptide masses after TPS calibration.
However, it is unbiased to assume a smooth transition between adjacent positions of the sample support. For example, Figure 2B demonstrates that the slope coefficient of the sample calibration-model obtained by the MST calibration methods increases for samples close to the support border. This change is due to alterations in the electric field E (Equation 1) influencing the flight velocity given by
where sa is the size of the acceleration region, z is the ion charge and m is the mass of the ion. We determined the systematic change of the slope using the Thin-Plate Spline (TPS) interpolation method [42,53]. At first, we computed the TPS with a degree of smoothing λ = 5·10-2 (see Equation 15). Calibration models with slope coefficient c1 that varies more than ± 1·10-4 or with intercept coefficient c0 varying more than 0.2Da from the one predicted by the TPS were discarded. Using the remaining calibration models, the TPS was recomputed with smaller degree of smoothing λ = 1·10-3. Figure 3B, demonstrates the Colour scheme coded slope coefficient c1, as estimated by the refined TPS. This model resembles the one generated by the MST method (Figure 2B). We corrected the peak-lists masses (black cross hairs, Figure 3C), using the TPS values as estimates of the slope coefficients, and as intercept estimate we used the average intercept of all coefficients of the refined calibration models to obtain the calibrated masses (red circles).
The TPS method reduced the MME of a peak-list compared to any other peak-list in the data set (vertical red, dashed line in Figure 3C) down to 0.3Da, as compared to 1.5Da for raw data. This is approximately a 5- fold increase of a mass measurement accuracy. This decrease of the MME enabled us to utilise the MST-algorithm with an accuracy of ± 0.15Da, reducing further the probability of false assignments of calibration masses. In addition, the histogram of dissimilarities computed for all peak-list pairs (Figure 4A) shows for TPS calibrated data lower values of dissimilarity (in red) as compared to the raw data (in grey), even if the first dissimilarities were computed with a search window of 0.15Da and the second ones with a search window of 0.45Da. A subsequent calibration using the MST method decreased further the MME (Figure 4B).
Figure 4 A:Histogram of pairwise peak-list similarities. In gray – raw data and similarities computed with an accuracy of ± 0.4Da. In red – similarities computed with accuracy of ± 0.15Da using LR/PR-TPS calibrated data. B: Strip chart of peak-lists. Grey triangles – masses after TPS-calibration, green circles – data after TPS-MST- calibration, red circles – data calibrated into the theoretical co-ordinate system, defined by theoretical tryptic autolysis masses (blue vertical lines.)
The mass measurement error
Prior to the calibration, the main error source is due to different drift velocities of the ions causing an increase of the absolute MME, proportional to mass and best described by the slope coefficient c1 ≠ 0 and measured as relative error using parts per million ppm (Table 1, row 1 and 2). After removal of this error using calibration methods, for example the TPS calibration (Table 1, row 3,4) or TPS with subsequent MST calibration (Table 1 row 5,6), the main contribution to the MME was due to peak detection performance. We were aware, however, of systematic changes of the MME, which can be described using higher order polynomials [37,54]. We have removed higher order terms of the MME, by applying external calibration before to other calibration procedures (cf. Methods : External Calibration). The change of peak-detection quality was negligible in the range of 500 – 4000Da. Figure 5, as well as Table 1, illustrates that after calibration the absolute MME was smaller for the peak with higher mass (2211.1) than that of the peak with a lower mass (842.508) if the peak intensity and consequently the Signal to noise ratio remained sufficiently high. Therefore, we performed the database searches by specifying the search window in Da instead of ppm.
Figure 5 Stick spectrum of the merged data set of 380 peak-lists. The black vertical lines represent peaks calibrated using the TPS and MST method. Their height equals their intensity. Green line – average mass of all peaks in the region 842 – 843Da (A) and 2210.5 – 2211.6Da (B). The orange vertical lines represent the average mass ±, the standard deviation of the peak masses in each region. Magenta line – density of peak-masses.
Table 1 Mass Measurement Error. Standard deviation (SN) observed for the trytpic autolysis peaks 842.508 and 2211.1. Raw data; TPS – Thin-Plate Spline (TPS) calibrated data; TPS-MST – The data, which undergone Thin-Plate Spline (TPS)(pre-processing), followed by Maximum Spanning Tree (MST) calibration
Calibration Mass SN[Da] SN [ppm]
Raw data 842.508 0.1 118
Raw data 2211.1 0.3 135
TPS 842.508 0.03 37
TPS 2211.1 0.057 26
TPS-MST 842.508 0.012 14.5
TPS-MST 2211.1 0.01 4.6
The optimal size of the search window
Figure 5 and Table 1 demonstrate that it is possible to reduce the mass measurement error to approximately ± 10 ppm for most of the peak-lists in a dataset consisting of 380 spectra, by applying the TPS-MST calibration sequence. Nevertheless, in this dataset one can observe peak-lists that do not exhibit such high mass measurement accuracy. Consequently, if the database searches were performed with a search window of 10 ppm, these PLs would not be identified.
The optimal size of the search window was determined by searching of four internally calibrated data sets with five different search window sizes, namely 0.5, 0.2, 0.1, 0.05 and 0.02Da using the Mascot [55] search algorithm. The search window of 0.2Da generated the highest identification rate. Figure 6 shows the relative identification rate (identification rate / max(identification rate)·100%). Allowing the search window to be larger e.g. 0.5Da, decreases the identification rate by increasing the rate of false negatives, while a smaller window e.g. ± 0.05Da decreases it by rejecting true matches [55]. Because the identification rate for a search window of 0.1Da is only slightly worse than one of 0.2Da, and since it minimizes the risk of false positive matches, we further compared the practical performance of the calibration methods with a search window of 0.1Da.
Figure 6 The optimal search window. Comparison of the relative identification rates of internally calibrated data (Y-axis) given a search window size of 0.5Da, 0.2Da, 0.1Da, 0.05Da and 0.02Da, respectively (X-axis). Red – Two Reflex (Pirellula) dataset, Black – Two Ultraflex (Mus Musculus) datasets.
Prior to the database searches we removed all masses that occur in more than 8% of spectra, as it significantly increased the identification rate [39,40] (cf. Methods – Filtering of ubiquitous masses prior to database search). The sequence data base search was performed using the Mascot [55] search software version 1.8.1. We interfaced the search server from within R using the in-house developed R package msmascot [56].
Combining different calibration methods and their comparison
All parameters were fitted to a data set optimised for internal calibration, measured on an Autoflex MALDI-TOF MS [44] instrument. We applied the calibration methods introduced (MST and TPS based calibration) without changing the parameters to two sample sets obtained using two different instruments, namely a Reflex MALDI-TOF MS and a Ultraflex MALDI-TOF MS instrument. This was executed to illustrate that our methods are robust with respect to different instruments even if the parameters were not optimised for the respective machines.
We combined the different pre-calibration and calibration methods resulting in six different calibration sequences (summarised in Table 2). We compared the performance of the MST and TPS calibration sequence to the internal calibration (IC), and the peptide rule based calibration methods (LR/PR). Furthermore, we investigated if the identification rate of the TPS based method could be improved further by subsequent internal (TPS-IC) or MST calibration (TPS-MST). The R [57] scripts implementing each sequence can be found in the samples directory of the mscalib BioConductor [58] package.
Table 2 Calibration sequences. LR/PR – linear regression on peptide rule, IC – Internal calibration with two iterations. (Bruker Reflex – mass measurement error (MME) window of 450 and 250 ppm, Bruker Ultraflex 250 and 125 ppm); MST – MST calibration method computed with an search window of ± 0.4Da; TPS-IC – Pre-processing (TPS calibration) and subsequent internal calibration with a MME window of 250 ppm; TPS-MST – pre-processing and an MST with a search window of ± 0.25Da;
Abbreviation Description
1 LR/PR peptide rule calibration.
2 IC internal calibration 450 ppm and 250 ppm.
3 MST minimum spanning tree calibration.
4 TPS LR/PR and subsequent thin-plate spline (TPS) calibration.
5 TPS-IC TPS calibration and subsequent internal calibration.
6 TPS-MST TPS calibration and subsequent MST calibration.
The only calibration method for which parameters were optimised with respect to the instrument was the standard internal calibration (IC) method, which employs a pre-compiled calibration list of theoretical trypsin autolysis peaks and a calibrated set of ubiquitous masses (cf. Methods – Standard internal calibration). In case of the peptide rule based calibration (LR/PR) method we applied an additional filtering of the calibration-models. Only models with an intercept coefficient c0 satisfying -0.4Da <c0 < 0.4Da and slope coefficients c1 with -5·10-3 <c1 < 5·10-3 were kept. In order to avoid falsely calibrated peak-lists we performed the filtering.
The identification rates were defined as the number of identified samples by at least one of the calibration sequences divided by the number of samples submitted for searches
where CSi indicates the set of identified samples by one of the calibration sequences (Table 2), and #{A} denotes the number of elements in a set A. The identification rates were 74%, 87%, 79%, 85% for the Pirellula (Reflex) data set, with an overall identification rate of 82%, whereas for the Mus musculus (Ultraflex) data set they were 51%, 72%, 35%, 51%, 27%, with an overall identification rate of 58%. The lower identification rate of the Mus musculus data set can possibly be explained by the fact that it was matched with a larger database. Therefore, more matching peaks are required to make significant assignments to a data base entry.
In order to directly compare the identification rates for both data sets and each calibration sequence, we computed the relative identification rate. It was defined as the ratio of the number of identified samples calibrated by a sequence (numerator) and of the number of identified samples, which could be identified by at least one method (denominator):
The relative identification rate is indicated by the dots, joined by continuous lines for readability purposes only, in Figure 7. The dashed lines denote the average of the sequence coverage of all identified samples. Figure 7A presents the results for the four Pirellula data sets, while Figure 7B shows the results of five Mus musculus data sets.
Figure 7 Relative identification rate in % (continuous line – left y-axis) and sequence coverage in % (dashed lines – right y-axis). LR/PR – linear regression on peptide rule, IC – two step internal calibration, MST – minimum spanning tree calibration, P – TPS calibration, TPS-IC – TPS calibration and subsequent internal calibration, TPS-MST – TPS calibration and subsequent MST calibration.
Only in one case of one data set was a single calibration sequence TPS-MST (see Table 2) able to identify all peak-lists (100% identification rate) and therefore it completely dominated over the other methods (black line, Figure 7A). In the case of the Ultraflex data set (Figure 7B) we observed that the TPS-MST method had the highest identification rate, while in Reflex data set (Figure 7A) it achieved the highest performance for approximately half of the data sets.
Figure 7C illustrates the averaged relative identification rate of the calibration methods for the Ultraflex and Autoflex data sets. In addition, it demonstrates that the ordering of the calibration methods according to the relative identification rate does not depend on the value of the Probability Based Mowse Score [55] (PBMS) used as identification threshold. The dashed lines (Figure 5) indicate the identification rates obtained for a PBMS 5 units higher than the one used to identify the samples with a 0.5% significance level (continuous lines).
Interestingly, the TPS smoothing method resulted in an overall higher identification rate than the other methods tested on raw data (peptide rule based calibration, internal calibration, MST-calibration), except for one case of the Ultraflex data set. Furthermore, a combination of the internal calibration with TPS calibration (TPS-IC) did not increase either the sequence coverage (dashed lines) or the identification rate of the TPS method applied alone.
In two out of the four Reflex data sets, the MST method applied on TPS-processed data (P-TPS Figure 7A, dashed lines) slightly decreased the sequence coverage indicating a reduction of calibration accuracy. For the Ultraflex data sets, the sequence coverage correlated well with the identification rate and the TPS-MST-method accomplished the highest performance.
Moreover, if similar identification rates of the peptide rule based calibration and the internal calibration were observed, the peptide rule based calibration method provided higher sequence coverage (Figure 7B). This could be explained by the fact that the peptide rule based method calibrated well the peak-lists possessing many peptide peaks. Such peak-lists potentially contain the higher sequence coverage.
The BioConductor package mscalib
All of the calibration methods are part of the mscalib programme, which is available as a BioConductor [59] package. The Bioconductor project is an initiative for the collaborative creation of extensible software for computational biology and bioinformatics [58]. The scripts carrying out the calibration sequences tested, can be found in the subdirectory/samples of the package. Furthermore, in the same directory and in the directory/doc there are two vignettes [60] with detailed descriptions of two selected calibration sequences.
Conclusion
While the methods described in this study significantly improve the calibration of raw data, they do not perform better than other published calibration routines which reduce the MME to 10 ppm or below. The real advantage of the methods described here is that they are not dependent on the presence of internal or external calibrants, required to correct for the affine component of the MME. Furthermore, the calibration methods described in this study allow a larger fraction of peak-lists in the datasets to be calibrated than the reference internal calibration method would do.
The TPS method deals with systematic detrimental calibration effects that are due to imperfections in the geometry of the electric field over the MALDI sample plates. Usage of TPS calibration results in up to 10% higher identification rates, at least for the Bruker mass spectrometers, than the internal calibration. The TPS calibration procedure enables, for most of the samples deposited on the sample support, to obtain mass accuracy in the range of ± 0.1Da. Moreover, the TPS method does not require the presence of internal calibrants since it relies on calibration coefficients acquired from a calibration method based on the peptide mass rule.
The MST method is able to increase the identification rates obtained by the TPS-method for protein samples separated by a 2D-Gel electrophoretic procedure. Furthermore, the parameters optimised for one instrument (Autoflex) can be directly utilised for other instruments (Reflex, Ultraflex).
In this work, we have only examined a version of the MST algorithm that builds a single tree for all peak-lists. This is adequate if the data are a set of peak-lists with smooth transitions in the similarity values. If this is not the case, it might be more appropriate to compute a forest of several MSTs. We have examined, however, only a single peak-list similarity measure (Equation 10) for peak-lists calibration. It is possible that better similarity measures can still be generated and subsequently applied for peak-lists calibration.
Complete utilisation of microtitre plates and sample supports is not only rational with respect to increased accuracy of the TPS method, but also with respect to the idea of high throughput experiments – maximal utilisation of energy and resources. Dense excision of spots from 2D-gels not only increases the performance of the MST method, but also identifies novel proteins. Hence, the main contribution of this manuscript is to present two calibration methods, compatible with the principle of high throughput sample processing and aims to identify a maximum of the proteins resolved on 2D-gels.
However, no single "best-calibration" method exists. Each of the methods utilises different properties of the peak-lists. Consequently, applying these methods in parallel and determining the total (union) of the identified samples provides the highest identification rate.
Methods
Data sets
In this study, we used three data sets generated in different proteome analyses:
1. A bacterial proteome Rhodopirellula baltica (unpublished data) (1,193 spectra) measured on a Reflex III [44] MALDI-TOF instrument.
2. A mammalian proteome Mus musclus (1,882 spectra) measured on Ultraflex [44] MALDI-TOF instrument.
3. A plant proteome Arabidopsis thaliana [43] measured on an Autoflex [44] MALDI-TOF instrument.
All PMF MS spectra derive from tryptic protein digests of individually excised protein spots. For this purpose, the whole tissue/cell protein extracts of the former mentioned organisms were separated by two-dimensional (2D) gel electrophoresis [4] and visualised with MS compatible Coomassie brilliant blue G250 [43]. The MALDI-TOF MS analysis was performed using delayed ion extraction and by employing the MALDI AnchorChip ™targets (Bruker Daltonics, Bremen, Germany). Positively charged ions in the range of 700 – 4, 500 m/z were recorded. Subsequently, the SNAP algorithm of the XTOF spectrum analysis software (Bruker Daltonics, Bremen, Germany) detected the monoisotopic masses of the measured peptides. The sum of the detected monoisotopic masses constitutes the raw peak-list. Before affine mass calibration, mass measurement errors which can be described by higher order polynomials and determined using external calibration (cf. Methods: External Calibration), were removed. Processed peak-lists were then used for the protein database searches with the Mascot search software (Version 1.8.1) [55], employing a mass accuracy of ± 0.1Da. Methionine oxidation was set as a variable and carbamidomethylation of cysteine residues as fixed modification. We allowed only one missed proteolytic cleavage site in the analysis.
Describing the Mass Measurement Error (MME) and predicting the correct mass
A mass difference can be described either in absolute ΔA = my - mx[m/z] or in relative ΔR = (my - mx)·106/my[ppm] units. The masses in two peak-lists X, Y were compared to each other and we considered two peaks to match, in the case of the absolute error if ΔA <a[m/z] and in the case of the relative errors if ΔR <a[ppm]. If we plotted ΔA or ΔR as a function of mtheo, we observed, besides a white noise component ε ∝ N(0, σ2), a systematic dependence. This dependence was modelled using a function . Given we corrected the experimental masses using the equations:
depending on whether the relative or absolute error was used, to obtain corrected masses mcorr.
Affine MME model
In the first approximation, the MME can be described by an affine function , where mi is the mass of the matching peaks. The intercept and slope coefficients of this function can be determined using linear regression.
If only one matching peak was found or the mass range enclosed by the matching masses was small (e.g. less than 200Da), as a remedy one can fix:
• the intercept at 0, if absolute difference ΔA[Da],
• the slope coefficient at 0, if relative difference ΔR[ppm]
and determine the slope or intercept respectively from the data.
To correct the experimental masses mexp we used Equation 5 for the absolute differences ΔA of matching peaks and Equation 4 in case of relative differences ΔR.
The difference between theoretical and measured masses is called a mass measurement error MME, while the alignment of mexp on mtheo an internal calibration [23,54,61].
Determining ubiquitous masses and their filtering
To determine the abundant masses we computed two histograms for each data set. The origin in the first histogram is x0 = min (M) - h and of the second histogram is x0 = min (M) - h/2, where M are all masses in the data set and the bandwidth h equals the measurement accuracy (in Da). We divided the range of M into bins of bandwidth h
Bj = [x0 + (j - 1)h, x0 + jh], with j ∈ 1,..., l, (6)
where l = (max(M) - x0) mod h. Formally the histogram of counts f is given by [62]
where n represented the number of masses in M. If a bin had more counts than a given threshold, the average mass of all peaks in the bin was computed. In the case of two adjacent or overlapping bins B1, B2 with a significant number of counts c, we first computed a weighted average of the bin midpoints using the number of counts in each bin as weight
where m1 and m2 are the bin midpoints. Afterwards, the average mass of all peaks in the range m ± h/2 was computed. All peaks with mass m ∈ [ ± h/2] were subsequently removed from the data set. Using two overlapping histograms allows the detection of clusters that are scattered over two adjacent bins in one of the histograms. Different ways to determine ubiquitous masses were used and reported by Levender et al. [40] and Kreitler [63].
Standard internal calibration – Alignment to a pre-compiled list of calibration masses
Instead of using a predefined list of calibration masses, we chose the calibration masses adaptively. The calibration list consisted of ubiquitous masses determined for the data set (cf. Determining ubiquitous masses). Some of the peaks in the list of ubiquitous masses could be assigned to tryptic autolysis products.
These matches were used to calibrate the abundant masses. The peak-lists in the data set were then aligned to the calibrated list of ubiquitous masses.
Filtering of ubiquitous masses prior to database search
We removed ubiquitous masses that occurred in more than 7.7% of peak-lists [39,40]. Filtering of ubiquitous masses was performed on a calibrated set of peak-lists. As a result, we could use a small bandwidth of h = 0.2Da (Equation 6) to determine ubiquitous masses. Next, we checked which of them can be assigned with a significant Probability Based Mascot Score (PBMS) to a sequence database entry and subsequently removed these masses from the filtering list. Abundant masses assigned to a database entry usually result from proteins multiply detected on a 2D-gel. The multiple identification is due to different localisation of the protein on the 2D-gel caused by: protein modifications (phosphorylation, glycosylation), different splice variants or by partial protein degradation. Finally, we removed all peaks within the range ± 0.1Da around the ubiquitous masses.
Linear regression and peptide mass rule algorithm
Wolski et al. (publication in preparation) defined the distance measure
which computes given λDB (the average peptide cluster distance for a sequence database DB against which the search is performed, e.g. λDB = 1.000495) the deviation of a peptide mass difference |mi - mj| from the closest monoisotopic mass predicted by the PM-rule [48]. If there was a linear dependence between |mi - mj| and dλ (mi, mj), then it was caused by the slope of the MME. If we computed all differences |mj - mi| and dλ (mi, mj) for peak pairs mi, mj with |mi, mj| < 1400, we could determine the slope coefficient c1 using linear regression, while fixing the intercept to zero [64]. In order to make the prediction robust against e.g. non-peptide peaks, we used a robust linear regression [65]. We removed the slope by multiplying each mass mi in the peak-list by (1 - c1). Next, we identified the intercept, which was the average of the distance dλ (mi, 0), and corrected for it.
External calibration
In order to model higher order systematic changes of mass dependent differences Δ of experimental mexp and reference masses mtheo, the measurements must be evenly distributed over the whole measurement range [37,66]. To model the dependence Δ ∝ m we used a cubic smoothing spline function [67,68], given by Δ = f(m) + εi, where f is a smooth function, and εi ~ N(0, σ2).
In our study, we used an implementation of the smoothing spline function, provided by B.D. Ripley and Martin Mächler (based on Fortran code of T. Hastie and R. Tibshirani) as part of the R-stats package. Other non-parametric regression methods like local polynomial regression [69] generated similar results for all types of instruments used in this study.
To obtain equidistantly spaced measurements of known masses, External calibration was employed. Some sample spots on the sample support are dedicated to calibration only. Calibration samples, of polymer mixtures [36], which yield equidistant peaks were used to precisely estimate the mass-dependent difference function.
Similarity/quality measures for internal calibration
Peak-lists can be easily aligned if they contain many matching peaks and the masses of these peaks span a wide mass range. The alignment of a peak-list pair (X, Y) fails if no matching peaks are found. We described these properties mathematically by the following similarity measure:
where n represented the number of matches, while mi and mj were the masses of matching peaks. This measure computed the sum of all mass differences of the matching peaks. The power p could be used to weight the large differences stronger.
Alignment of a set of peak-list using a Minimum Spanning Tree
To align a whole data-set to a single peak-list and to align the peak-lists with the highest similarity given by Equation 10, we computed for all peak-lists pairs a distance matrix D by casting the similarities into dissimilarities. This distance matrix can be represented by a complete, weighted graph G, where the vertices V correspond to peak-lists and the edges are weighted with the pairwise dissimilarity. To connect all vertices in the graph G with edges e of maximal similarity, the Dijkstra-Prim algorithm for finding the Minimum Spanning Tree(MST) [50] was implemented. We present here a modified version of this algorithm (see Figure 8). The algorithm was modified with respect to the starting conditions. As a starting-vertex s we chose a vertex incident to an edge of smallest distance. In addition to the MST tree T, the algorithm returns also a list of calibration coefficients C, which align all peak-lists V in the data set to the starting vertex (peak-list) s, and a list with connection weights W.
By traversing the edges in T, we reached each vertex in G, starting at s via edges with the highest possible calibration similarity (smallest distance). This is because we picked D(uv) with the smallest possible distance (Figure 8, line 5).
To align peak-list v to the starting peak-list s we needed to determine the coefficients C(v, s) of the difference function (Equation 5). We could obtain them from the coefficients C(v, u) and C(u, s) of the pairwise difference function and by:
where e.g. denotes the slope coefficient, and the intercept of the function .
Proof
The masses of the peak-list pairs (v, u) as well as (u, s) can be aligned given the C(v, u) and C(u, s) using the equations
Hence,
C(v, s) was computed online using Equation 11 while growing the tree (Figure 8, line 8). Subsequently, the algorithm returned a list C of calibration constants, where C(v, s) described the calibration coefficients allowing to transform peak-list v into the co-ordinate system of the peak-list of origin s.
In order to gain more confidence in the calibration constants in C, the MST algorithm was iterated n times. For computing the consecutive. Ti, Ci, Wi, Di with i = 2,..., n we applied the dissimilarity matrix Di-1 and set as a starting vertex si= s1 – the vertex incident to the edge of highest similarity in D1. The returned Ti, Ci, Wi, Di differed since we removed in iteration i - 1 each visited edge (Figure 8, line 6).
The calibration constants Ci(v, s) with i = 1,.., n should ideally be the same. It is known that Ci(v, s) differ due to alignment errors. Therefore, we computed a weighted average of the coefficients of the difference model. As weight of each model Ci(v, s) we utilised the smallest pairwise calibration similarity Wi(v) (Figure 8, line 9), on the path from s to v:
We applied the calibration constants in Cw to align all peak-lists to the peak-list s.
Abbreviations
• MME – mass measurement error
• MST – minimum spanning tree.
• MS – Mass Spectrometry.
• TOF – Time of Flight.
• MALDI – Matrix Assisted Laser Desorption Ionization.
• mod – modulo operator.
• TPS – Thin plate spline.
Authors' contributions
ML, KR and PJ gave initial input to the research.
WEW implemented the BioConductor package mscalib, msmascot, carried out the analysis, visualised the results and wrote the manuscript.
ML wrote essential parts of the manuscript
All authors contributed to the final version of the manuscript and approved it.
Appendix
Thin-plate spline
The thin-plate spline is the two-dimensional analogue to the cubic spline in one dimension [42,71]. Let vi denote one of the error model coefficients, e.g. intercept, at a target location (xi, yi). A thin-plate spline f(x, y) is a smooth function which interpolates a surface that is fixed at the landmark points Pi = (xi, yi) at a specific height hi A thin-plate spline interpolation function can be written as
where U(r) = r2 ln(r) is the radial basis function with . This equation is used to predict an unknown v for location (x, y), and is the unique solution [42,71] which minimises the equation:
This quantity was called the bending energy of the thin-plate spline function. If noise in the determined coefficients vi is detected, one may wish to relax the exact interpolation requirement (Equation 14). This can be accomplished by multiplying equation 14 with a regularization parameter λ, a positive scalar, and by adding the residual sum of squares, which gives:
Again, as in case of the cubic smoothing spline with the parameter λ, the degree of smoothing can be determined. In our study, we utilised an implementation of the TPS [72], according to Doug Nychka [53].
Acknowledgements
We would like to thank the members of Algorithmic Bioinformatics group at FU-Berlin for valuable discussion, especially Dr. Clemens Gröpl. We would like to thank Dr. Johan Gobom, Dr. Patrick Giavalisco and Thomas Kreitler for providing the PMF-MS data and for valuable discussion. We thank Carole Procter, Stale Nygard, Richard Boys and Daniel Henderson for proofreading the manuscript. We thank Prof. Dr. Hans Lehrach, at whose department part of the work was performed. This project was funded by the National Genome Research Network (NGFN) of the German Ministry for Education and Research (BMBF), and the Max Planck Society.
==== Refs
Gevaert K Vandekerckhove J Protein identification methods in proteomics Electrophoresis 2000 21 1145 54 10786887 10.1002/(SICI)1522-2683(20000401)21:6<1145::AID-ELPS1145>3.0.CO;2-Z
Kaltschmidt E Wittmann HG Ribosomal proteins. XII. Number of proteins in small and large ribosomal subunits of Escherichia coli as determined by two-dimensional gel electrophoresis Proc Natl Acad Sci USA 1970 67 1276 82 4922286
O'Farrell PH High resolution two-dimensional electrophoresis of proteins J Biol Chem 1975 250 4007 21 236308
Klose J Kobalz U Two-dimensional electrophoresis of proteins: an updated protocol and implications for a functional analysis of the genome Electrophoresis 1995 16 1034 59 7498127 10.1002/elps.11501601175
Blackstock W Weir M Proteomics: quantitative and physical mapping of cellular proteins Trends Biotech 1999 17 121 127 10.1016/S0167-7799(98)01245-1
Quadroni M James P Proteomics and automation Electrophoresis 1999 20 664 677 10344232 10.1002/(SICI)1522-2683(19990101)20:4/5<664::AID-ELPS664>3.0.CO;2-A
Nordhoff E Egelhofer V Giavalisco P Eickhoff H Horn M Przewieslik T Theiss D Schneider U Lehrach H Gobom J Large-gel two-dimensional electrophoresis-matrix assisted laser desorption/ionization-time of flight-mass spectrometry: an analytical challenge for studying complex protein mixtures Electrophoresis 2001 22 2844 2855 [(eng)]. 11565779 10.1002/1522-2683(200108)22:14<2844::AID-ELPS2844>3.0.CO;2-7
Tanaka K Waki H Ido Y Akita S Yoshida Y Yoshida T Matsuo T Protein and polymer analyses up to m/z 100 000 by laser ionization time-of-flight mass spectrometry Rapid Communications in Mass Spectrometry 1988 2 151 153 10.1002/rcm.1290020802
Karas M Hillenkamp F Laser Desorption Ionization of proteins with molecular masses exceeding 10 000 daltons Anal Chem 1988 60 2299 2301 3239801 10.1021/ac00171a028
Fenyo D Identifying the proteome: software tools Current Opinion in Biotechnology 2000 11 391 395 10975459 10.1016/S0958-1669(00)00115-4
Griffin TJ Aebersold R Advances in proteome analysis by mass spectrometry J Biol Chem 2001 276 45497 500 11585843 10.1074/jbc.R100014200
Patterson SD Data analysis-the Achilles heel of proteomics Nat Biotechnol 2003 21 221 2 12610558 10.1038/nbt0303-221
Aebersold R Mann M Mass spectrometry-based proteomics Nature 2003 422 198 207 12634793 10.1038/nature01511
Lottspeich F Eckerskorn C Internal amino acid sequence analysis of proteins separated by gel electrophoresis after tryptic digestion in polyacrylamide matrix Chromatographia 1989 92 94
Mann M Hojrup P Roepstorff P Use of mass spectrometric molecular weight information to identify proteins in sequence databases Biol Mass Spectrom 1993 22 338 345 8329463 10.1002/bms.1200220605
Pappin DJC Hojrup P Bleasby AJ Rapid identification of proteins by peptide-mass fingerprinting Curr Biol 1993 3 327 332 15335725 10.1016/0960-9822(93)90195-T
Colby SM King TB Reilly JP Improving the Resolution of MALDI TOF Mass Spectrometry by Exploiting the Correlation Between Ion Position and Velocity Rapid Comm Mass Spectrom 1994 8 865 868 10.1002/rcm.1290081102
Whittal RM Li L High-resolution matrix-assisted laser desorption/ionization in a linear time-of-flight mass spectrometer Anal Chem 1995 67 1950 4 8694244 10.1021/ac00109a007
Brown RS Lennon JJ Mass resolution improvement by incorporation of pulsed ion extraction in a matrix-assisted laser desorption/ionization linear time-of-flight mass spectrometer Anal Chem 1995 67 1998 2003 8694246 10.1021/ac00109a015
Takach EJ Hines WM Patterson DH Juhasz P Falick AM Vestal ML Martin SA Accurate mass measurements using MALDI-TOF with delayed extraction J Protein Chem 1997 16 363 9 9246615 10.1023/A:1026376403468
Fenn J Mann M Meng C Wong S Whitehouse C Electrospray Ionization for Mass Spectrometry of Large Biomolecules Science 1989 246 64 71 2675315
Guilhaus M Principles and Instrumentation in Time-of-Flight Mass Spectrometry JOURNAL OF MASS SPECTROMETRY 1995 30 1519 1532 10.1002/jms.1190301102
Gras R Muller M Gasteiger E Gay S Binz PA Bienvenut W Hoogland C Sanchez JC Bairoch A Hochstrasser DF Appel RD Improving protein identification from peptide mass fingerprinting through a parameterized multi-level scoring algorithm and an optimized peak detection Electrophoresis 1999 20 3535 3550 [(eng)]. 10612280 10.1002/(SICI)1522-2683(19991201)20:18<3535::AID-ELPS3535>3.0.CO;2-J
Wool A Smilansky Z Precalibration of matrix-assisted laser desorption/ionization-time of flight spectra for peptide mass fingerprinting Proteomics 2002 2 1365 1373 12422354 10.1002/1615-9861(200210)2:10<1365::AID-PROT1365>3.0.CO;2-9
Strittmatter EF Rodriguez N Smith RD High mass measurement accuracy determination for proteomics using multivariate regression fitting: application to electrospray ionization time-of-flight mass spectrometry Anal Chem 2003 75 460 8 12585471 10.1021/ac026057g
Samuelsson J Dalevi D Levander F Rognvaldsson T Modular, scriptable, and automated analysis tools for high-throughput peptide mass fingerprinting Bioinformatics 2004 20 3628 3635 15297302
Apweiler R Bairoch A Wu CH Protein sequence databases Curr Opin Chem Biol 2004 8 76 80 15036160 10.1016/j.cbpa.2003.12.004
Pappin J Hojrup P Bleasby A Rapid Identification of Proteins by Peptide-Mass Fingerprinting Current Biology 1993 3 327 332 15335725 10.1016/0960-9822(93)90195-T
Zhang W Chait BT ProFound: an expert system for protein identification using mass spectrometric peptide mapping information Anal Chem 2000 72 2482 2489 10857624 10.1021/ac991363o
Eriksson J Fenyo D A Model of random mass-matching and its use for automated significance testing in mass spectrometric proteome analysis Proteomics 2002 2 262 70 11921442 10.1002/1615-9861(200203)2:3<262::AID-PROT262>3.0.CO;2-W
Parker KG Scoring methods in MALDI peptide mass fingerprinting: ChemScore, and the ChemApplex program J Am Soc Mass Spectrom 2002 13 22 39 11777197 10.1016/S1044-0305(01)00320-8
Tabb DL Huang Y Wysocki VH Yates JRr Influence of basic residue content on fragment ion peak intensities in low-energy collision-induced dissociation spectra of peptides Anal Chem 2004 76 1243 8 14987077 10.1021/ac0351163
Pevzner PA Dancik V Tang CL Mutation-Tolerant Protein Identification by Mass Spectrometry Journal of Computational Biology 2000 7 777 787 11382361 10.1089/10665270050514927
Egelhofer V Gobom J Seitz H Giavalisco P Lehrach H Nordhoff E Protein identification by MALDI-TOF-MS peptide mapping: A new strategy Analytical Chemistry 2002 74 1760 1771 11985306 10.1021/ac011204g
Schuerenberg M Luebbert C Eickhoff H Kalkum M Lehrach H Nordhoff E Prestructured MALDI-MS sample supports Anal Chem 2000 72 3436 42 10952524 10.1021/ac000092a
Gobom J Mueller M Egelhofer V Theiss D Lehrach H Nordhoff E A calibration method that simplifies and improves accurate determination of peptide molecular masses by MALDI-TOF MS Anal Chem 2002 74 3915 3923 [(eng)]. 12175185 10.1021/ac011203o
Bantscheff M Duempelfeld B Kuster B An improved two-step calibration method for matrix-assisted laser desorption/ionization time-of-flight mass spectra for proteomics Rapid Commun Mass Spectrom 2002 16 1892 5 12271455 10.1002/rcm.798
Moskovets E Chen HS Pashkova A Rejtar T Andreev V Karger BL Closely spaced external standard: a universal method of achieving 5 ppm mass accuracy over the entire MALDI plate in axial matrix-assisted laser desorption/ionization time-of-flight mass spectrometry Rapid Commun Mass Spectrom 2003 17 2177 87 14515315 10.1002/rcm.1158
Chamrad DC Koerting G Gobom J Thiele H Klose J Meyer HE Blueggel M Interpretation of mass spectrometry data for high-throughput proteomics Anal Bioanal Chem 2003 376 1014 22 12845399 10.1007/s00216-003-1995-x
Levander F Rognvaldsson T Samuelsson J James P Automated methods for improved protein identification by peptide mass fingerprinting Proteomics 2004 4 2594 601 15352234 10.1002/pmic.200300804
Wolski WE Lalowski M Martus P Herwig R Giavalisco P Sickmann A Lehrach H Gobom J Reinert K Transformation and other factors of the biological Mass Spectrometry pairwise peak-list Comparison Process BMC Bioinformatics 2005
Bookstein F Principal Warps: Thin-Plate Splines and the Decomposition of Deformations IEEE Transactions on Pattern Analysis and Machine Intelligence 1989 11 567 585 10.1109/34.24792
Giavalisco P Nordhoff E Kreitler T Kloppel KD Lehrach H Klose J Gobom J Proteome analysis of Arabidopsis thaliana by two-dimensional gel electrophoresis and matrix-assisted laser desorption/ionisation-time of flight mass spectrometry Proteomics 2005 5 1902 13 15815986 10.1002/pmic.200401062
Bruker Daltonics – enabling life science tools based on mass spectrometry 2004
Glockner FO Kube M Bauer M Teeling H Lombardot T Ludwig W Gade D Beck A Borzym K Heitmann K Rabus R Schlesner H Amann R Reinhardt R Complete genome sequence of the marine planctomycete Pirellula sp. strain 1 Proc Natl Acad Sci USA 2003 100 8298 303 12835416 10.1073/pnas.1431443100
Pruitt KD Tatusova T Maglott DR NCBI Reference Sequence project: update and current status Nucleic Acids Res 2003 31 34 7 12519942 10.1093/nar/gkg111
Thiede B Lamer S Mattow J Siejak F Dimmler C Rudel T Jungblut PR Analysis of missed cleavage sites, tryptophan oxidation and N-terminal pyroglutamylation after in-gel tryptic digestion Rapid Commun Mass Spectrom 2000 14 496 502 10717661 10.1002/(SICI)1097-0231(20000331)14:6<496::AID-RCM899>3.0.CO;2-1
Gay S Binz PA Hochstrasser DF Appel RD Modeling peptide mass fingerprinting data using the atomic composition of peptides Electrophoresis 1999 20 3527 3534 [(eng)]. 10612279 10.1002/(SICI)1522-2683(19991201)20:18<3527::AID-ELPS3527>3.0.CO;2-9
Schmidt F Schmid M Jungblut PR Mattow J Facius A Pleissner KP Iterative data analysis is the key for exhaustive analysis of peptide mass fingerprints from proteins separated by two-dimensional electrophoresis J Am Soc Mass Spectrom 2003 14 943 56 12954163 10.1016/S1044-0305(03)00345-3
Schrijver A Combinatorial Optimization – Polyhedra and Efficiency 2003 Berlin: Springer-Verlag
Härdle W Simar L Applied Multivariate Statistical Analysis 2003 Springer, Heidelberg
Handl A Multivariate Analysemethoden – Theorie und Praxis multivariater Verfahren unter besonderer Berücksichtigung von S-PLUS 2003 Springer, Heidelberg
Nychka D fields – A collection of programs based in [R,S] for curve and function fitting with an emphasis on spatial data 2004
Gobom J Mueller M Egelhofer V Theiss D Lehrach H Nordhoff E A Calibration Method that Simplifies and Improves Accurate Determination of Peptide Molecular Masses by MALDI-TOF-MS Analytical Chemistry 2002 74 3915 3923 12175185 10.1021/ac011203o
Perkins DN Pappin DJ Creasy DM Cottrell JS Probability-based protein identification by searching sequence databases using mass spectrometry data Electrophoresis 1999 20 3551 3567 10612281 10.1002/(SICI)1522-2683(19991201)20:18<3551::AID-ELPS3551>3.0.CO;2-2
R for Proteomics
R Development Core Team R: A language and environment for statistical computing 2004 R Foundation for Statistical Computing, Vienna, Austria [ISBN 3-900051-00-3].
Gentleman RC Carey VJ Bates DM Bolstad B Dettling M Dudoit S Ellis B Gautier L Ge Y Gentry J Hornik K Hothorn T Huber W Iacus S Irizarry R Li FLC Maechler M Rossini AJ Sawitzki G Smith C Smyth G Tierney L Yang JYH Zhang J Bioconductor: Open software development for computational biology and bioinformatics Genome Biology 2004 5 R80 15461798 10.1186/gb-2004-5-10-r80
Bioconductor – open source software for bioinformatics 2004
Leisch F Hornik K, Leisch F, Zeileis A Sweave and Beyond: Computations on Text Documents Proceedings of the 3rd International Workshop on Distributed Statistical Computing 2003 Technische Universität Wien, Vienna, Austria
Lee K Bae D Lim D Evaluation of parameters in peptide mass fingerprinting for protein identification by MALDI-TOF mass spectrometry Mol Cells 2002 13 175 84 12018838
Härdle W Müller M Sperlich S Werwatz A Nonparametric and Semiparametric Models – An Introduction 2004 Springer, Heidelberg
Kreitler T Oral Communication 2003
Chambers JM Chambers J, Hastie T Linear models Statistical Models in S 1992 Wadsworth & Brooks/Cole
Venables WN Ripley BD Modern Applied Statistics with S 2002 4 Springer-Verlag New York Inc [ISBN 0-387-95457-0].
Gobom J Schürenberg M Mueller M Theiss D Lehrach H Nordhoff E alpha-cyano-4-hydroxycinnamic acid affinity sample preparation. A protocol for MALDI-MS peptide analysis in proteomics Analytical Chemistry 2001 73 434 438 11217742 10.1021/ac001241s
Chambers JM Hastie TJ Statistical Models in S 1992 London: Chapman & Hall
Hastie T Tibshirani R Generalized Additive Models 1990 Chapman and Hall
Cleveland W Grosse E Shyu W Chambers J, Hastie T Local Regression Models Statistical Models in S 1992 Wadsworth & Brooks/Cole
Hastie T Tibshirani R Friedman J The Elements of Statistical Learning 2001 Springer [ISBN:0387952845].
Donato G Belongie S Heyden A, Sparr G, Nielsen M, Johansen P Approximation Methods for Thin Plate Spline Mappings and Principal Warps Computer Vision – ECCV 2002: 7th European Conference on Computer Vision, Copenhagen, Denmark, May 28–31, 2002 Proceedings, Part III, Lecture Notes in Computer Science 2002 Springer-Verlag Heidelberg 21 31
Green P Silverman B Nonparametric Regression and Generalized Linear Modes: A Roughness Penalty Approach 1994 Chapman and Hall
|
16102175
|
PMC1199585
|
CC BY
|
2021-01-04 16:27:27
|
no
|
BMC Bioinformatics. 2005 Aug 15; 6:203
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-203
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2091612238510.1186/1471-2105-6-209SoftwareFiatFlux – a software for metabolic flux analysis from 13C-glucose experiments Zamboni Nicola [email protected] Eliane [email protected] Uwe [email protected] Institute of Biotechnology, ETH Zurich, 8093 Zurich, Switzerland2005 25 8 2005 6 209 209 11 3 2005 25 8 2005 Copyright © 2005 Zamboni et al; licensee BioMed Central Ltd.2005Zamboni et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Quantitative knowledge of intracellular fluxes is important for a comprehensive characterization of metabolic networks and their functional operation. In contrast to direct assessment of metabolite concentrations, in vivo metabolite fluxes must be inferred indirectly from measurable quantities in 13C experiments. The required experience, the complicated network models, large and heterogeneous data sets, and the time-consuming set-up of highly controlled experimental conditions largely restricted metabolic flux analysis to few expert groups. A conceptual simplification of flux analysis is the analytical determination of metabolic flux ratios exclusively from MS data, which can then be used in a second step to estimate absolute in vivo fluxes.
Results
Here we describe the user-friendly software package FiatFlux that supports flux analysis for non-expert users. In the first module, ratios of converging fluxes are automatically calculated from GC-MS-detected 13C-pattern in protein-bound amino acids. Predefined fragmentation patterns are automatically identified and appropriate statistical data treatment is based on the comparison of redundant information in the MS spectra. In the second module, absolute intracellular fluxes may be calculated by a 13C-constrained flux balancing procedure that combines experimentally determined fluxes in and out of the cell and the above flux ratios. The software is preconfigured to derive flux ratios and absolute in vivo fluxes from [1-13C] and [U-13C]glucose experiments and GC-MS analysis of amino acids for a variety of microorganisms.
Conclusion
FiatFlux is an intuitive tool for quantitative investigations of intracellular metabolism by users that are not familiar with numerical methods or isotopic tracer experiments. The aim of this open source software is to enable non-specialists to adapt the software to their specific scientific interests, including other 13C-substrates, labeling mixtures, and organisms.
==== Body
Background
Genome-wide measurements of cellular mRNA, protein or metabolite concentrations (or their differential concentrations) are current workhorse technologies in functional genomics and systems biology. For a comprehensive analysis of metabolic networks, however, typically also knowledge on the molecular traffic between the metabolites is necessary. These time-dependent in vivo fluxes are the functional complement to the metabolite concentrations, but, in contrast to the concentrations, cannot be detected directly [1]. Instead, intracellular fluxes must be inferred indirectly from measurable quantities, such as nutrient uptake and secretion rates and/or 13C-labeling pattern, through methods of metabolic flux analysis [2,3].
To reliably identify a unique distribution of intracellular fluxes, highly controlled culture conditions, extensive physiological, and 13C-data are a prerequisite [2]. Although many laboratories have access to the necessary instrumentation, flux analysis remained largely restricted to a handful of expert groups because flux quantification required the simultaneous interpretation of physiological and 13C-data. Briefly, complicated isotopomer models of metabolism were used to balance the labeling state of metabolic intermediates or protein-bound amino acids and to identify a best fit of intracellular fluxes to the available data. Several (non-open source) software tools for flux analysis with isotopomer models of varying complexity are available for academic research [4-6], with 13C-FLUX as the probably most advanced one [7]. Furthermore, software tools for automated processing of raw MS [8,9] or NMR data for flux anaylsis are available [10], in the latter case allowing also to calculate flux ratios. Although valuable biological insights can be obtained by isotopomer balancing [11-16], the required expertise in computational analysis and quantitative biology as well as the complexity of the models restricted broader application and wider use as a routine tool.
A conceptual simplification of flux analysis and an appropriate analytical throughput was obtained by splitting the problem in two separate tasks. Firstly, MS-detected 13C data are analytical interpreted with probabilistic equations that quantify flux partitioning ratios in so-called metabolic flux ratio analysis [17], akin to an earlier NMR-based approach [18]. In the second step, these flux ratios are used as constraints for a flux balancing calculation in a comparatively simple metabolic network model to estimate absolute intracellular fluxes from the measured extracellular fluxes [19,20]. For non-expert users, the major advantage of this 13C-constrained flux balancing is the relative simplicity of the employed models, rapid computation, and a more intuitive data treatment. This also allows to simplify the experimental set-up because the flux ratios are calculated from MS data exclusively. Hence, simple shake flask experiments suffice for standard analyses – although at the cost of flux resolution – thus restricting the use of laborious bioreactor experiments to specific applications. Intuitively, less data suggest less reliable flux estimates, which indeed would be the case if an isotopomer models was used. However, since the flux ratios are analytically determined in a strictly local data interpretation and not in a global fitting procedure, most ratios are from independent measurements and can partly validate each other. For a more comprehensive treatise of flux ratio and net flux analysis please see [3,14,19,21]. Recently, 13C-constrained flux balancing was successfully applied to various microorganisms [22-25] and was also the key methodology for higher-throughput flux analyses in our lab [22,26,27].
Based on these conceptual advances, the availability of a user-friendly and robust software for flux analyses becomes the major limitation for wider use. Here we describe the open-source software package FiatFlux that consists of two separate modules for analytical metabolic flux ratio analysis and for 13C-constrained flux analysis. FiatFlux condenses our accumulated knowhow and experience on metabolic flux analysis, and was used successfully for teaching and in collaborations with biologically-oriented groups.
Implementation
We developed the FiatFlux software on a Matlab basis to exploit the Optimization toolbox and the open source environment. FiatFlux consists of two parts with distinct functions: (i) computation of metabolic flux ratios exclusively from MS data in the RATIO module and (ii) estimation of net carbon fluxes within a comprehensive model of metabolite balances from measured extracellular fluxes, previously determined flux ratios, and biomass requirements in the NETTO module. The two modules are run independently, calling either the functions ratio.m or netto.m, respectively.
The RATIO module affords the integration of raw MS data that are passed to the software using the netCDF standard (network Common Data Form) [28] (Figure 1). This format was chosen because it is supported by the proprietary software of most mass spectrometer manufacturers. From a netCDF file, FiatFlux generates a matrix with the total ion counts for each scan (timepoint) and considered m/z value, and searches automatically for known compounds based on their predefined fragmentation pattern. For each recognized analyte, a mass isotopomer distribution vector MDVα is extracted from the matrix and normalized such that
Figure 1 Procedure for derivation of metabolic flux ratios from raw MS data in RATIO (see text for details). For each stage of the analysis, exemplary data and corresponding computation time in seconds are shown on the right and the left, respectively. Times were measured for the analysis of a GC-MS sample on a Pentium 4 1.6 GHz processor.
where m0 is the fractional abundance of molecules with monoisotopic mass and mi>0 the abundances of fragments with heavier masses. The mass isotope distribution vector specific to the carbon backbone (MDVA) is obtained from MDVα upon correction (a) for naturally occurring isotopes of O, N, H, P, S, Si, and C atoms in the derivatization agent [29] and (b) for the presence of unlabeled biomass in the sample, e.g. the inoculum [17]. The MDVA are, in turn, used to estimate by least square fitting the mass distribution of their precursors (MDVM) in central carbon metabolism [17], along with covariance matrices for each MDVM, which are calculated from the experimental error (i. e. comparison of the MDVA of fragments with identical carbon skeletons). Faulty MDVα measurements are diagnosed by visual inspection of the residuals that result for each MDVA in the MDVM fitting. In the case of uniformly labeled tracer experiments, diagnosis is based on the fractional labeling of MDVA (and MDVM) that should equal that of the substrate [30]. Finally, the flux ratios are calculated from the MDVM with probabilistic equations [17]. Standard deviations for each flux ratio are calculated using the covariance matrices of MDVM by applying the Gaussian law of error propagation [17]. For a more complete treatise of the mathematical/analytical background and the experimental protocols please refer to [30]. User monitoring and intervention is possible at every stage from the graphical user interface (Figure 2).
Figure 2 The main window of RATIO. Upon loading of MS data, analytes are first automatically recognized and assigned (A). When necessary, manual assignment of analytes is performed in a different window. The experimental parameters are set by the user (B), then MDVM (C) and flux ratios (D) are calculated. Abnormal residuals indicate that the corresponding fragments are outlier, and they can be excluded (white) or reactivated (blue) by a single mouse click on the corresponding bar. Finally, the flux ratios, the MDVA and the MDVM are exported to a text file or Excel workbook (E).
The set of calculable flux ratios is a function of the biochemical reaction network, the carbon substrates and their corresponding 13C-labeling, and the analyte fragments that can be detected by MS. The software is preconfigured to derive metabolic flux ratios for a variety of microorganisms such as yeasts [31,32], Escherichia coli [17], Bacillus subtilis [23], and others [25] for growth on [1-13C]glucose, [U-13C]glucose or mixtures thereof. The preconfigured analytes are the TBDMSTFA-derivatized proteinogenic amino acids that are typically detected by robust GC-MS analysis [33]. Notably, FiatFlux is not limited to glucose substrates and can be extended to cope with additional analytes, different derivatization agents or separations, i.e. liquid chromatography or capillary electrophoresis.
The second module (NETTO) addresses the estimation of absolute in vivo (net) fluxes through a reaction network. This is achieved by global material balances derived from a stoichiometric model and accounting for the withdrawal of precursors during growth (Figure 3). Because of the presence of redundant or interconnected pathways, this system of linear contraints is typically underdetermined [34]. In so-called 13C-constrained flux balancing [19,20], additional linearly independent constraints are obtained from the experimentally determined flux ratios in the RATIO module that are used to solve the system without further assumptions on energy or redox balances. NETTO features a platform to integrate metabolite balances and 13C-derived equality or inequality constraints; i.e. flux ratios that are exactly determined or for which only reaction bounds are available, respectively [19]. Depending on the active set of constraints and reactions, the system may either be underdetermined, determined, or overly constrained. In underdetermined system, NETTO offers either to search within the solution space for the flux distribution that maximizes a particular flux or the product of an intermediate, or estimate all calculable fluxes using the procedure outlined by Klamt et al. [35]. Exactly determined and overly constrained systems are both solved by a least square optimization using Matlab fmincon function. This approach permits to simultaneously integrate equality and reaction bound constraints in the calculation, and weight the constraints with the experimental uncertainty [19]. Confidence intervals for each calculated flux are estimated as a function of the experimental errors from the Jacobian matrix of the output function. Inequality constraints (reaction bounds), only contribute to the error criterion if the flux solution would otherwise exceed the upper or lower bounds set by the flux ratio data. This asymmetrical error consideration is described elsewhere [19]. If the boundary constraint is inactive, the confidence interval of the resulting flux (e.g. malic enzyme), is a result of the (stoichiometric) dependence on other fluxes.
Figure 3 Schematic representation of the analysis workflow in NETTO.
In NETTO, metabolic models can be constructed from scratch and error-prone operations such as introduction or modification of reactions are executed by the software. In a text file, the user provides a list of reactions, ratios, and biomass composition with a user-friendly syntax (Figure 4). The information is then automatically translated into properly formatted structures and matrices and saved in a Matlab m-file, that is called by NETTO during computation. The graphical user interface of NETTO permits to freely remove a reaction or modify its reversibility, submit extracellular fluxes or metabolic parameters such as the P/O respiratory coupling, or define which metabolites have to be excluded from balancing, for example oxygen or ATP (Figure 5). Alternatively, default preferences can be defined in the saved model m-file. Whenever a session of RATIO is running in parallel, NETTO imports the value for matching flux ratios.
Figure 4 Example of syntax for definition of a model for NETTO. (A) Reactions are described with a unique identifier, educts, products and an operator to set reversibility. (B) Ratios are entered either as equality constraints (=), upper bounds (>), or lover bounds (<), and are defined using the reaction indentifiers. (C) Precursor requirement for biomass formation is expressed with a list of growth-rate dependent withdrawals of metabolites in μmol/gCDW. Separate statements are used for each macromolecular class such as protein, DNA, etc.
Figure 5 The graphical user interface of NETTO.
Both modules offer functions to save all variables and recover work at a later point. Results are visualized directly on the graphical user interface and can be stored to text files or to Microsoft Excel spreadsheets.
Results and discussion
FiatFlux is the first publicly available software for flux ratio analysis from MS data and, consequently, no comparison can be done with other programs. The scientific value and accuracy of FiatFlux-calculated flux ratios has already been discussed extensively [14,17,25,26,36,37], and consistency between net flux estimates obtained either with 13C-constrained flux balancing as in FiatFlux or with global isotopomer balancing was demonstrated previously [19]. Notably, both the calculation of flux ratios from raw MS data in RATIO and the estimation of net fluxes in NETTO is typically completed in a few seconds (Figure 1). This constitutes a major advantage compared to isotopomer balancing, since computation time becomes negligible in relation to the time required by the user to set the experimental parameters. In addition, interpretation of MS data and the integration with measured fluxes are executed independently in FiatFlux. In contrast to methods of isotopomer balancing, this enables the user to discern problems arising from bad measurements or from incomplete metabolic models.
In FiatFlux, user supervision is necessary only when MS-signals are low, saturated, or overlapping. This affects the ion statistics of the corresponding fragment and results in relatively high residuals after inferring MDVM from the MDVA. Since the residuals are graphically represented on the graphic user interface of RATIO, bad fragments are rapidly identified and excluded with a single click. Also when the quality of the fragments has to be diagnosed in detail, and MDVM fitting and flux ratios estimation have therefore to be iterated several times, a correct estimate is obtained within some minutes. Using FiatFlux, a typical user with moderate experience will be able to determine intracellular net fluxes for hundreds of samples per day from previously generated MS data.
The open source nature of FiatFlux, and in particular of the RATIO module, permits to modify and extend its capabilities beyond the predefined features. Although the necessary skills strongly depend on the functionalities to be modified, fundamental biochemical knowledge of the reactions investigated is paramount for every user to understand initial assumptions and critically interpret outcomes. Provided that metabolism of a new organisms to be investigated is similar to that of any of the 4 preconfigured models,, very few adaptations are necessary and the task is manageable by any biochemically-trained biologist. In fact, in previous works we already demonstrated the analysis of about 20 different species with the 4 core models [25,32]. The implementation of new flux ratios or new substrates, however, requires detailed information on mapping of atoms in biochemical pathways, understanding of error propagation, and advanced experience with Matlab syntax, thus is probably limited to experts. Hence, at this stage, we decided to restrict free modification of the preconfigured models by precompiling the corresponding routine. In case a user requires extensions, we encourage to contact the authors to collaborate on a proper integration that ensures correct estimation of metabolic flux ratios and confidence intervals.
Finally, introduction of new GC methods or derivatization procedures is very simple, and can be attained by users with basic familiarity with the Matlab environment. In principle, the same applies to implementing other separation techniques, such liquid-phase systems. Currently, RATIO is not compatible with MS/MS product ion scans.
Conclusion
FiatFlux condenses the know-how developed over years in our lab and has become our workhorse for quantitative analyses of microbial central carbon metabolism. The software is preconfigured for the most widely used substrate (glucose), the most frequently used (and informative) tracer mixtures, and several model microbes. While this covers about 80% of all current flux applications, it is, of course, not complete. The aim of this open source software is to enable non-specialists to adapt the software to their specific scientific interests, including other substrates and or labeling mixtures. In particular, we aim at biologists that are not familiar with numerical methods or isotopic tracer experiments. In fact, with the availability of this software, the only burden for such studies remains the access to a GC-MS instrument. We hope that this transparent and flexible framework will support further developments.
Availability
Project name: FiatFlux
Operating system: preferably Microsoft Windows. Some minor problems were encountered using Matlab's graphic user interface with Linux.
Programming language: Matlab R14 (The Mathworks).
Other requirements: Matlab Optimization Toolbox
License: source code is freely available from the authors for academic purposes.
Any restriction to use by non-academics: license required.
Abbreviations
MDVα Mass distribution vector of analyte
MDVA Carbon-specific mass distribution vector of analyte
MDVM Mass distribution vector of metabolite
TBDMSTFA N-(tert-butyldimethelsylil)-N-methyl-trifluoroacetamide
Authors' contributions
NZ and EF developed the software. US supported the work. NZ and US wrote the manuscript. All authors read and approved the final version.
Acknowledgements
We thank the members of the Sauer Lab for continuous testing.
==== Refs
Hellerstein MK In vivo measurement of fluxes through metabolic pathways: The missing link in functional genomics and pharmaceutical research Annu Rev Nutr 2003 23 379 402 12704218 10.1146/annurev.nutr.23.011702.073045
Wiechert W 13C metabolic flux analysis Metabolic Eng 2001 3 195 206 10.1006/mben.2001.0187
Sauer U High-throughput phenomics: experimental methods for mapping fluxomes Curr Opin Biotechnol 2004 15 58 63 15102468 10.1016/j.copbio.2003.11.001
Schmidt K Nielsen J Villadsen J Quantitative analysis of metabolic fluxes in Escherichia coli using two-dimensional NMR spectroscopy and complete isotopomer models J Biotechnol 1999 71 175 190 10483105 10.1016/S0168-1656(99)00021-8
Wittmann C Heinzle E Mass spectroscopy for metabolic flux analysis Biotechnol Bioeng 1999 62 739 750 10099575 10.1002/(SICI)1097-0290(19990320)62:6<739::AID-BIT13>3.0.CO;2-E
Dauner M Bailey JE Sauer U Metabolic flux analysis with a comprehensive isotopomer model in Bacillus subtilis Biotechnol Bioeng 2001 76 144 156 11505384 10.1002/bit.1154
Wiechert W Möllney M Petersen S de Graaf AA A universal framework for 13C metabolic flux analysis Metab Eng 2001 3 265 283 11461148 10.1006/mben.2001.0188
Talwar P Wittmann C Lengauer T Heinzle E Software tool for automated processing of 13C labeling data from mass spectrometric spectra Biotechniques 2003 35 1214 1215 14682055
Wahl A Dauner M Wiechert W New tools for mass isotopomer data evaluation in 13C flux analysis: mass isotope correction, data consistency checking and precursor relationships Biotechnol Bioeng 2004 85 259 268 14748080 10.1002/bit.10909
Szyperski T Glaser RW Hochuli M Fiaux J Sauer U Bailey JE Wüthrich K Bioreaction network topology and metabolic flux ratio analysis by biosynthetic fractional 13C-labeling and two-dimensional NMR spectroscopy Metabolic Eng 1999 1 189 197 10.1006/mben.1999.0116
Dauner M Storni T Sauer U Bacillus subtilis metabolism and energetics in carbon-limited and carbon-excess chemostat culture J Bacteriol 2001 183 7308 7317 11717290 10.1128/JB.183.24.7308-7317.2001
Marx A Hans S Möckel B Bathe B de Graaf AA Metabolic phenotype of phosphoglucose isomerase mutants of Corynebacerium glutamicum J Biotechnol 2003 104 185 197 12948638 10.1016/S0168-1656(03)00153-6
Gunnarsson N Mortensen UH Sosio M Nielsen J Identification of the Entner-Doudoroff pathway in an antibiotic-producing actinomycete species Mol Microbiol 2004 52 895 902 15101992 10.1111/j.1365-2958.2004.04028.x
Fischer E Sauer U A novel metabolic cycle catalyzes glucose oxidation and anaplerosis in hungry Escherichia coli J Biol Chem 2003 278 46446 46451 12963713 10.1074/jbc.M307968200
Wittmann C Heinzle E Genealogy profiling through strain improvement by using metabolic network analysis: metabolic flux genealogy of several generations of lysine-producing Corynebacteria Appl Environ Microbiol 2002 68 5843 5859 12450803 10.1128/AEM.68.12.5843-5859.2002
Hua Q Yang C Baba T Mori H Shimizu K Responses of the central carbon metabolism in Escherichia coli to phosphoglucose isomerase and glucose-6-phosphate dehydrogenase knockouts J Bacteriol 2003 185 7053 7067 14645264 10.1128/JB.185.24.7053-7067.2003
Fischer E Sauer U Metabolic flux profiling of Escherichia coli mutants in central carbon metabolism by GC-MS Eur J Biochem 2003 270 880 891 12603321 10.1046/j.1432-1033.2003.03448.x
Szyperski T Biosynthetically directed fractional 13C-labeling of proteinogenic amino acids: an efficient analytical tool to investigate intermediary metabolism Eur J Biochem 1995 232 433 448 7556192 10.1111/j.1432-1033.1995.tb20829.x
Fischer E Zamboni N Sauer U High-throughput metabolic flux analysis based on gas chromatography-mass spectrometry derived 13C constraints Anal Biochem 2004 325 308 316 14751266 10.1016/j.ab.2003.10.036
Sauer U Hatzimanikatis V Bailey JE Hochuli M Szyperski T Wüthrich K Metabolic fluxes in riboflavin-producing Bacillus subtilis Nature Biotechnol 1997 15 448 452 9131624 10.1038/nbt0597-448
Emmerling M Dauner M Ponti A Fiaux J Hochuli M Szyperski T Wüthrich K Bailey JE Sauer U Metabolic flux responses to pyruvate kinase knockout in Escherichia coli J Bacteriol 2002 184 152 164 11741855 10.1128/JB.184.1.152-164.2002
Blank LM Kuepfer L Sauer U Large-scale 13C-flux analysis reveals mechanistic principles of metabolic network robustness to null mutations in yeast Genome Biol 2005 6 R49 15960801 10.1186/gb-2005-6-6-r49
Zamboni N Sauer U Knockout of the high-coupling cytochrome aa3 oxidase reduces TCA cycle fluxes in Bacillus subtilis FEMS Microbiol Lett 2003 226 121 126 13129617 10.1016/S0378-1097(03)00614-1
Sauer U Canonaco F Heri S Perrenoud A Fischer E The soluble and membrane-bound transhydrogenases UdhA and PntAB have divergent functions in NADPH metabolism of Escherichia coli J Biol Chem 2004 279 6613 6619 14660605 10.1074/jbc.M311657200
Fuhrer T Fischer E Sauer U Experimental identification and quantification of glucose metabolism in seven bacterial species J Bacteriol 2005 187 1581 1590 15716428 10.1128/JB.187.5.1581-1590.2005
Fischer E Sauer U Large-scale in vivo flux analysis shows rigidity and suboptimal performance of Bacillus subtilis metabolism Nat Genet 2005 37 636 640 15880104 10.1038/ng1555
Perrenoud A Sauer U Impact of global transcriptional regulation by ArcA, ArcB, Cra, Crp, Cya, Fnr and Mlc on glucose catabolism in Escherichia coli J Bacteriol 2005 187 3171 3179 15838044 10.1128/JB.187.9.3171-3179.2005
Unidata: the NetCDF library [http://wwwunidataucaredu/packages/netcdf/]
van Winden WA Wittmann C Heinzle E Heijnen JJ Correcting mass isotopomer distributions for naturally occurring isotopes Biotechnol Bioeng 2002 80 477 479 12325156 10.1002/bit.10393
Perrenoud A Fuhrer T Sauer U Determination of metabolic flux ratios from 13C-experiments and GC-MS-data: protocols and principles Meth Mol Biol 2005 in press
Blank LM Sauer U TCA cycle activity in Saccharomyces cerevisiae is a function of the environmentally determined specific growth and glucose uptake rates Microbiology 2004 150 1085 1093 15073318 10.1099/mic.0.26845-0
Blank LM Lehmbeck F Sauer U Metabolic-flux and network analysis in 14 hemiascomycetous yeasts FEMS Yeast Res 2005 5 545 558 15780654 10.1016/j.femsyr.2004.09.008
Dauner M Sauer U GC-MS analysis of amino acids rapidly provides rich information for isotopomer balancing Biotechnol Prog 2000 16 642 649 10933840 10.1021/bp000058h
Bonarius HPJ Schmid G Tramper J Flux analysis of underdetermined metabolic networks: The quest for the missing constraints Trends Biotechnol 1997 15 308 314 10.1016/S0167-7799(97)01067-6
Klamt S Schuster S Gilles ED Calculability analysis in underdetermined metabolic networks illustrated by a model of the central metabolism in purple nonsulfur bacteria Biotechnol Bioeng 2002 77 734 751 11835134 10.1002/bit.10153
Zamboni N Fischer E Muffler A Wyss M Hohmann HP Sauer U Transient expression and flux changes during a shift from high to low riboflavin production in continuous cultures of Bacillus subtilis Biotechnol Bioeng 2005 89 219 232 15584023 10.1002/bit.20338
Zamboni N Fischer E Laudert D Aymerich S Hohmann HP Sauer U The Bacillus subtilis yqjI gene encodes the NADP+-dependent 6-P-gluconate dehydrogenase in the pentose phosphate pathway J Bacteriol 2004 186 4528 4534 15231785 10.1128/JB.186.14.4528-4534.2004
|
16122385
|
PMC1199586
|
CC BY
|
2021-01-04 16:27:24
|
no
|
BMC Bioinformatics. 2005 Aug 25; 6:209
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-209
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-1011609553110.1186/1471-2407-5-101Research ArticleNo significant role for beta tubulin mutations and mismatch repair defects in ovarian cancer resistance to paclitaxel/cisplatin Mesquita Bárbara [email protected] Isabel [email protected] Deolinda [email protected] Ana [email protected] Isabel M [email protected] Carla [email protected] Manuel R [email protected] Sérgio [email protected] Department of Genetics, Portuguese Oncology Institute, 4200-072 Porto, Portugal2 Department of Medical Oncology, Portuguese Oncology Institute, 4200-072 Porto, Portugal3 Department of Pathology, Portuguese Oncology Institute, 4200-072 Porto, Portugal2005 11 8 2005 5 101 101 26 4 2005 11 8 2005 Copyright © 2005 Mesquita et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The mechanisms of chemoresistance in ovarian cancer patients remain largely to be elucidated. Paclitaxel/cisplatin combination is the standard chemotherapeutic treatment for this disease, although some patients do not respond to therapy. Our goals were to investigate whether TUBB mutations and mismatch repair defects underlie paclitaxel and cisplatin resistance.
Methods
Thirty-four patients with primary ovarian carcinomas (26 serous and eight clear cell carcinomas) treated with paclitaxel/cisplatin were analysed. TUBB exon 4 was analysed by nested PCR after a first round PCR using intronic primers. Microsatellite analysis was performed with the quasimonomorphic markers BAT 26 and BAT 34.
Results
Twenty-two of the 34 ovarian cancers (64.7%) presented residual tumour after surgery, seven of which (7/22; 31.8%) were shown to be chemoresistant (five serous and two clear cell tumours). Sequence analysis did not find any mutation in TUBB exon 4. Microsatellite instability was not detected in any of the ovarian carcinomas.
Conclusion
We conclude that TUBB exon 4 mutations and mismatch repair defects do not play a significant role in paclitaxel/cisplatin resistance.
==== Body
Background
Ovarian cancer is the fourth most common cancer in women [1]. The standard treatment for ovarian cancer is cytoreductive surgery followed by combination systemic chemotherapy [2]. Since the middle 90's, the combination paclitaxel/cisplatin became the standard chemotherapeutic treatment for poor prognosis ovarian cancer [3]. Nevertheless, some patients are resistant to this chemotherapeutic treatment, making it important to clarify the underlying mechanisms of resistance [4].
Paclitaxel binds to microtubules and causes kinetic suppression (stabilisation) of microtubule dynamics, promoting their polymerisation and cell cycle arrest in mitosis (antimitotic activity), which probably leads to apoptosis [5,6]. Microtubules are composed of a dimeric protein, tubulin, with alpha (α) and beta (β) tubulin heterodimers in dynamic equilibrium. The β-tubulin gene (TUBB), mapped to 6p21.3 [7], is composed of four exons and encodes a 445-aminoacid protein with GTPase function to which paclitaxel preferentially binds [8]. Cisplatin is activated intracellularly and establishes inter- and intra-strand DNA adducts that block replication and translation. The fate of cells after cisplatin exposure depends both on the extent of DNA damage and the cellular response to it, and apoptosis can be induced as a consequence [5,9]. Although the specific mechanism that triggers apoptosis is not totally clear, some evidence suggests that this process can be mediated by the DNA mismatch repair system (MMR) [5,9].
Drug resistance is considered a multifactorial process, but the detailed mechanisms are still unknown. Recently, point mutations in the β-tubulin gene, predominantly in exon 4, were associated with resistance to paclitaxel [10-12]. Resistance to cisplatin was linked with anomalies in the DNA MMR system resulting in microsatellite instability (MSI) [13-16]. In order to evaluate the relevance of these mechanisms to ovarian cancer chemoresistance, we screened TUBB exon 4 for mutations and performed MSI analysis in 34 ovarian carcinomas treated with paclitaxel/cisplatin and evaluated patients' response to chemotherapy.
Methods
Patient data
Thirty-four primary ovarian carcinoma patients, of serous (26 cases) or clear cell (eight cases) histological types (invasive or borderline), consecutively admitted at the Portuguese Oncology Institute – Porto and treated with the adjuvant chemotherapy scheme paclitaxel/cisplatin, were analysed. Patients previously treated with other chemotherapeutic regimens or radiotherapy were excluded from the study.
Evaluation of treatment responses was done by an oncologist using computerized tomography or magnetic resonance and CA125 quantification, according to international guidelines [17]. Investigators performing laboratory analysis were not aware of chemotherapy response or resistance until the study was completed.
DNA extraction
Genomic DNA was extracted from chemo-naïve, paraffin embedded tumours, after dissection. Tissue blocks were sectioned, mounted on glass slides, deparaffinised, and stained with haematoxylin and eosin. Tumour areas were identified under the microscope on each slide and marked. Areas with at least 70% of cancer cells were identified on tissue sections. Selected paraffin blocks were sectioned (5μm sections) and mounted in microscope slides. Marked tumour areas were selected with a sterile razor blade. Three sections of tissue were incubated in a solution of 10 mM Tris-HCl buffer (pH 8.0), 2.5 mM MgCl2, 50 mM KCl, 0.5% (p/v) Tween 20, and 1 mg/mL proteinase K for 48 hours at 55°C.
TUBB exon 4 sequencing
For polymerase chain reaction (PCR) assays, different sets of oligonucleotides were designed to amplify specific regions of TUBB exon 4 that code for the GTP and paclitaxel binding sites. To assure that the amplicon was not a pseudogene, the following intronic primer set was used in the first round PCR: 5'AAG-GAG-ATA-CAT-CCG-AGG-GAA-TT3' and 5'AAG-GTA-TTC-ATG-ATG-CGA3'. After checking for first round PCR product in an agarose gel, a 1:10 dilution was used for nested PCRs with the following primers: set 1, 5'AGA-GAG-CTG-TGA-CTG-CCT-G3' and 5'AAG-GTA-TTC-ATG-ATG-CGA3'; set 2, 5'GCT-CTG-GAA-TGG-GCA-CTC3' and 5'CCG-TAG-GTT-GGT-TGT-GGT-CA3'; set 3, 5'CGG-GGA-TCT-GAA-CCA-CCT-T3' and 5'GAG-TGT-CAC-GGC-CTG-GAG-T3'. The PCR products were separated by electrophoresis in agarose gels stained with ethidium bromide and analysed in a transiluminator. All DNA samples were analysed in an automatic DNA sequencer ABI PRISM 310™ Genetic Analyser. The sequences were compared with the genomic sequence GenBank AF070600.
MSI evaluation
For microsatellite analysis, tumour DNA was amplified with primers for two quasimonomorphic markers BAT 26 (5'GAG-TGT-CAC-GGC-CTG-GAG-T3'; 5'AAC-CAT-TCA-ACA-TTT-TTA-ACC-C3') and BAT 34 (5'ACC-CTG-GAG-GAT-TTC-ATC-TC3'; 5'AAC-AAA-GCG-AGA-CCC-AGT-CT3') [18-20]. Fragments were analysed in an ABI PRISM 310™ Genetic Analyser.
Results
Twenty-two of the 34 ovarian cancers (64.7%) presented residual tumour after surgery, seven of which (7/22; 31.8%) were shown to be chemoresistant (five serous and two clear cell tumours) (Table 1).
Amplicons of 700 bp were observed for all cases after amplification with the intronic primers (data not shown). Nested PCR with primer sets 1–3 specific for TUBB exon 4 resulted in amplicons with 129, 254, and 201 bp, respectively (Figure 1). Sequencing analysis of all 34 cases showed no TUBB exon 4 mutations (Figure 2).
Microsatellite analysis with markers BAT 26 and BAT 34 (Figure 3) showed the normal pattern in all cases, so no evidence for microsatellite instability was detected in this series.
Discussion
Tumour resistance to chemotherapy or disease relapses resistant to further treatment after an initial response are common events in current cytotoxic cancer treatment regimens [5]. With regard to ovarian cancer chemotherapy, the current major challenge is to understand why histologically similar tumours behave so differently when treated with the same chemotherapeutic regimen. The action of a drug potentially depends on several mechanisms, namely, metabolisation, access into the tumour microenvironment, intracellular uptake, interaction with the target, and subsequent signalling events [5]. It is therefore important to study the different molecular mechanisms that can be involved in chemotherapy resistance.
The rationale for studying the relationship between TUBB gene mutations with paclitaxel resistance came from the studies of Giannakakou et al and Gonzalez-Garay et al [10,11], who found TUBB mutations in ovarian cancer cell lines and in hamster cells, respectively. Subsequently, Monzó et al [12] reported TUBB mutations in 16 out of 36 (44.4%) paclitaxel resistant tumour samples from patients with advanced non-small cell lung cancer and proposed that TUBB mutations could represent a possible mechanism of paclitaxel resistance in that tumour type.
However, our findings in the present study argue against a significant role of TUBB gene mutations in paclitaxel resistance in ovarian cancer. In keeping with our results, Sale et al [22] and Lamendola et al [23] did not detect TUBB gene mutations in ovarian cancer samples. Similar findings have recently been obtained for several other tumour types, namely, lung [24], breast [25,26], and gastric cancer [27]. Taken together, our and several other investigations concur that TUBB exon 4 mutations are not an important mechanism underlying paclitaxel resistance. To explain the early findings of Monzó et al [12], which were not reproduced by other authors, including the present study, Kelley et al [24] suggested that the primers used by Monzó et al [12] did not allow to discriminate TUBB from its pseudogenes [24]. Furthermore, the first studies reporting TUBB gene mutations were made in hamster cells [11] and in ovarian cancer cell lines [10] after selection by paclitaxel exposure, something that makes difficult a direct extrapolation of these findings to human tumours.
We have also evaluated the MSI status of the 34 ovarian carcinomas with quasimonomorphic BAT 26 and BAT 34 markers, but did not find any microsatellite unstable tumours. Some authors have described an association between cisplatin resistance and MMR system anomalies in ovarian adenocarcinomas [28-30], as well as in colon cancer cell lines [31-33]. Additionally, other studies found MSI in ovarian cancer, namely, in serous and clear cell histological types [34-40]. However, the frequency of MSI identified in those studies was quite low (0–14.3%), which is in agreement with our findings. To completely rule out any relationship between deficient mismatch repair and cisplatin resistance, one would have to analyse more microsatellite markers in a larger series of tumours paired with normal DNA.
Conclusion
We conclude that, contrarily to earlier suggestions, TUBB exon 4 mutations and MMR defects are not major mechanisms underlying paclitaxel and/or cisplatin resistance in ovarian cancer. Further investigation on alternative mechanisms of resistance to these drugs is warranted. Possible mechanisms to paclitaxel resistance are P-glycoprotein overexpression [5,41], differential β-tubulin isotype expression [5,42,43], and apoptosis deregulation [5,44-46]. Decrease in intracellular cisplatin level, increase of tolerance or repair of DNA lesions, and alterations in the apoptotic cascade have also been related with cisplatin resistance [5,47]. These studies are necessary to predict individual response of patients to these chemotherapeutic agents.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
BM carried out the molecular genetic studies and drafted the manuscript. IV participated in the design of the study and coordination. DP was responsible for clinical surveillance. AS was responsible for paraffin slides. IMP was responsible for the histopathologic analysis. CP carried out sequence alignment. MT coordinated the study. SC conceived the study and participated in its design and coordination.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We thank the Portuguese Health Ministry for the financial support (Project nr. 191/99). B. Mesquita was a fellow of Fundação para Ciência e Tecnologia. We thank Mariano Monzó and collaborators for useful discussions during this work.
Figures and Tables
Figure 1 Nested PCR products (in duplicate) obtained with the three primer sets specific for TUBB exon 4 in an ovarian carcinoma (from right to left, lane 1: 100 bp step ladder; lanes 2 and 3: 129 bp amplicon – set 1 primers; lanes 4 and 5: 254 bp amplicon – set 2 primers; lanes 6 and 7: 201 bp amplicon – set 3 primers).
Figure 2 Electrophorogram of part of TUBB exon 4 without any sequence variation in an ovarian carcinoma.
Figure 3 Electrophorograms of BAT 26 (A) and BAT 34 (B) markers in an ovarian carcinoma, showing the same pattern found in normal control DNA.
Table 1 Clinical, pathological and genetic data of 34 ovarian cancer patients.
Patient
Histological type
Stage
Grade
Residual tumour
Treatment response
Exon 4 TUBB mutation
MSI status
1 Serous III 1 > 2 cm CR Not present Stable
2 Serous III 2 < 2 cm CR Not present Stable
3 Serous III 3 > 2 cm CR Not present Stable
4 Serous III 1 > 2 cm WR Not present Stable
5 Serous III 2 > 2 cm CR Not present Stable
6 Serous III 2 < 2 cm CR Not present Stable
7 Serous III 3 > 2 cm CR Not present Stable
8 Serous III 3 > 2 cm CR Not present Stable
9 Serous III 3 > 2 cm CR Not present Stable
10 Serous III 3 > 2 cm CR Not present Stable
11 Serous III 3 > 2 cm CR Not present Stable
12 Serous III 3 > 2 cm WR Not present Stable
13 Serous III 3 > 2 cm WR Not present Stable
14 Serous IV 3 > 2 cm PR Not present Stable
15 Serous IV 3 > 2 cm CR Not present Stable
16 Serous IV 3 > 2 cm WR Not present Stable
17 Serous IV 3 > 2 cm WR Not present Stable
18 Serous IV 3 < 2 cm CR Not present Stable
19 Serous I 1 Absent - Not present Stable
20 Serous II 3 Absent - Not present Stable
21 Serous III 1 Absent - Not present Stable
22 Serous III 3 Absent - Not present Stable
23 Serous III 1 Absent - Not present Stable
24 Serous II B Absent - Not present Stable
25 Serous II B Absent - Not present Stable
26 Serous III B Absent - Not present Stable
27 Clear cell I 3 < 2 cm CR Not present Stable
28 Clear cell II 3 > 2 cm WR Not present Stable
29 Clear cell III 3 > 2 cm PR Not present Stable
30 Clear cell IV 3 > 2 cm WR Not present Stable
31 Clear cell I 3 Absent - Not present Stable
32 Clear cell I 3 Absent - Not present Stable
33 Clear cell II 3 Absent - Not present Stable
34 Clear cell III 3 Absent - Not present Stable
B: borderline tumours; CR: complete response; PR: partial response; WR: without response. The stage classification was according to the Federation Internationale de Gynecologie [21] and grading was defined as well differentiated (1), moderately differentiated (2), and poorly differentiated (3).
==== Refs
Parkin DM Bray FI Devesa SS Cancer burden in the year 2000: the global picture Eur J Cancer 2001 37 S4 66 11602373 10.1016/S0959-8049(01)00267-2
Crown J O'Leary M The taxanes: an update Lancet 2000 355 1176 1178 10791395 10.1016/S0140-6736(00)02074-2
Piccart MJ Bertelsen K James K Cassidy J Mangioni C Simonsen E Stuert G Kaye S Vargote I Blom R Grimshaw R Atkinson RJ Swenerton KD Trope C Nardi M Kaern J Tumolo S Timmers P Roy JA Lhoas F Lindvall B Bacon M Birt A Andersen JE Zee B Paul J Baron B Pecorelli S Randomized intergroup trial of cisplatin-paclitaxel versus cisplatin-cyclophosphamide in women with advanced epithelial ovarian cancer: three year results J Natl Cancer Inst 2000 92 699 708 10793106 10.1093/jnci/92.9.699
McGuire Ozols RF Chemotherapy in advanced ovarian cancer Semin Oncol 1998 25 340 348 9633846
Agarwal R Kaye SB Ovarian cancer: strategies for overcoming resistance to chemotherapy Nat Rev Cancer 2003 3 502 516 12835670 10.1038/nrc1123
Wang TH Wang HS Soong YK Paclitaxel-induced cell death: where the cell cycle and apoptosis come together Cancer 2000 88 2619 2628 10861441 10.1002/1097-0142(20000601)88:11<2619::AID-CNCR26>3.0.CO;2-J
Floyd-Smith GA de Martinville B Francke U An expressed tubulin gene, TUBB, is located on the short arm of human chromosome 6 and two related sequences are dispersed on chromosome 8 and 13 Exp Cell Res 1986 163 539 548 3007184 10.1016/0014-4827(86)90084-4
Ludueña RF Multiple forms of tubulin: different gene products and covalent modifications Int Rev Cytol 1998 178 207 275 9348671
Perez RP Cellular and molecular determinants of cisplatin resistance Eur J Cancer 1998 34 1535 1542 9893624 10.1016/S0959-8049(98)00227-5
Giannakakou P Sackett DL Kang YK Zhan Z Butters JT Fojo T Poruchynsky MS Paclitaxel-resistant human ovarian cancer cells have mutant β-tubulins that exhibit impaired paclitaxel-driven polymerization J Biol Chem 1997 272 17118 17125 9202030 10.1074/jbc.272.27.17118
Gonzalez-Garay ML Chang L Blade K Menick DR Cabral F A β-tubulin leucine cluster involved in microtubule assembly and paclitaxel resistance J Biol Chem 1999 271 23875 23882 10.1074/jbc.274.34.23875
Monzó M Rosell R Sanchéz JJ Lee JS O'Brate A Gonzalez-Larriba JL Alberola V Lorenzo JC Nunez L Ro JY Martin C Paclitaxel resistance in non-small cell lung cancer associated with β-tubulin gene mutations J Clin Oncol 1999 17 1786 1793 10561216
de las Alas MM Aebi S Fink D Howell SB Los G Loss of DNA mismatch repair: effects on the rate of mutation to drug resistance J Natl Cancer Inst 1997 89 1537 1541 9337351 10.1093/jnci/89.20.1537
Vaurs C Bignon YJ Quoi de neuf sur la signification des instabilités microssatellitaires dans les pathologies tumorales humaines? Bull Cancer 1997 84 1061 1071 9536988
Fink D Aebi S Howell SB The role of DNA mismatch repair in drug resistance Clin Cancer Res 1998 4 1 6 9516945
Lage H Dietel M Involvement of the DNA mismatch repair system in antineoplastic drug resistance J Cancer Res Clin Oncol 1999 125 156 165 10235469 10.1007/s004320050258
Soignet SL Spriggs DR Barakat RR, Bevers MW, Gershenson DM General principles of chemotherapy Handbook of Gynecologic Oncology 2001 London: Martin Dunitz 243 263
Hoang JM Cottu PH Thuille B Salmon RJ Thomas G Hamelin R BAT – 26, an indicator of the replication error phenotype in colorectal cancers and cell lines Cancer Res 1997 57 300 303 9000572
Zhou XP Hoang JM Li YJ Seruca R Carneiro F Sobrinho-Simoes M Lothe RA Gleeson CM Russel SE Muzeau F Flejou JF Hoang-Xuan L Lidereau R Thomas G Hamelin R Determination of the replication error phenotype in human tumors without the requirement for matching normal DNA by analysis of mononucleotide repeat microsatellites Genes Chromosomes Cancer 1998 21 101 107 9491320 10.1002/(SICI)1098-2264(199802)21:2<101::AID-GCC4>3.0.CO;2-4
Lamb P Crawford L Characterization of the human P53 gene Mol Cell Biol 1986 6 1379 1385 2946935
Benedet JL Bender H Jones H IIINgan HYS Pecorelli S FIGO staging classifications and clinical practice guidelines in the management of gynecologic cancers. FIGO Committee on Gynecologic Oncology Int J Gynaecol Obstet 2000 70 209 262 11041682 10.1016/S0020-7292(00)90001-8
Sale S Sung R Shen P Yu K Wang Y Duran GE Kim JH Fojo T Oefner PJ Sikic BI Conservation of the class I β-tubulin gene in human populations and lack of mutations in lung cancers and paclitaxel-resistant ovarian cancers Mol Cancer Ther 2002 1 215 225 12467216
Lamendola DE Duan Z Penson RT Oliva E Seiden MV Beta tubulin mutations are rare in ovarian carcinoma Anticancer Res 2003 23 681 686 12680167
Kelley MJ Li S Harpole DH Genetic analysis of the β-tubulin gene, TUBB, in non-small-cell lung cancer J Nat Cancer Inst 2001 93 1886 1888 11752014 10.1093/jnci/93.24.1886
Hasegawa S Miyoshi Y Egawa C Ishitobi M Tamaki Y Monden M Nogushi S Mutational analysis of the class I beta-tubulin gene in human breast cancer Int J Cancer 2002 101 46 51 12209587 10.1002/ijc.10575
Maeno K Ito K Hama Y Shingu K Kimura M Sano M Nakagomi H Tsuchiaya S Fujimori M Mutation of the class I β-tubulin gene does not predict response to paclitaxel for breast cancer Cancer Lett 2003 198 89 97 12893435 10.1016/S0304-3835(03)00279-9
Urano N Fujiwara Y Hasegawa S Miyoshi Y Nogushi S Takiguchi S Yasuda T Yano M Monden M Absence of beta-tubulin gene mutation in gastric carcinoma Gastric Cancer 2003 6 108 112 12861402
Orth K Hung J Gazdar A Bowcock A Mathis JM Sambrook J Genetic instability in human ovarian cancer cell lines Proc Natl Acad Sci USA 1994 91 9495 9499 7937795
Aebi S Haidar BK Gordon R Cenni B Zheng H Fink D Christen RD Boland CR Koi M Fishel R Howell SB Loss of DNA mismatch repair in acquired resistance to cisplatin Cancer Res 1996 56 3087 3090 8674066
Fink D Nebel S Norris PS Baergen RN Wilczynski SP Costa MJ Haas M Cannistra SA Howell SB Enrichment for DNA mismatch repair deficient cells during treatment with cisplatin Int J Cancer 1998 77 741 746 9688308 10.1002/(SICI)1097-0215(19980831)77:5<741::AID-IJC13>3.0.CO;2-4
Fink D Nebel S Aebi S Zheng H Cenni B Nheme A Christen RD Howell SB The role of DNA mismatch repair in platinum drug resistance Cancer Res 1996 56 4881 4886 8895738
Aebi S Fink D Gordon R Kim HK Zheng H Finj JL Howell SB Resistance to cytotoxic drugs in DNA mismatch repair-deficient cells Clin Cancer Res 1997 3 1763 1767 9815561
Fink D Zheng H Nebel S Norris PS Aebi S Lin TP Nehme A Christen RD Haas M MacLeod CL Howell SB In vitro and in vivo resistance to cisplatin in cells that have lost DNA mismatch repair Cancer Res 1997 57 1841 1845 9157971
Fujita M Enomoto T Yoshino K Nomura T Buzard GS Inoue M Okudaira Y Microsatellite instability and alterations in the hMSH2 gene in human ovarian cancer Int J Cancer 1995 64 361 366 8550235
Pieretti M Cavalieri C Conway PS Gallion HH Powell DE Turker MS Genetic alterations distinguish different types of ovarian tumors Int J Cancer 1995 64 434 440 8550247
Arzimanoglou II Lallas T Osborne M Barber H Gilbert F Microsatellite instability differences between familial and sporadic ovarian cancers Carcinogenesis 1996 17 1799 1804 8824498
Haas CJ Diebold J Hirschmann A Rohrbach H Schid S Lohrs U Microsatellite analysis in serous tumors of the ovary Int J Gynecol Pathol 1999 18 158 162 10202674
Geisler JP Goodheatr MJ Sood AK Holmes RJ Hatterman-Zogg MA Buller RE Mismatch repair gene expression defects contribute to microsatellite instability in ovarian carcinoma Cancer 2003 98 2199 2206 14601090 10.1002/cncr.11770
Cai KQ Albarracin C Rosen D Zhong R Zheng W Luthra R Broaddus R Liu J Microsatellite instability and alteration of the expression of the hMLH1 and hMSH2 in ovarian clear cell carcinoma Hum Pathol 2004 35 552 559 15138928 10.1016/j.humpath.2003.12.009
Singer G Kallinowski T Hartmann A Dietmaier W Wild PJ Schraml P Sauter G Different types of microsatellite instability in ovarian carcinoma Int J Cancer 2004 112 643 646 15382045 10.1002/ijc.20455
Parekh H Wiesen K Simpkins H Acquisition of taxol resistance via P-glycoprotein and non-P-glycoprotein-mediated mechanisms in human ovarian carcinoma cells Biochem Pharmacol 1997 53 461 470 9105396 10.1016/S0006-2952(97)83383-7
Ranganathan S Benetatos CA Colarusso PJ Dexter DW Hudes GR Altered β-tubulin isotype expression in paclitaxel resistant human prostate carcinoma cells Brit J Cancer 1998 77 562 566 9484812
Kavallaris M Burkhart CA Horowitz SB Antisense oligonucleotides to class III β-tubulin sensitize drug-resistant cells to Taxol Brit J Cancer 1999 80 1020 1025 10362110 10.1038/sj.bjc.6690507
Haldar S Basu A Croce CM Bcl2 is the guardian of microtubule integrity Cancer Res 1997 57 229 233 9000560
Liu JR Fletcher B Page C Hu C Nunez G Baker V Bcl-xL is expressed in ovarian carcinoma and modulates chemotherapy-induced apoptosis Gynecol Oncol 1998 70 398 403 9790794 10.1006/gyno.1998.5125
Tai YT Lee S Niloff E Weisman C Strobel T Cannistra SH BAX protein expression and clinical outcome in epithelial ovarian cancer J Clin Oncol 1998 16 2583 2590 9704707
Perego P Romanelli S Carenini N Magnani I Leone R Bonetti A Paolicchi A Zunino F Ovarian cancer cisplatin-resistant cell lines: multiple changes including collateral sensitivity to taxol Ann Oncol 1998 9 423 430 9636834 10.1023/A:1008265012435
|
16095531
|
PMC1199587
|
CC BY
|
2021-01-04 16:03:06
|
no
|
BMC Cancer. 2005 Aug 11; 5:101
|
utf-8
|
BMC Cancer
| 2,005 |
10.1186/1471-2407-5-101
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-1021610517110.1186/1471-2407-5-102Research ArticleAbsence of pathogenic mitochondrial DNA mutations in mouse brain tumors Kiebish Michael A [email protected] Thomas N [email protected] Department of Biology, Boston College, Chestnut Hill, MA 02467 USA2005 16 8 2005 5 102 102 19 5 2005 16 8 2005 Copyright © 2005 Kiebish and Seyfried; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Somatic mutations in the mitochondrial genome occur in numerous tumor types including brain tumors. These mutations are generally found in the hypervariable regions I and II of the displacement loop and unlikely alter mitochondrial function. Two hypervariable regions of mononucleotide repeats occur in the mouse mitochondrial genome, i.e., the origin of replication of the light strand (OL) and the Arg tRNA.
Methods
In this study we examined the entire mitochondrial genome in a series of chemically induced brain tumors in the C57BL/6J strain and spontaneous brain tumors in the VM mouse strain. The tumor mtDNA was compared to that of mtDNA in brain mitochondrial populations from the corresponding syngeneic mouse host strain.
Results
Direct sequencing revealed a few homoplasmic base pair insertions, deletions, and substitutions in the tumor cells mainly in regions of mononucleotide repeats. A heteroplasmic mutation in the 16srRNA gene was detected in a spontaneous metastatic VM brain tumor.
Conclusion
None of the mutations were considered pathogenic, indicating that mtDNA somatic mutations do not likely contribute to the initiation or progression of these diverse mouse brain tumors.
==== Body
Background
Mitochondrial DNA (mtDNA) is essential for the production of subunits for the electron transport chain, since it encodes peptides for Complex I, III, IV, and V. The mitochondrial genome consists of a 16.5 kb piece of circular DNA encoding 37 genes on the L and H strands. These include 22 tRNAs, 2 rRNAs, and 13 respiratory chain subunits [1]. The mitochondrial genetic code varies slightly from the nuclear genetic code and the mutation rate of the mitochondrial genome is 10 fold higher than that of the nuclear genome [2-4]. This higher mutation rate may be due in part to a lack of protective histones, reduced fidelity of polymerase γ, or to the localization of the mitochondrial genome to the inner mitochondrial membrane near damaging reactive oxygen species [5-9]. mtDNA mutations may also contribute to a decrease in metabolic efficiency that enhances aging, promotes neurodegeneration, and is conducive to carcinogenesis [10-12].
mtDNA mutations have been studied in various types of cancers, including breast [13], bladder, lung, head and neck [14], gastric [15], esophageal [16], pancreatic [17], and colon [18]. The majority of the mutations in the mitochondrial genome of human tumors were found in the hypervariable regions of the displacement loop (D-loop) [19-21]. Alterations in these hypervariable regions are unlikely to alter mitochondrial function or produce a diseased phenotype [22]. These hypervariable regions correspond to the OL and Arg tRNA in mouse mtDNA [23]. Otto Warburg originally emphasized that the high glycolytic rate of tumors resulted from diminished or disturbed respiration [24,25]. Recent studies suggest that mtDNA mutations may contribute to the respiratory defects in cancer [18]. Furthermore, studies of human tumors have demonstrated an association between mtDNA somatic mutations and tumor progression [26-28].
Few studies have evaluated mtDNA mutations in brain cancer. Mitochondrial genome instability was found in the hypervariable regions in the D-loop of human gliomas, meningiomas, schwannomas, gliomatosis cerebri, glioblastomas, astrocytomas, and neurofibromas [29-33]. Other studies of the mitochondrial genome focused on mtDNA copy number in brain tumors [34]. In the only study examining the entire mitochondrial genome, Wong et al. found that the majority of mtDNA somatic mutations in medulloblastomas existed in regions of mononucleotide repeats, rather than in genes encoding proteins for the respiratory chain complexes [35]. A recent study in mouse tumorigenic fibroblasts, suggest that nuclear mutations rather than mtDNA mutations contribute to the tumor phenotype [36]. It is not yet clear if mtDNA mutations contribute to brain tumor progression.
In contrast to most tissues where mitochondrial morphology and function is relatively homogeneous, mitochondria in brain are heterogeneous and exist as non-synaptic free mitochondria (FM), synaptic light mitochondria (LM), and synaptic heavy mitochondria (HM) [37,38]. This morphological heterogeneity is thought to reflect the metabolic heterogeneity of mitochondria within and between the various neuronal and non-neuronal (glial) cell populations [39]. Besides morphological heterogeneity, genetic heterogeneity in mtDNA has been reported in normal human brain tissue [40]. In this study, we sequenced the entire mitochondrial genome in a series of chemically induced and spontaneous mouse brain tumors that differ in metastasis, malignancy, and vascularity. To our knowledge, no prior studies have compared the sequence of the entire mitochondrial genome in brain tumor cell lines with that of the syngeneic brain tissue. Although we found several mutations, none of these were considered pathogenic.
Methods
Mice and brain tumors
The inbred VM and C57BL/6J (B6) mouse strains were obtained originally from Professor H Fraser, University of Edinburgh, and from the Jackson Laboratory, Bar Harbor, ME, respectively. All mice used in the study were propagated in the animal care facility at Boston College using animal husbandry conditions described previously [41]. All animal-use procedures were in accordance with the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care Committee.
The EPEN and CT-2A brain tumors were produced originally from implantation of 20-methylcholanthrene into the brains of B6 mice as previously described [42,43]. The EPEN tumor arose in the cerebral ventricle and was characterized as an ependymoblastoma, whereas the CT-2A tumor arose in the cerebral cortex and was characterized as a poorly differentiated astrocytoma [42,44]. The CT-2A tumor contains complex gangliosides, is highly angiogenic in vivo, and grows in vitro as a fusiform monolayer [42,45,46]. In contrast, the EPEN tumor expresses only the simple ganglioside GM3, is less angiogenic in vivo, and grows in vitro as cuboidal cell islands [41,42,46-49]. The nonmetastatic (VM-NM) and the metastatic (VM-M) tumors arose spontaneously in the brains of VM mice as previously described for other VM brain tumors [50,51].
Tumor cell culture
Cultured cell lines were prepared from each tumor as we previously described [52,53]. The number of passages for CT-2A and EPEN cell lines were greater than 40, where as VM-NM and VM-M were passaged 9 and 11 times, respectively. The VM-M cell line maintained metastatic capacity when implanted into VM mice. The tumor cells were grown in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS) in a humidified atmosphere containing 95% air and 5% CO2 at 37°C. Upon reaching confluency, the cells were trypsinized and collected.
Mitochondrial isolation from brain tissue and cultured tumor cells
Mitochondria were isolated from VM and B6 cerebral cortex according to previously described procedures [38,54]. Cerebral cortex mitochondria were separated into three populations following ficoll gradient centrifugation. These populations included non-synaptic free mitochondria (FM), synaptic light mitochondria (LM), and synaptic heavy mitochondria (HM). Mitochondria were isolated from the tumor cell lines as a single enriched mitochondrial fraction as previously described [54,55]. Mitochondria were isolated in duplicate independent samples from each of the brain tumor cell lines for sequence comparison.
Isolation and amplification of mouse mtDNA
mtDNA from cerebral cortex and cultured tumor cells was isolated using the DNeasy Tissue Kit (QIAGEN, Valencia, CA, USA) and was amplified in 12 overlapping segments covering the entire mitochondrial genome. PCR reactions were performed in a volume of 50 μl using the Eppendorf Master Taq system (Brinkmann Eppendorf, Westbury, NY). Primers used for mtDNA amplification are given in Table 1. PCR conditions for all primer pairs consisted of an initial denaturation at 94°C for 120 s followed by 34 cycles of 94°C for 60 s, an annealing temperature of 57–66°C for 35 s and 72°C for extension at 105 s. A final extension was for 5 min. Optimal annealing temperatures were determined by performing gradient PCR on each of the primer sets.
mtDNA sequencing
Amplified regions of mtDNA were purified using a PCR purification kit (MO BIO Laboratories, Solana Beach, CA) and were then visualized by 1% agarose gel electrophoresis. DNA concentration was estimated using a Low DNA Mass Ladder (Invitrogen, Carlsbad, CA) by gel electrophoresis. Approximately 25 fmol of the purified PCR product was used in the sequencing reaction following manufacturer's instructions for the CEQ DTCS kit (Beckman Coulter, Fullerton, CA). The amount of DNA used varied depending on the size of the PCR fragment. The primers in Table 1, together with 75 nested primers, were used in the sequencing reactions to obtain double-stranded sequence. Nested primer information can be made available upon request to corresponding investigators.
Sequencing reactions were performed at 96°C for 20 s, 50°C for 20 s, and 60°C for 240 s based on a 32 cycle reaction. Sequencing products were ethanol precipitated in cold 95% ethanol, washed twice with cold 70% ethanol, dried for 20 minutes, and then re-dissolved in sample loading solution provided by the manufacturer (Beckman Coulter, Fullerton, CA). The mtDNA sequence alignments were then compared to those from Mus Musculus [GenBank: V00711] [56] and from B6 mice [GenBank: AY172335] [23].
Results
Analysis of mtDNA variation in B6 and VM mouse strains
Sequence analysis of the entire mitochondrial genome of the VM and B6 mouse strains revealed a single nucleotide base pair change at position 11780, found in both strains, compared to the republished B6 sequence [23]. This resulted in a T->A transversion, causing an amino acid change of an isoleucine to methionine in the ND5 gene (Table 2). This transversion was found in all three brain mitochondrial populations for both strains and existed in all brain tumor cell lines corresponding to the host strain. This was the only variation found between the B6 strain sequence and the republished B6 sequence [23]. The VM strain also had two additional nucleotide variations compared to the B6 strain. The first variation was found in the ND3 gene at position 9461. This caused a T->C transition, but resulted in no amino acid change. A second variation existed in the poly-A tract of the Arg tRNA, where an adenine was inserted at position 9829 (Table 2). The data show that few mtDNA variations exist between the B6 and VM inbred mouse strains.
mtDNA somatic mutations in mouse brain tumor cell lines compared to syngeneic host mtDNA sequence
Analysis of the EPEN brain tumor cell line revealed two somatic mutations compared to the syngeneic B6 host strain. One mutation involved an insertion of two adenines in the Arg tRNA at position 9829 (Table 3). This is not expected to alter Arg tRNA function because the mutation occurred in a highly variable site in mouse strains [23]. A second mutation involved a T->C transition in the ND3 gene at position 9461, but did not change the amino acid. Analysis of the CT-2A brain tumor cell line revealed only a single mutation compared to the syngeneic B6 host strain. This involved an insertion of an adenine in the OL at position 5182 (Table 3). This mutation is also not likely to have functional consequence [23].
As in the chemically induced EPEN and CT-2A brain tumor cell lines, only a few somatic mutations were found in the VM brain tumor cell lines (VM-NM and VM-M) compared to the syngeneic VM host (Table 3). Two somatic mutations were found in the nonmetastatic VM-NM cell line. The first mutation involved a deletion of an adenine in the OL at position 5182. This is the same location where the only variation was found in the CT-2A tumor and is consistent with hypervariability in this region [23]. The second mutation in this cell line involved a T->C transition in the D-loop at position 15584. Neither of these mutations have functional consequence.
A single mutation was found in the metastatic VM-M cell line involving an A->G transition in the 16srRNA gene at position 2159. In contrast to the other mtDNA mutations found in either the B6 or the VM tumor cell lines, the A->G transition in the VM-M cell line was heteroplasmic (Figure 1). This heteroplasmic mutation, however, was not found in another independently derived cell line from this metastatic tumor. Viewed together, these data indicate that only a few mtDNA somatic mutations exist in the chemically induced tumors in the B6 mice or in the spontaneous brain tumors in VM mice. Also, most of these mutations were found in hypervariable regions and none are considered pathogenic.
Discussion
The purpose of this study was to determine if mtDNA mutations existed in chemically induced and spontaneous mouse brain tumors and whether such mutations might provide insight on the development or progression of these experimental brain tumors. To reduce the probability of detecting spurious pathogenic mutations, we compared the mtDNA sequence of each brain tumor cell line with that of normal mtDNA from the syngeneic mouse host strains. Furthermore, to avoid the coamplification of mitochondrial pseudogenes that exist in nuclear DNA, we amplified only large mtDNA fragments (1–1.8 kb in length) in isolated mitochondria [18,57,58]. Because the solid tumors become infiltrated with host stromal cells containing wildtype mitochondria [59], we used cultured tumor cell lines for these studies instead of whole tumor tissue. Hence, our experimental design will detect any mtDNA mutation that could alter mitochondrial function.
Several mtDNA mutations were found in the mouse brain tumor cell lines, but most of these were located in hypervariable mononucleotide hotspot regions, such as in the OL and in the Arg tRNA for mouse mtDNA. Moreover, none of these mutations were considered pathogenic since they did not change amino acid sequence and therefore could not alter gene function. Tumor cells in each line were selected from original tumors that may have contained multiple transformed cell types. The lines therefore represent only those cells with the most aggressive malignant growth in vitro and in vivo. The absence of pathogenic mutations in these tumor cells, further suggests that mtDNA mutations do not likely contribute to the initiation or the progression of these diverse mouse brain tumors.
Most tumors including brain tumors express abnormalities in the number and function of their mitochondria [60,61]. Warburg originally emphasized that the high glycolytic rate of tumors results from diminished or disturbed respiration [24,25]. Later studies in a variety of neural and non-neural tumor systems showed that these respiratory disturbances could involve abnormalities in TCA cycle components, alterations in electron transport, and deficiencies in oxidative phosphorylation [62-66]. Recent studies also suggest that mtDNA mutations may contribute to the respiratory defects in cancer [18]. Our findings indicate that diminished respiration and the glycolytic dependence of the chemically induced CT-2A and EPEN brain tumors or the two spontaneous VM brain tumors studied here do not result from somatic mutations in their mtDNA.
In contrast to previous reports of genetic heterogeneity in normal human brain mtDNA, [40,67] we found no evidence for genetic heterogeneity in mtDNA isolated from synaptic and nonsynaptic mitochondria in the C57BL/6J and VM mouse strains by direct sequencing. On the other hand, our findings that most mtDNA changes in the mouse brain tumors were in hypervariable regions of mononucleotide repeats are in general agreement with previous studies in human brain tumors (Table 4). We detected a single heteroplasmic mutation in the 16srRNA gene of the VM-M metastatic cell line, which was determined by sequence analysis [68]. Since this heteroplasmic mutation was not detected in another independently derived cell line from this tumor, it is unlikely that the mutation is associated with either tumorigenesis or metastasis. The EPEN cell line contained a nucleotide variation compared to the B6 genotype at position 9461 in the ND3 gene. This nucleotide variation in the EPEN cell line also was found in the VM-NM and VM-M cell lines. This polymorphism could be due to a base pair wobble at that location or could result from a nucleotide variation in the original C57BL/6J mouse from which the EPEN tumor arose [44].
The four mouse brain tumor cell lines we examined represent both spontaneous and chemically induced tumors that differ in cell origin, biochemical composition, and angiogenesis. The selection of the cell lines provides a comparison for defining the contribution that mtDNA might play in carcinogenesis. Although we did not find any mtDNA pathogenic mutations in these brain tumors, we do not rule out the possibility that such mutations might occur in other spontaneous mouse brain tumors or in mouse brain tumors induced with other chemical carcinogens. Our findings suggest that mtDNA mutations do not likely contribute to the development or progression of CT-2A or EPEN chemically induced tumors or the metastatic and nonmetastatic spontaneous VM brain tumors.
With the exception of a single variation at nucleotide 11780 in the ND5 gene, our whole mitochondrial genome sequence for the B6 strain was in complete agreement with the recently republished sequence for this mouse strain [23]. This single nucleotide difference involved a T->A transversion that changed isoleucine to methionine. This difference was also found in the VM mouse strain as well as in all examined mouse brain tumor cell lines. It is not yet clear if this variation represents an error in the republished B6 mtDNA sequence or a population genetic variation among B6 mouse strains. Brain mitochondria were fractionated into different populations for genotyping to confirm that there was no mtDNA heterogeneity in hypervariable regions that might be associated with metabolic heterogeneity. Sequencing of all three brain populations also served as an internal control for the genotype of the mouse strains.
In summary, our results show that the mitochondrial genome of these mouse brain tumor cell lines contain mutations in regions of mononucleotide repeats. These mtDNA mutations in the mouse brain tumors are similar to those found previously in hypervariable regions of the D-loop in human brain tumors. However, none of the mouse mutations were pathogenic suggesting that mtDNA changes do not contribute to the carcinogenic progression of these chemically induced and spontaneous mouse brain tumors.
List of abbreviations
The abbreviations used are: OL, origin of replication of the light strand; mtDNA, mitochondrial DNA; kb, kilobase; D-loop, displacement loop; B6, C57BL/6J
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MK carried out mitochondrial isolation, amplification and sequencing of brain and tumor mtDNA, sequence analysis and drafted the manuscript. TS participated in the study's design, coordination, and editing of the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We would like to thank Laura Abate and Kristen Gorham for technical assistance. This work was supported from NIH grants HD39722 and CA102135, a grant from the American Institute of Cancer Research and the Boston College Research Expense Fund.
Figures and Tables
Figure 1 Heteroplasmy in the metastatic VM-M tumor cell line. Mitochondrial 16srRNA revealed a heteroplasmic mutation at position 2159 of an A to G transition. The three VM brain mitochondrial populations (LM, HM, and FM) displayed the same chromotagraph profile. The same level of heteroplasmy was observed in duplicate independent samples for the VM-M cell line. Arrow indicates heteroplasmic nucleotide at position 2159 compared to control.
Table 1 Primers used for amplification of Mus Musculus mitochondrial genome
Primer pair mtDNA bp Forward primer Reverse primer
1 15309-00040 GGTCTTGTAAACCTGAAATGAAG GCATTTTCAGTGCTTTGCTTTG
2 16173-01436 CTGATCAATTCTAGTAGTTCC CAAGCTCGTTAGGCTTTTCAC
3 01314–02880 GAATTACAGCTAGAAACCCCG CGTATGGACCAACAATGTTAGG
4 02673–03873 GTTATTAGGGTGGCAGAGCCAG GCCCGATAGCTTAATTAGCTGAC
5 03731–05324 CTTTGATAGAGTAAATTATAGAGG TAAAATGGCTGAGTAAGCATTAG
6 05194–06593 GTCTTAGTAGAGATTTCTCTACAC GTTTACTCCTACGAATATGATGG
7 06477–07996 GGAGCAGTGTTTGCTATCATAGC TGATGGCTACAACGATTGGGAATC
8 07847–09438 GTCTCATCACAAACATTCCCAC CAGAATCTACTAATTGGAAGTCAG
9 09307–10806 GATTTGAAGCCGCAGCATGATAC GTCATAGGTGAACTCCATATAATGG
10 10647–12399 GCCCTCATCTTAATCCAAAACC CTGTAGCTGCGATTAATAGGC
11 12247–13788 CAATCCTCTATAACCGCATCG GTATTGGGGGTGATTATAGAG
12 13670–15498 CCAACATCATCAACCTCATAC GTACTAGCTTATATGCTTGGG
The nucleotide positions were based on Mus Musculus sequence from Bibb et al. 1981 [GenBank: V00711]
Table 2 Summary of mtDNA nucleotide variations in C57BL/6J and VM mouse strainsa
Mouse strain Nucleotide position Gene DNA change Amino acid change
C57BL/6J 11780 ND5 T->A Ile->Met
VM 11780 ND5 T->A Ile->Met
VM 9829 Arg tRNA InsA NAb
VM 9461 ND3 T->C Ile->Ile
Accession numbers for C57BL/6J and VM strains are DQ106412 and DQ106413, respectively
a Nucleotide variation compared to the republished C57BL/6J mtDNA sequence from Bayona-Bafaluy et al. 2003
b NA, not applicable
Table 3 Summary of mtDNA mutations in experimental mouse brain tumor cell linesa
Brain tumor cell line Metastatic Host genetic background Nucleotide position Gene DNA change Stateb Amino acid change
EPEN - B6 9829 Arg tRNA InsAA Homo NAc
9461 ND3 T->C Homo Ile->Ile
CT-2A - B6 5182 OL InsA Homo NA
VM-NM - VM 5182 OL DelA Homo NA
15584 D-loop T->C Homo NA
VM-M + VM 2159 16srRNA A->G Hetero NA
a Tumor cell lines mtDNA sequences were compared to host strain brain mtDNA populations
b Homo, homoplasmy; hetero, heteroplasmy
c NA, not applicable
Table 4 mtDNA alterations in human brain tumors compared to matched tissue controls
Type of tumor Number of tumors Number of tumors with mtDNA somatic mutationsa Number of somatic mutations causing amino acid substitutionsc Region of mutation Reference
Neurofibroma 37 23/37 NAd D-loop Kurtz et al. 2004
Glioma 53 20/53 NA HVRII Vega et al. 2004
Meningioma 11 5/11 NA
Schwannoma 5 1/5 NA
Medulloblastoma 15 6/15 2/18 D-loop, Asp tRNA, COXI, CoxII, Thr tRNA, ND4 Wong et al. 2003
Gliomatosis cerebri 6 2/6 NA HVRII Kirches et al. 2003
Glioma 55 12/55 NA HVRII, D-loop Kirches et al. 2001
Astrocytoma 12 11/12b NA HVRII Kirches et al. 1999
a Number of tumors with mutations/number of tumors
b 2 of the 12 astrocytomas were compared to adjacent brain, the remaining 10 were aligned to Anderson et al. 1981 for mutation analysis
c Number of mutations causing amino acid substitutions/total number of mutations found in all tumors
d NA, not applicable because in hypervariable region of the D-loop (HVR)
==== Refs
Wolstenholme DR Animal mitochondrial DNA: structure and evolution Int Rev Cytol 1992 141 173 216 1452431
Barrell BG Bankier AT Drouin J A different genetic code in human mitochondria Nature 1979 282 189 194 226894 10.1038/282189a0
Neckelmann N Li K Wade RP Shuster R Wallace DC cDNA sequence of a human skeletal muscle ADP/ATP translocator: lack of a leader peptide, divergence from a fibroblast translocator cDNA, and coevolution with mitochondrial DNA genes Proc Natl Acad Sci U S A 1987 84 7580 7584 2823266
Wallace DC Ye JH Neckelmann SN Singh G Webster KA Greenberg BD Sequence analysis of cDNAs for the human and bovine ATP synthase beta subunit: mitochondrial DNA genes sustain seventeen times more mutations Curr Genet 1987 12 81 90 2896550 10.1007/BF00434661
Backer JM Weinstein IB Mitochondrial DNA is a major cellular target for a dihydrodiol-epoxide derivative of benzo[a]pyrene Science 1980 209 297 299 6770466
Allen JA Coombs MM Covalent binding of polycyclic aromatic compounds to mitochondrial and nuclear DNA Nature 1980 287 244 245 7432460 10.1038/287244a0
Kunkel TA Loeb LA Fidelity of mammalian DNA polymerases Science 1981 213 765 767 6454965
Croteau DL Bohr VA Repair of oxidative damage to nuclear and mitochondrial DNA in mammalian cells J Biol Chem 1997 272 25409 25412 9325246 10.1074/jbc.272.41.25409
Bianchi NO Bianchi MS Richard SM Mitochondrial genome instability in human cancers Mutat Res 2001 488 9 23 11223402 10.1016/S1383-5742(00)00063-6
Brown WM George MJ Wilson AC Rapid evolution of animal mitochondrial DNA Proc Natl Acad Sci U S A 1979 76 1967 1971 109836
Cavalli LR Liang BC Mutagenesis, tumorigenicity, and apoptosis: are the mitochondria involved? Mutat Res 1998 398 19 26 9626961
Augenlicht LH Heerdt BG Mitochondria: integrators in tumorigenesis? Nat Genet 2001 28 104 105 11381247 10.1038/88800
Tan DJ Bai RK Wong LJ Comprehensive scanning of somatic mitochondrial DNA mutations in breast cancer Cancer Res 2002 62 972 976 11861366
Fliss MS Usadel H Caballero OL Wu L Buta MR Eleff SM Jen J Sidransky D Facile detection of mitochondrial DNA mutations in tumors and bodily fluids Science 2000 287 2017 2019 10720328 10.1126/science.287.5460.2017
Maximo V Soares P Seruca R Rocha AS Castro P Sobrinho-Simoes M Microsatellite instability, mitochondrial DNA large deletions, and mitochondrial DNA mutations in gastric carcinoma Genes Chromosomes Cancer 2001 32 136 143 11550281 10.1002/gcc.1175
Hibi K Nakayama H Yamazaki T Takase T Taguchi M Kasai Y Ito K Akiyama S Nakao A Mitochondrial DNA alteration in esophageal cancer Int J Cancer 2001 92 319 321 11291064 10.1002/ijc.1204
Jones JB Song JJ Hempen PM Parmigiani G Hruban RH Kern SE Detection of mitochondrial DNA mutations in pancreatic cancer offers a "mass"-ive advantage over detection of nuclear DNA mutations Cancer Res 2001 61 1299 1304 11245424
Polyak K Li Y Zhu H Lengauer C Willson JK Markowitz SD Trush MA Kinzler KW Vogelstein B Somatic mutations of the mitochondrial genome in human colorectal tumours Nat Genet 1998 20 291 293 9806551 10.1038/3108
Okochi O Hibi K Uemura T Inoue S Takeda S Kaneko T Nakao A Detection of mitochondrial DNA alterations in the serum of hepatocellular carcinoma patients Clin Cancer Res 2002 8 2875 2878 12231530
Ha PK Tong BC Westra WH Sanchez-Cespedes M Parrella P Zahurak M Sidransky D Califano JA Mitochondrial C-tract alteration in premalignant lesions of the head and neck: a marker for progression and clonal proliferation Clin Cancer Res 2002 8 2260 2265 12114429
Kumimoto H Yamane Y Nishimoto Y Fukami H Shinoda M Hatooka S Ishizaki K Frequent somatic mutations of mitochondrial DNA in esophageal squamous cell carcinoma Int J Cancer 2004 108 228 231 14639607 10.1002/ijc.11564
Sanchez-Cespedes M Parrella P Nomoto S Cohen D Xiao Y Esteller M Jeronimo C Jordan RC Nicol T Koch WM Schoenberg M Mazzarelli P Fazio VM Sidransky D Identification of a mononucleotide repeat as a major target for mitochondrial DNA alterations in human tumors Cancer Res 2001 61 7015 7019 11585726
Bayona-Bafaluy MP Acin-Perez R Mullikin JC Park JS Moreno-Loshuertos R Hu P Perez-Martos A Fernandez-Silva P Bai Y Enriquez JA Revisiting the mouse mitochondrial DNA sequence Nucleic Acids Res 2003 31 5349 5355 12954771 10.1093/nar/gkg739
Warburg O On the origin of cancer cells Science 1956 123 309 314 13298683
Warburg O The Metabolism of Tumours 1931 New York, Richard R. Smith Inc. 327
Shay JW Werbin H Are mitochondrial DNA mutations involved in the carcinogenic process? Mutat Res 1987 186 149 160 3306359
Baggetto LG Role of mitochondria in carcinogenesis Eur J Cancer 1992 29A 156 159 1445736
Cavalli LR Varella-Garcia M Liang BC Diminished tumorigenic phenotype after depletion of mitochondrial DNA Cell Growth Differ 1997 8 1189 1198 9372242
Kirches E Michael M Woy C Schneider T Warich-Kirches M Schneider-Stock R Winkler K Wittig H Dietzmann K Loss of heteroplasmy in the displacement loop of brain mitochondrial DNA in astrocytic tumors Genes Chromosomes Cancer 1999 26 80 83 10441009 10.1002/(SICI)1098-2264(199909)26:1<80::AID-GCC11>3.0.CO;2-4
Kirches E Krause G Warich-Kirches M Weis S Schneider T Meyer-Puttlitz B Mawrin C Dietzmann K High frequency of mitochondrial DNA mutations in glioblastoma multiforme identified by direct sequence comparison to blood samples Int J Cancer 2001 93 534 538 11477557 10.1002/ijc.1375
Kirches E Mawrin C Schneider-Stock R Krause G Scherlach C Dietzmann K Mitochondrial DNA as a clonal tumor cell marker: gliomatosis cerebri J Neurooncol 2003 61 1 5 12587789 10.1023/A:1021296212233
Vega A Salas A Gamborino E Sobrido MJ Macaulay V Carracedo A mtDNA mutations in tumors of the central nervous system reflect the neutral evolution of mtDNA in populations Oncogene 2004 23 1314 1320 14647457 10.1038/sj.onc.1207214
Kurtz A Lueth M Kluwe L Zhang T Foster R Mautner VF Hartmann M Tan DJ Martuza RL Friedrich RE Driever PH Wong LJ Somatic mitochondrial DNA mutations in neurofibromatosis type 1-associated tumors Mol Cancer Res 2004 2 433 441 15328370
Liang BC Hays L Mitochondrial DNA copy number changes in human gliomas Cancer Lett 1996 105 167 173 8697440 10.1016/0304-3835(96)04276-0
Wong LJ Lueth M Li XN Lau CC Vogel H Detection of mitochondrial DNA mutations in the tumor and cerebrospinal fluid of medulloblastoma patients Cancer Res 2003 63 3866 3871 12873974
Akimoto M Niikura M Ichikawa M Yonekawa H Nakada K Honma Y Hayashi J Nuclear DNA but not mtDNA controls tumor phenotypes in mouse cells Biochem Biophys Res Commun 2005 327 1028 1035 15652499 10.1016/j.bbrc.2004.12.105
Battino M Gorini A Villa RF Genova ML Bovina C Sassi S Littarru GP Lenaz G Coenzyme Q content in synaptic and non-synaptic mitochondria from different brain regions in the ageing rat Mech Ageing Dev 1995 78 173 187 7596200 10.1016/0047-6374(94)01535-T
Genova ML Bovina C Marchetti M Pallotti F Tietz C Biagini G Pugnaloni A Viticchi C Gorini A Villa RF Lenaz G Decrease of rotenone inhibition is a sensitive parameter of complex I damage in brain non-synaptic mitochondria of aged rats FEBS Lett 1997 410 467 469 9237684 10.1016/S0014-5793(97)00638-8
Lai JC Oxidative metabolism in neuronal and non-neuronal mitochondria Can J Physiol Pharmacol 1992 70 Suppl S130 7 1295663
Trounce I Schmiedel J Yen HC Hosseini S Brown MD Olson JJ Wallace DC Cloning of neuronal mtDNA variants in cultured cells by synaptosome fusion with mtDNA-less cells Nucleic Acids Res 2000 28 2164 2170 10773087 10.1093/nar/28.10.2164
Ranes MK El-Abbadi M Manfredi MG Mukherjee P Platt FM Seyfried TN N -butyldeoxynojirimycin reduces growth and ganglioside content of experimental mouse brain tumours Br J Cancer 2001 84 1107 1114 11308262 10.1054/bjoc.2000.1713
Seyfried TN el-Abbadi M Roy ML Ganglioside distribution in murine neural tumors Mol Chem Neuropathol 1992 17 147 167 1418222
Zimmerman HM Arnould H Experimental brain tumors: I. Tumors produced with methylcholanthrene Cancer Res 1941 1 919 938
Rubin R Sutton C The ultrastructure of the mouse ependymoblastoma and its contained virus-like particles J Neuropathol Exp Neurol 1968 27 136 5656551
Mukherjee P El-Abbadi MM Kasperzyk JL Ranes MK Seyfried TN Dietary restriction reduces angiogenesis and growth in an orthotopic mouse brain tumour model Br J Cancer 2002 86 1615 1621 12085212 10.1038/sj.bjc.6600298
Mukherjee P Abate LE Seyfried TN Antiangiogenic and proapoptotic effects of dietary restriction on experimental mouse and human brain tumors Clin Cancer Res 2004 10 5622 5629 15328205
Manfredi MG Lim S Claffey KP Seyfried TN Gangliosides influence angiogenesis in an experimental mouse brain tumor Cancer Res 1999 59 5392 5397 10537325
Seyfried TN Yu RK Saito M Albert M Ganglioside composition of an experimental mouse brain tumor Cancer Res 1987 47 3538 3542 3581087
Cotterchio M Seyfried TN The influence of ImuVert, a biological response modifier, on the growth and ganglioside composition of murine neural tumors Mol Chem Neuropathol 1993 20 163 172 8297420
Fraser H Brain tumours in mice, with particular reference to astrocytoma Food Chem Toxicol 1986 24 105 111 3957158 10.1016/0278-6915(86)90344-3
El-Abbadi M Seyfried TN Yates AJ Orosz C Lee MC Ganglioside composition and histology of a spontaneous metastatic brain tumour in the VM mouse Br J Cancer 2001 85 285 292 11461091 10.1054/bjoc.2001.1909
Bai H Seyfried TN Influence of ganglioside GM3 and high density lipoprotein on the cohesion of mouse brain tumor cells J Lipid Res 1997 38 160 172 9034210
el-Abbadi M Seyfried TN Influence of growth environment on the ganglioside composition of an experimental mouse brain tumor Mol Chem Neuropathol 1994 21 273 285 8086038
Pon LA Schon EA Wilson L and Matsudaira P Mitochondria Methods in Cell Biology 2001 65 , Academic Press 20 23
Yang J Liu X Bhalla K Kim CN Ibrado AM Cai J Peng TI Jones DP Wang X Prevention of apoptosis by Bcl-2: release of cytochrome c from mitochondria blocked Science 1997 275 1129 1132 9027314 10.1126/science.275.5303.1129
Bibb MJ Van Etten RA Wright CT Walberg MW Clayton DA Sequence and gene organization of mouse mitochondrial DNA Cell 1981 26 167 180 7332926 10.1016/0092-8674(81)90300-7
Heerdt BG Chen J Stewart LR Augenlicht LH Polymorphisms, but lack of mutations or instability, in the promotor region of the mitochondrial genome in human colonic tumors Cancer Res 1994 54 3912 3915 8033115
Parfait B Rustin P Munnich A Rotig A Co-amplification of nuclear pseudogenes and assessment of heteroplasmy of mitochondrial DNA mutations Biochem Biophys Res Commun 1998 247 57 59 9636653 10.1006/bbrc.1998.8666
Ecsedy JA Yohe HC Bergeron AJ Seyfried TN Tumor-infiltrating macrophages influence the glycosphingolipid composition of murine brain tumors J Lipid Res 1998 39 2218 2227 9799808
Meixensberger J Herting B Roggendorf W Reichmann H Metabolic patterns in malignant gliomas J Neurooncol 1995 24 153 161 7562002 10.1007/BF01078485
Pedersen PL Tumor mitochondria and the bioenergetics of cancer cells Prog Exp Tumor Res 1978 22 190 274 149996
Cuezva JM Chen G Alonso AM Isidoro A Misek DE Hanash SM Beer DG The bioenergetic signature of lung adenocarcinomas is a molecular marker of cancer diagnosis and prognosis Carcinogenesis 2004 25 1157 1163 14963017 10.1093/carcin/bgh113
Floridi A Paggi MG Fanciulli M Modulation of glycolysis in neuroepithelial tumors J Neurosurg Sci 1989 33 55 64 2674359
Mangiardi JR Yodice P Metabolism of the malignant astrocytoma Neurosurgery 1990 26 1 19 2403663 10.1097/00006123-199001000-00001
Ikezaki K Black KL Conklin SG Becker DP Histochemical evaluation of energy metabolism in rat glioma Neurol Res 1992 14 289 293 1360622
Aisenberg AC The Glycolysis and Respiration of Tumors 1961 New York, Academic Press 224
Jazin EE Cavelier L Eriksson I Oreland L Gyllensten U Human brain contains high levels of heteroplasmy in the noncoding regions of mitochondrial DNA Proc Natl Acad Sci U S A 1996 93 12382 12387 8901590 10.1073/pnas.93.22.12382
Taylor RW Taylor GA Durham SE Turnbull DM The determination of complete human mitochondrial DNA sequences in single cells: implications for the study of somatic mitochondrial DNA point mutations Nucleic Acids Res 2001 29 E74 4 11470889 10.1093/nar/29.15.e74
|
16105171
|
PMC1199588
|
CC BY
|
2021-01-04 16:03:06
|
no
|
BMC Cancer. 2005 Aug 16; 5:102
|
utf-8
|
BMC Cancer
| 2,005 |
10.1186/1471-2407-5-102
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-781602661010.1186/1471-2407-5-78Research ArticleCancer cell adaptation to chemotherapy Di Nicolantonio Federica [email protected] Stuart J [email protected] Louise A [email protected] Francis G [email protected] Pauline A [email protected] Sanjay [email protected] Augusta [email protected] Sharon [email protected] Palma Silvana [email protected] Penny [email protected] Shaw S [email protected] Simon [email protected] Bernie [email protected] Alan [email protected] Tim [email protected] Jeremy [email protected] Constantinos [email protected] Ian A [email protected] Translational Oncology Research Centre, Department of Histopathology, Queen Alexandra Hospital, Portsmouth PO6 3LY, UK2 Department of Surgery, Queen Alexandra Hospital, Portsmouth PO6 3LY, UK3 Department of Mathematics and Statistics, University of Portsmouth, Buckingham Building, Lion Terrace, Portsmouth PO1 3HE, UK4 Department of Radiotherapy and Oncology, Southend Hospital, Prittlewell Chase, Westcliff-on-Sea, Essex SS0 0RY, UK5 Department of Radiotherapy and Oncology, St Mary's Hospital, Milton Road, Portsmouth PO3 6AD, UK2005 18 7 2005 5 78 78 23 11 2004 18 7 2005 Copyright © 2005 Di Nicolantonio et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Tumor resistance to chemotherapy may be present at the beginning of treatment, develop during treatment, or become apparent on re-treatment of the patient. The mechanisms involved are usually inferred from experiments with cell lines, as studies in tumor-derived cells are difficult. Studies of human tumors show that cells adapt to chemotherapy, but it has been largely assumed that clonal selection leads to the resistance of recurrent tumors.
Methods
Cells derived from 47 tumors of breast, ovarian, esophageal, and colorectal origin and 16 paired esophageal biopsies were exposed to anticancer agents (cisplatin; 5-fluorouracil; epirubicin; doxorubicin; paclitaxel; irinotecan and topotecan) in short-term cell culture (6 days). Real-time quantitative PCR was used to measure up- or down-regulation of 16 different resistance/target genes, and when tissue was available, immunohistochemistry was used to assess the protein levels.
Results
In 8/16 paired esophageal biopsies, there was an increase in the expression of multi-drug resistance gene 1 (MDR1) following epirubicin + cisplatin + 5-fluorouracil (ECF) chemotherapy and this was accompanied by increased expression of the MDR-1 encoded protein, P-gp. Following exposure to doxorubicin in vitro, 13/14 breast carcinomas and 9/12 ovarian carcinomas showed >2-fold down-regulation of topoisomerase IIα (TOPOIIα). Exposure to topotecan in vitro, resulted in >4-fold down-regulation of TOPOIIα in 6/7 colorectal tumors and 8/10 ovarian tumors.
Conclusion
This study suggests that up-regulation of resistance genes or down-regulation in target genes may occur rapidly in human solid tumors, within days of the start of treatment, and that similar changes are present in pre- and post-chemotherapy biopsy material. The molecular processes used by each tumor appear to be linked to the drug used, but there is also heterogeneity between individual tumors, even those with the same histological type, in the pattern and magnitude of response to the same drugs. Adaptation to chemotherapy may explain why prediction of resistance mechanisms is difficult on the basis of tumor type alone or individual markers, and suggests that more complex predictive methods are required to improve the response rates to chemotherapy.
==== Body
Background
Tumor resistance to chemotherapy is a well-known clinical phenomenon that is now yielding its secrets to investigation at the molecular level in biopsy material. Studies in cell lines do not always correlate well with results from tumor tissue [1], which may consist largely of non-neoplastic cells that support and modify the biology of neoplastic cells. Thus it is important to validate the mechanisms important in vitro with the situation in the patient. Nevertheless, cell line studies and immunohistochemical studies of tumors suggest that resistance is a selective process: only those cells that survive a drug-induced insult will re-grow.
We have previously shown development of such resistance to combination chemotherapy in tumor-derived cells from matched biopsies collected from breast cancer patients before and after administration of doxorubicin-containing chemotherapy [2]. In this study we show similar results in patients with esophageal cancer from biopsies obtained prior to and several months after chemotherapy. Two cycles of the combination of epirubicin, cisplatin and 5-FU (ECF) are given to these patients prior to resection, allowing studies to be performed with paired samples before and after chemotherapy. We have used real-time quantitative RT-PCR (qRT-PCR) and immunohistochemistry (IHC) to assess targets known to be of importance to resistance to these agents.
The mechanisms involved in resistance to chemotherapy usually involve up-regulation of resistance mechanisms, or down-regulation of target genes. Examples of the former include drug efflux pump molecules such as multi-drug resistance gene 1/P-glycoprotein (MDR1/P-gp), while the latter include topoisomerases (TOPOs), targets of drugs such as etoposide and doxorubicin. Many papers attest to the importance of clonal selection in this process: it is for instance possible to expose cell lines to low concentrations of drugs and, over time, to produce highly resistant sub-clones [3]. However, there is another potential mechanism that does not require clonal selection: cells may be able to adapt by regulation of expression of resistance or target molecules individually if they survive the initial exposure to the drug. This could be a more rapid process and would require changes in molecular expression, possibly due to epigenetic change, rather than genetic mechanisms such as mutation [4]. As a result, resistance may therefore arise rapidly following treatment with chemotherapy.
Recent studies have shown that the expression of MDR1/P-gp is up-regulated within hours of anti-cancer drug treatment in vivo in patient samples [5-8], although this effect was not observed in all patients. We therefore wished to examine how quickly and in how many cases these resistance molecules were up-regulated in tumor-derived cells from several tumor types. We have used selective short-term cell culture (6 days) to examine the changes in expression that occur following exposure to chemotherapy compared to medium-only control cells from the same samples. Our short-term culture system employs a serum-free medium and polypropylene 96 'U' well microplates. This inhibits the proliferation and survival of normal cells and allows selective survival of a neoplastic cell population [9]. The short incubation period also limits the possibility of selection of clones or sub-populations in vitro. However, the presence of non-neoplastic cells for most of the incubation period allows the interaction between stromal cells and neoplastic cells, a factor that appears to be important to maintain the chemosensitivity profile [10].
Methods
Patients and tissue samples for in vitro studies
Tumor derived cells were obtained from 17 breast cancer patients (16 primaries; 1 pre-treated with mitoxantrone and paclitaxel), 13 ovarian cancer patients (all pre-treated with a cisplatin-based regimen), 10 colorectal cancer patients (all primaries) and 7 esophageal cancer patients (3 untreated; 4 treated with ECF). Cells were grown for 6 days in serum-free medium with or without drugs, before RNA extraction and further PCR analysis.
Patients and tissue samples for in vivo study
Thirty-four esophageal adenocarcinoma biopsies, thirty-two of which were paired samples were obtained from patients (17M:1F; median age 57, range 42–81) before and after administration of 2 cycles of ECF chemotherapy. After enzymatic digestion, tumor derived cells were centrifuged over Ficoll (Sigma Chemical Co, Poole, UK, Cat. No. 1077-1) to remove blood contaminating cells, washed in PBS, and stored in RNA later (Ambion, Huntingdon, UK) at -80°C until further molecular analysis was performed. Tissue sections from these samples were stained for GST-π, MRP1, P-gp and TS. All tumor samples were removed as part of patient treatment, with consent for tissue donation and local research ethics committee approval for use of the tissue surplus to diagnostic requirements for cellular and molecular assays. Chemosensitivity data were available for 9 patients with recurrent ovarian cancer, before treatment and on relapse, though in only one case was material available from both samples for quantitative RT-PCR.
Drugs
Cisplatin, 5-fluorouracil (5-FU), epirubicin, doxorubicin, irinotecan, paclitaxel (Taxol®) and topotecan (Hycamptin®) were obtained from the pharmacy at Queen Alexandra Hospital (Portsmouth, UK). Cisplatin, 5-FU, irinotecan and paclitaxel were stored at room temperature, while all other drugs were stored at -20°C, as previously reported [11]. Test drug concentrations (TDC) were 10.0 μM for cisplatin, 345 μM for 5-FU, 148 μM for irinotecan, 15.9 μM for paclitaxel, 2.5 μM for doxorubicin, 0.862 μM for epirubicin, and 1.64 μM for topotecan. Combinations were made up by adding two or three drugs concurrently at their 200% TDC at the beginning of the ATP-cell viability assay and diluted in a constant ratio: sequential studies were not performed.
Short-term cell culture
Briefly, tumor tissue or fluid was taken by a histopathologist or surgeon under sterile conditions, and transported to the laboratory in cell culture medium of Dulbecco's modified Eagle's medium (DMEM; Sigma Cat No. D5671) with antibiotics (100 U/ml penicillin and 100 μg/ml streptomycin, Sigma, Cat No. P0781) at 4°C. Cells were obtained from solid tumors by enzymatic dissociation, usually 0.75 mg/ml collagenase (Sigma Cat No. C-8051) overnight. Viable tumor-derived cells were purified by density centrifugation (Histopaque 1077-1, Sigma), washed, counted and resuspended to 100,000 cells/ml in case of effusions or 200,000 cells/ml for solid biopsies. In the meantime 96-well polypropylene microplates (Corning-Costar, High Wycombe, UK; Cat No. 3790) were prepared with each drug/combination at six doubling dilutions in triplicate from 200% TDC to 6.25% TDC, according to Andreotti et al. [9]. Approximately 10,000–20,000 cells/well were added to the plates to a final volume of 200 μl/well. The plates were then incubated at 37°C in 5% CO2 for 6 days, after which the degree of cell inhibition was assessed by measurement of the remaining ATP in comparison with negative control (no drug, MO) and positive control (maximum inhibitor, MI) rows of 12 wells each. Prior to cell lysis with an ATP-extracting reagent, an aliquot of 150 μl of cell suspension were removed from each well, centrifuged, washed with phosphate buffered saline (PBS) and stored at -80°C in a GTIC-containing solution (lysis buffer RA1, Macherey-Nagel, Düren, Germany; Cat. No.740961) until further molecular analysis was carried out. The RNA was subsequently extracted from aliquots that had been exposed to a drug concentration capable of inhibiting cell growth by 40–60%. ATP was extracted from the remaining 50 μl cell suspension and measured by light output in a microplate luminometer (Berthold Diagnostic Systems GmbH, Pforzheim, Germany) following addition of luciferin-luciferase.
RNA extraction
Cells obtained after enzymatic dissociation from endoscopic esophageal biopsies or short term cell culture were either resuspended in RNA later (Ambion, Huntingdon, UK; Cat No. 7020) or lysed with buffer RA1 and stored at -80°C until RNA extraction. Total RNA was extracted from at least 50,000 cells with a commercially available kit (NucleoSpin® RNA II mini, Macherey-Nagel; Cat No. 740955) according to the manufacturer's instructions. The protocol included a DNase digestion step to prevent carry-over of genomic DNA in further analysis.
qRT-PCR
A two-step protocol was employed. Firstly, total RNA was reverse-transcribed using the Promega reverse transcription system (Promega, Southampton, UK; Cat No. A3500) including 8 μl RNA, 0.5 μg random primers, 20 units of recombinant RNasin® ribonuclease inhibitor and 15 units of reverse transcriptase AMV (Promega, Cat No. M9004) to each 20 μl reaction. The resulting c-DNA was amplified by qPCR on a Biorad iCycler instrument (BioRad Laboratories, Hemel Hampstead, UK). The constituents of each PCR reaction (25 μl) were 1 μl of cDNA (or H2O), 200–500 nM of each primer (Table 1), 200 μM each dATP, dCTP, dGTP, 400 μM dUTP, 3.0–5.0 mM MgCl2, 0.125 units AMPErase® UNG, 0.625 units of AmpliTaq Gold DNA polymerase and 1 × SYBR Green PCR buffer (all reagents were from Applied Biosystems, Warrington, UK). Product amplification was performed up to 45 PCR cycles, after uracil removal (2 min at 50°C) and polymerase activation (10 min at 95°C). Each two-step PCR cycle comprised denaturing (15 s at 95°C), annealing, and extending (1 min at 60°C). At the end of each run a final melt curve cycle (cooling to 50°C and then increasing stepwise 1°C to 95°C) was performed to exclude the presence of primer-dimer artefacts. At least 3 housekeeping genes were used for each experiment chosen from the following: glyceraldehyde-3 phosphate dehydrogenase (GAPDH), hypoxanthine phosphoribosyltransferase 1 (HPRT1), human porphobilinogen deaminase (PBGD), succinate dehydrogenase complex-subunit A (SDHA) and TATA box binding protein (TBP). The internal reference genes were selected due to their relative low abundance in normal tissue [12]. The housekeeping genes were amplified parallel to the target genes in separate wells. When possible, primer sequences (Table 1) were chosen to span exon boundaries to the following targets: breast cancer resistance protein (BCRP), dihydropyrimidine dehydrogenase (DPD), epidermal growth factor receptor (EGFR), excision repair cross-complementing 1 (ERCC1) gene, glutathione-S-transferase isoform π (GST-π), multi-drug resistance gene 1 (MDR1), MLH1, multi-drug resistance related protein 1 (MRP1), multi-drug resistance related protein 2 (MRP2), metallothionein (MT), major vault protein (MVP), thymidine phosphorylase (TP), thymidylate synthase (TS), topoisomerase I (TOPO I), topoisomerase IIα and β (TOPO IIα and TOPO IIβ). Each amplicon was amplified in a separate reaction to prevent competition among multiple sets of primers. A positive control (pooled c-DNA from a variety of human tumors, including breast, ovarian, colorectal and esophageal carcinoma) and negative controls with no template and RT-negative as template were added in every experiment. All assays were run in triplicate. Validation experiments were run to show that the efficiencies of the target and reference genes amplifications were approximately equal, and in the range 95–105%. The PCR cycle number that generated the first fluorescence signal above a threshold (threshold cycle, Ct; 10 standard deviations above the mean fluorescence generated during the baseline cycles) was determined, and a comparative Ct method was then used to measure relative gene expression [13]. The following formula was used to calculate the relative amount of the transcript in the sample: 2-ΔΔCt, where ΔCt is the difference in Ct between the gene of interest and the mean of the at least two reference genes, and ΔΔCt = ΔCt of drug non-exposed cells – ΔCt of drug-exposed cells, for the ex-vivo experiments, or ΔΔCt = ΔCt of pre-chemotherapy sample – ΔCt of post-chemotherapy sample, for matched tumor biopsies.
Immunohistochemistry
The monoclonal antibodies (Abs) for GST-π, MRP1, Pg-p, and TP (Table 2) were detected using the Chemicon IHC Select™ – Immuno Peroxidase secondary detection system (Chemicon International, Chandlers Ford, Southampton, UK, Cat# Det-HP1000) and stained with 3,3' diaminobenzidine (DAB; HD Supplies, Aylesbury, UK, Cat. 4170) on a Dako Autostainer instrument (Dako, Ely, UK). Positive and negative controls were included with each run. Volumes of 50 μl avidin/ml goat serum and 50 μl biotin/ml primary Ab were used to block endogenous avidin binding. The slides were counterstained with Gills Haematoxylin, dehydrated and cleared using the Leica© XL slide staining machine and mounted in Styrolite® mounting medium (BDH, Poole, Dorset, UK; Cat No. 361704Y).
Data analysis
The luminometer readings obtained from the ATP-TCA were entered into an Excel 2000 spreadsheet (Microsoft) which calculated the percentage of cell inhibition for each drug concentration according to the previous published formula: 1 - [(Test - MI)/(MO - MI)]*100 [9]. For each drug-response curve, the 50% inhibitory concentration (IC50) and the 90% inhibitory concentration (IC90) were also calculated as previously described [14].
Assessment of slides was done using the H-score. Staining intensity (none, 0 points; weak, 1 point; moderate, 2 points; strong, 3 points) and percentage of positive tumor cells were multiplied to achieve a score between 0 and 300. A H-score of 100 or more was regarded as positive and results less than 100 were regarded as negative. The correlation coefficients were calculated by the method of the least squares, and the correlation between the IC90 and IC50 values and immunohistochemistry indices was assessed using univariate linear regression (Statsdirect, Sale, UK).
Non-parametric statistical methods were used. The calculated and descriptive data were entered into an Access 2000 database (Microsoft) and analysed using a Wilcoxon two-tailed paired rank sum test for paired data or the Mann-Whitney U test for unpaired data, as appropriate (Statsdirect). IC50 and IC90 values for each compound were correlated to the relative mRNA levels of target genes using Spearman's rank correlation coefficient and multivariate analysis. On statistical advice, we chose not to use a Bonferroni's correction, but it should be noted that some technically statistically significant results could have arisen by chance.
Results
Chemotherapy-induced changes in mRNA levels in biopsy material
Figures 1a–e illustrate changes to IC50 μM values in pre- and post chemotherapy ovarian tumor-derived cells tested with a variety of chemotherapeutic agents. There is a general increase in resistance to doxorubicin when anthracycline-based regimens (mitoxantrone + paclitaxel or liposomal doxorubicin) are administered to ovarian cancer patients (Fig. 1a). We observed little cross-resistance for mitoxantrone, topotecan, paclitaxel, and cisplatin following anthracycline-based chemotherapy (Fig. 1b–e), [15], although some patients do show alteration of sensitivity to other agents. Similar changes can be seen in breast cancer biopsies before and after administration of a chemotherapy regimen [2]. The increase in resistance to doxorubicin shown in Figure 1a was accompanied by changes in drug resistance gene expression. Figure 1f highlights such up-regulation of MDR1 and breast cancer resistance protein (BCRP) observed in ovarian-tumor derived cells from a single patient.
Membrane proteins such as MDR1/P-gp, multi-drug resistance protein 1 (MRP1) and major vault protein (MVP) have been demonstrated to confer resistance to epirubicin by pumping the cytotoxic drug out of cells [16]. Paired esophageal samples were obtained from the same six patients both before and after chemotherapy; in all 6 biopsy pairs (Table 3), there was an increase of the mRNA levels of MDR1 following ECF chemotherapy (Fig. 2a). In a larger series of 20 esophageal samples (including those that did not have a paired biopsy), divided in two groups according to the patients' exposure to treatment, there was a 5-fold increase in MDR1 expression in the post-chemotherapy group (median expression index 0.07 pre- and 0.37 post-chemotherapy, p < 0.0001, Mann-Whitney non-parametric unpaired U test). The paired samples from patients before and after ECF chemotherapy also showed a trend towards increased expression of MRP1 and MVP (Table 3), with three tumors showing concomitant up-regulation of both genes. Furthermore, the increased expression of MRP1 paralleled that of glutathione-S-transferase isoform π (GST-π) in 2 of these 3 samples, consistent with the mechanism of detoxification of MRP1 that involves transport of glutathione-conjugated molecules [17].
We then determined the relative mRNA levels of enzymes that have previously been correlated with 5-FU sensitivity: dihydropyrimidine dehydrogenase (DPD), thymidine phosphorylase (TP) and thymidylate synthase (TS) [18]. No significant difference was found for these 3 genes when we compared their expression pre- and post-chemotherapy in unpaired samples (Table 3). However, we found a modest increase of TS expression in 5/6 paired samples (Fig. 2b), 3 of which also showed a concomitant decrease in TP levels. One sample showed a paradoxical decrease in TS, indicating some heterogeneity. Although the sample size is small, the trend of increased TS levels in ECF exposed tumors is consistent with a number of previous reports that demonstrated an acute induction of TS expression in cell lines, animal models and human tumors following 5-FU treatment [18,19].
It should be noted that these qRT-PCR results were all obtained from samples that included normal cells present in the tumor as well as neoplastic cells, and could be affected by changes in normal cells as well as malignant cells.
Chemotherapy-induced changes in protein levels in biopsy material
IHC was performed on 16 paired esophageal biopsies: qRT-PCR data were obtained for four of these pairs. Overall, we were able to confirm an increased expression of the MDR-1 encoded protein, P-gp (Table 4), in 8/16 paired tumor samples obtained from patients who had been exposed to ECF chemotherapy. TS positivity (Table 4) was found in 10/16 samples obtained before treatment and 10/16 post-chemotherapy specimens; in this instance the modest induction of TS seen at the mRNA level did not reflect an increase in protein level (data not shown). We found GST-π (Table 4) positivity in 5/16 pre-chemotherapy biopsies. In the post-chemotherapy specimens we detected strong expression of GST-π in all but one of the samples. In one case that had been negative at diagnosis we measured a marginal increase of mRNA levels with qRT-PCR. A total of 13/16 pre- and 12/16 post-chemotherapy biopsies were found to be positive for MRP1 (Table 4). Overall, these data show little correlation between the mRNA and the degree of expression of protein levels determined by IHC. Confounding factors may be the presence of non-cancerous cells as mentioned above, but also the nature of IHC, which is at best a semi-quantitative technique.
Changes in short-term cell culture
Of the 47 solid tumor samples studied in short-term cultures, 7 were esophago-gastric, 17 were breast carcinomas, 13 were ovarian carcinomas and 10 were colorectal tumors. A total of 93 experiments were performed, as in most cases the same sample was treated with 2 or more drugs.
Doxorubicin effects in short-term cell culture
In the first instance, we decided to study the effects of anthracycline exposure for 6 days on tumour-derived cells. Our in vitro experiments showed a significant down-regulation of TOPO IIα; its median levels decreased from 0.546 (range 0.005–1.782) to 0.017 (range 0.0004–0.595) units in a group of 14 breast samples (p < 0.0001, Wilcoxon matched pairs test) and from 1.491 (range 0.080–6.884) to 0.089 (range 0.003–3.605) units in the ovarian cancer subgroup (p < 0.0015, Wilcoxon matched pairs test). The down-regulation of TOPO IIα levels was greater than 2-fold in 13/14 breast samples and in 9/12 ovarian samples (Fig. 3a). It has been suggested that cancer cells can concomitantly down-regulate TOPO IIα and up-regulate TOPO I in response to treatment with topoisomerase II inhibitors [20]. In our series we noted a general trend of diminished levels of TOPO I, particularly in the ovarian subgroup (Fig 3b). It is possible that doxorubicin exposure resulted in quiescence of cells which were not killed, and which would not therefore require higher levels of TOPO I, or that doxorubicin may also partly inhibit TOPO I, as suggested by a previous study [21].
We observed a significant up-regulation of MRP-1 (p < 0.0001, Wilcoxon matched pairs test; Fig. 3c) and of MVP (p < 0.0023, Wilcoxon; Table 5) in the breast cancer subgroup. The same effect was not found in the ovarian cancer group, in which the levels of MRP-1 and MVP increased over 0.5 fold in only 4/12 and 3/12 samples, respectively (Table 5). No significant changes were found in the expression of MDR-1 and BCRP, although it should be noted that there was substantial heterogeneity among individual tumors (Table 5). For example, we measured a greater than 2-fold increase of MDR-1 levels in 2/12 ovarian samples (none of whom had received a MDR-1 pumped chemotherapeutic agent before), and a significant up-regulation of BCRP in 3/12 ovarian samples.
Topoisomerase I inhibitors effects in short-term cell culture
Subsequently we looked at the effect of 2-camptothecin derivatives, irinotecan and topotecan, on colorectal and ovarian tumor cells, respectively (Table 6). As both compounds act by TOPO I inhibition, we firstly measured the levels of their target enzyme. We noted a trend towards down-regulation of TOPO I in treated cells: exposure to irinotecan decreased TOPO I levels >2-fold in 3/7 colorectal tumors, while topotecan caused down-regulation in 4/10 ovarian samples (Fig. 3d). The decrease of TOPO I was accompanied by a concomitant reduction of TOPO IIα expression, which was particularly pronounced (>4-fold) in 6/7 colorectal tumors and 8/10 ovarian tumors (Fig. 3e).
No significant changes were observed in the expression of the drug efflux molecules, MDR-1, BCRP and MRP-1 (Table 6), though, as in the case of doxorubicin, considerable heterogeneity was noted. We observed an increase of BCRP levels after irinotecan or topotecan exposure in 2/7 colorectal samples and 3/9 ovarian samples.
Amongst the genes implicated in DNA repair (Table 6), the modest down-regulation of MLH1 by topotecan exposure was not found statistically significant, although it was noted in 7/10 ovarian samples. Up-regulation of ERCC1 expression was found in all 10 ovarian cancer samples exposed to topotecan (p < 0.002, Wilcoxon), and in all 7 colorectal specimens treated with irinotecan, although in this group the increase was modest (p = 0.016, Wilcoxon).
Finally, we looked at EGFR as a marker of tumor growth and progression. After topotecan exposure we found a decrease (more than 2-fold) in the expression of this growth factor in 9/10 ovarian samples (Table 6), with the median levels decreasing from 0.470 to 0.163 units (p < 0.0039, Wilcoxon).
5-FU effects in short-term cell culture
Preliminary data obtained from 2 esophageal biopsies indicated an increase of TS levels following 5-FU treatment. Subsequent in vitro experiments performed on material derived from 10 colorectal tumor samples and 13 breast tumor samples confirmed the expected increase of TS levels (Table 7) in the samples exposed to 5-FU (p < 0.0001, Wilcoxon; Fig. 3f). The results indicated a general trend towards DPD down-regulation in the cells that had been exposed to 5-FU in vitro, though this effect was modest in most cases and we observed a greater than 2-fold decrease in only 7/13 breast samples and 2/10 colorectal samples.
Cisplatin effects in short-term cell culture
We examined the mRNA expression levels of genes previously highlighted as important in cisplatin resistance in tumor-derived cells from 13 breast and 7 ovarian tumors (Table 8). Up-regulation of the nucleoside excision repair (NER) gene ERCC1 has previously been shown to increase the removal of platinum affected DNA [22]. We observed a significant increase in ERCC1 levels in both tumor-derived breast (p = 0.0266, Wilcoxon matched pairs test) and ovarian cells (p = 0.0156, Wilcoxon matched pairs test) (Fig 3g). We also observed significant down regulation of a second DNA repair gene, MSH2 in tumor-derived breast cells (p = 0.0479, Wilcoxon matched pairs test), but not in tumor-derived ovarian cells. Samimi et al., [23] have previously demonstrated that following cisplatin therapy, there is selection for cells expressing lower hMLH1 and hMSH2. However, no significant changes were observed in the other mismatch repair genes, MLH1 and MSH6 in either tumor type (Table 8).
There was no significant change in MGMT in tumor-derived breast or ovarian cells. However, it has previously been shown that MGMT mRNA levels begin to recover after 24 hours in the absence of drug [24]. There was a paradoxical decrease in the copper export pump ATP7B following cisplatin exposure (Table 8), in tumor-derived breast cells (p = 0.0327, Wilcoxon matched pairs test), but not in ovarian cells. No significant changes were observed in the heavy-metal binding protein MTII, in either tumor type.
ECF effects in short-term cell culture
We were able to study tumor-derived cells from 7 esophageal cancer patients, 4 of which had already been given ECF in vivo. The results mirror those obtained from paired esophageal biopsies: we noticed increased expression of TS in the cells that had been exposed to ECF (Fig 3h). However, the expected up-regulation of MDR1 was only detected in 4/7 samples (data not shown), and may therefore be occurring in non-neoplastic cells that do not normally express MDR1, rather than in the neoplastic cells, which we found to express this molecule to the same degree pre- and post-treatment.
Correlation of in vitro cytotoxicity with molecular expression
Lastly, for 5-FU, doxorubicin and irinotecan, IC90 and IC50 data were obtained by measuring the ATP levels in a small aliquot of cell suspension at the end of the incubation period. This allowed a comparison to be made between drug sensitivity and the expression of putative resistance genes in the cells that had been exposed to chemotherapeutic agents. Using multivariate analysis, the IC90 of 5-FU was correlated with the median change of mRNA levels of both TS and DPD measured in 5-FU treated tumor-derived cells compared to control cells (expressed as 2-ΔΔCt) (R2 = 0.872704; p = 0.0006 for DPD; p < 0.0001 for TS). In addition, expression of ERCC1 mRNA correlated with the IC50 values determined for doxorubicin in 11 breast samples (R = 0.7204, p < 0.0124). No other correlations were noted between sensitivity to the drugs and gene expression levels.
Discussion
Our results suggest that rapid adaptation to chemotherapy may result in a resistant phenotype. This is mediated by down- or up-regulation of genes that are usually correlated to the mechanism of action of the individual chemotherapeutic agent or relevant resistance mechanisms. Our data suggest that short-term cell culture of tumor-derived cells with drugs could provide a suitable model for studying resistance mechanisms. The mechanisms observed appeared to be more specific to the drug used than to the tumor type: constitutive resistance probably reflects pre-chemotherapy expression of resistance mechanisms (e.g. drug efflux molecule expression in esophageal carcinoma). Acquired resistance develops rapidly and is likely to reflect changes in gene regulation rather than mutation-dependent selection of clones. Clonal selection may be important in some rapidly growing tumors and cell lines grown in serum-containing media, but is unlikely to be the major factor in solid tumors which have relatively low doubling rates. While mutation-mediated resistance can be much more profound than that observed here, our data suggest that the functional effects may still be sufficient to render the patient's tumor resistant to treatment within one cycle of chemotherapy.
Changes in resistance and target molecules
A large proportion of the published studies on resistance to chemotherapy have investigated the development of resistance using cell lines generated in the lab after prolonged and step-wise exposure to anti-cancer drugs. These in vitro models are not necessarily representative of the in vivo situation, when patients are usually administered one cycle of chemotherapy every 3–4 weeks. There are few studies in clinical samples. Our approach allows us to expose the tumor cells to single drugs under carefully controlled conditions, even if this would be an inappropriate drug for that particular patient. We are then able to look concomitantly at cytotoxicity, and molecular markers of resistance in the same experiment.
Anthracyclines have a mechanism of action that includes TOPO IIα inhibition via DNA intercalation. Resistance to anthracyclines is thought to be mediated by a number of different mechanisms, which include mutation or alteration of its target enzyme, TOPO IIα, and up-regulation of drug efflux proteins, such as BCRP, MRP1, MVP and MDR-1 [25]. Our data show that many of the cell line data are correct: in most solid tumors, it appears that anthracycline and topotecan exposure do lead to decreased topoisomerase II and I expression respectively, while inducing the expression of drug efflux pump molecules. The reason for the alteration in topoisomerase II expression following topotecan exposure is not clear, but topotecan does affect cell proliferation, and any reduction in proliferation would indirectly affect the expression of topoisomerase II alpha induced during S phase. However, our results also suggest that the heterogeneity of chemosensitivity between tumors is reflected by heterogeneity of molecular determinants of resistance/sensitivity.
The increased TS levels in ECF and 5-FU exposed cells is consistent with a number of previous reports. Gene amplification of TS with consequent increases in TS mRNA and protein has been observed in cell lines that are resistant to 5-FU and fluorodeoxyuridine (FUDR) [26,27]. Treatment with 5-FU has been shown to acutely induce TS expression in cell lines, animal models and human tumors [19,28-30]. In general there is strong evidence that the expression of DPD and TS in GI cancers is predictive of response to 5-FU. It should be noted that the concomitant measurement of both these markers markedly enhanced the ability to predict tumor response to 5-FU-based chemotherapy in a number of studies [31-34].
Gene profiling and drug response
There are few studies comparing gene expression before and after chemotherapy. A few studies have reported microarray data in biopsy material taken before and after (or even during) chemotherapy. These studies are of great interest, though normal cell effects cannot be excluded from the results. Buchholz et al. [35] employed cDNA microarray to measure gene expression changes during chemotherapy in 5 patients with breast cancer. Clarke et al. [36] studied gene expression changes in 18 rectal cancer patients undergoing therapy with Mitomycin C or 5-FU. This study reported a number of genes implicated in protein synthesis and RNA metabolism to be significantly decreased during drug treatment. These studies are not directly comparable: tumors of different types respond better to different drugs, and differences in their adaptation to these drugs are therefore expected on the basis of their innate sensitivity or resistance.
The potential for positive selection: molecular chess
It is common to show a cross-over effect with clinical trials of treatments with differing mechanisms of action, in which patients treated with one type of chemotherapy show sensitivity to the alternative regimen following failure of the one to which they were allocated. The recognition that selection of a molecular phenotype by exposure to one anti-cancer agent may leads to the expression of molecular targets for other drugs raises the possibility that it might be possible to enhance sensitivity to second-line or maintenance therapy by careful selection of patients for first-line therapy [37]. This approach would provide patients with a "backstop" for their first-line chemotherapy. One can envisage a series of interlocking treatments using drugs with specific molecular targets, monitored by molecular assays, which would allow the oncologist to employ a form of molecular chess to defeat the tumor. This approach might overcome the inherent heterogeneity, which is likely to underlie the variable results obtained from sequential chemotherapy to date. Assessment of this process in tumors could provide predictive assays allowing the oncologist to tailor therapy to the patient and avoid the development of resistance within the tumor.
Conclusion
In summary, this study suggests that up-regulation of resistance genes or down-regulation in target genes may occur rapidly in human solid tumors, within days of the start of treatment, and that similar changes are present in pre- and post-chemotherapy biopsy material. The molecular processes used by each tumor appear to be linked to the drug used, but there is heterogeneity between individual tumors, even those with the same histological type, in the pattern and magnitude of response to the same drugs. Adaptation to chemotherapy may explain why prediction of resistance mechanisms is difficult on the basis of tumor type alone or individual markers, and suggests that more complex predictive methods are required to improve the response rates to chemotherapy.
List of abbreviations
Gene names were abbreviated as follows: glyceraldehyde-3 phosphate dehydrogenase (GAPDH), hypoxanthine phosphoribosyltransferase 1 (HPRT1), human porphobilinogen deaminase (PBGD), succinate dehydrogenase complex-subunit A (SDHA) and TATA box binding protein (TBP), breast cancer resistance protein (BCRP), dihydropyrimidine dehydrogenase (DPD), excision repair cross-complementing 1 (ERCC1) gene, epidermal growth factor receptor (EGFR), glutathione-S-transferase isoform π (GST-π), multi-drug resistance gene 1 (MDR1), mutL homologue 1 (MLH1), multi-drug resistance related protein 1 (MRP1), multi-drug resistance related protein 2 (MRP2), metallothionein II (MT II), major vault protein (MVP), thymidine phosphorylase (TP), thymidylate synthase (TS), topoisomerase I (TOPO I), topoisomerase IIα and β (TOPO II).
Competing interests
IAC is a director of CanTech Ltd. The remaining authors declare that they have no competing interests.
Authors' contributions
IAC, FDN and SJM conceived and designed the study. FDN, SJM, LAK, PAW, SS, AF and SG participated in the short-term cell culture studies. FDN, SJM, LAK and FGG also carried out the qRT-PCR studies. PJ carried out the immunohisotochemical studies and IAC and SDP carried out histological analysis. IAC, FDN, SJM and BH participated in the statistical analysis. All authors participated in the data analysis, drafting of the manuscript and read and approved the final version.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This project was funded by Portsmouth Hospitals NHS Trust, CanTech Ltd, the BBSRC (Ref 31/ABY14513), the Royal Navy (SJM), the European Commission (grant number BMH4-CT98-9522; FDN), and was supported by a donation from Schering Plough Ltd (LAK). We are grateful to the NHS predictive oncology programme and all the patients, oncologists and surgeons who submitted material for chemosensitivity testing to make this study possible. We thank Christine Seddon for assistance with data entry.
Figures and Tables
Figure 1 Changes due to chemotherapy in biopsies taken before and after chemotherapy. Cytotoxicity of a) doxorubicin; b) mitoxantrone; c) topotecan; d) paclitaxel and e) cisplatin (expressed as IC50 μM) in paired samples obtained from ovarian cancer patients before and after they were treated with an anthracycline containing regimen (paclitaxel+mitoxantrone n = 7; liposomal doxorubicin n = 2). Each line represents an individual patient. (f) Gene expression changes pre- and post-treatment (paclitaxel plus mitoxantrone chemotherapy) in one patient were analysed by qRT-PCR following in vitro exposure to doxorubicin.
Figure 2 Changes in relative gene expression in 6 paired esophageal tumor biopsies. a) Changes in MDR1 relative gene expression and b) changes in TS relative gene expression. Each dot represents the relative mRNA level for an individual tumor, measured before and after ECF chemotherapy.
Figure 3 Changes in relative expression of putative chemoresistance genes in tumor-derived cells. Each dot represents the relative mRNA level for an individual sample, measured after in vitro drug exposure compared with untreated control cells. (a) TOPO IIα expression in breast (black lines) and ovarian (red lines) tumor cells after doxorubicin exposure. (b) TOPO I expression in breast (black lines) and ovarian (red lines) tumor cells after doxorubicin exposure. (c) MRP1 expression in breast (black lines) and ovarian (red lines) tumor cells after doxorubicin exposure. (d) TOPO I expression in ovarian tumor cells after topotecan exposure. (e) TOPO IIα expression in ovarian tumor cells after topotecan exposure. (f) TS expression in breast (red lines) and colorectal (black lines) tumor cells after 5-FU exposure. (g) ERCC1 expression in breast (black lines) and ovarian (red lines) tumor cells after cisplatin exposure. (h) TS expression in esophageal tumor cells after ECF exposure. The numerical data for these graphs is summarized in Tables 5-8.
Table 1 List of primers for qRT-PCR. Sequence of primers (forward and reverse) used for qRT-PCR experiments. GenBank accession numbers for each gene are indicated in brackets. The primers were designed using an old version of the software Primer 3.0, available at the following website:
Name Sequence 5'-3' (forward and reverse primer) Product length Primer nM Mg2+ mM Reference
GAPDH* (NM_002046) GAA GGT GAA GGT CGG AGT C 226 200 4 -
GAA GAT GGT GAT GGG ATT TC 200
HPRT1* (NM_000194) TCA GGC AGT ATA ATC CAA AGA TGG T 84 400 4 [38]
AGT CTG GCT TAT ATC CAA CAC TTC G 400
PBGD* (NM_000190) CTG CAC GAT CCC GAG ACT CT 98 400 4 [39]
GCT GTA TGC ACG GCT ACT GG 400
SDHA* (NM_004168) TGG GAA CAA GAG GGC ATC TG 400 4 [12]
CCA CCA CTG CAT CAA ATT CAT G 400
TBP* (X54993) CAC GAA CCA CGG CAC TGA TT 89 400 4 [40]
TTT TCT TGC TGC CAG TCT GGA C 400
BCRP (AF098951) CAC AAC CAT TGC ATC TTG GC 74 400 4 [41]
GCT GCA AAG CCG TAA ATC CA 400
DPD (NM_000110) CCA AAG GCA GTA AAG CAG GAA 66 400 4 [42]
TCA CGA CTC CCC GTA TCG A 400
EGFR (NM_005228) TGG TCA AGT GCT GGA TGA TAG A 156 400 4 -
GGT AGA AGT TGG AGT CTG TAG GA 400
ERCC-1 (NM_001983) GGG AAT TTG GCG ACG TAA TTC 71 400 3 [43]
GCG GAG GCT GAG GAA CAG 400
GST-π (NM_000852) CGG AGA CCT CAC CCT GTA 169 400 5 -
CGC CTC ATA GTT GGT GTA GA 400
MDR1 (AF016535) TGG TTC AGG TGG CTC TGG AT 72 300 4 [44]
CTG TAG ACA AAC GAT GAG CTA TCA CA 300
MLH-1 (NM_000249) GGC ACA GCA TCA AAC CAA GT 147 400 4 -
GCA AGC ATG GCA AGG TCA A 400
MRP1 (L05628) CAA TGC TGT GAT GGC GAT G 70 400 4 [39]
GAT CCG ATT GTC TTT GCT CTT CA 400
MRP2 (NM_000392) TGC AGC CTC CAT AAC CAT GAG 80 400 3 [45]
GAT GCC TGC CAT TGG ACC TA 400
MT-II (NM_005953) GAT CCC AAC TGC TCC TGC 127 400 4 -
ACT TGG CAC AGC CCA CAG 400
MVP (NM_017458) CAG CTG GCC ATC GAG ATC A 68 400 4 [39]
TCC AGT CTC TGA GCC TCA TGC 400
TP (NM_001953) CCT TGG ATA AGC TGG AGT CT 107 400 4 -
CCT ACT CTG ACC CAC GAT AC 400
TS (NM_001071) CCA GAG ATC GGG AGA CAT GG 66 400 4 -
TAC GTG AGC AGG GCG TAG CT 400
TOPO I (J03250) CTC CAC AAC GAT TCC CAG AT 149 400 3 -
TTA TGT TCA CTG TTG CTA TGC TT 400
TOPO IIα (NM_001067) GTA ATT TTG ATG TCC CTC CAC GA 223 400 3 -
TCA AGG TCT GAC ACG ACA CTT 400
TOPO IIβ (NM_001068) GCA GCC GAA AGA CCT AAA TA 85 400 3 -
AAT CAT TAT TGT CAT CAT CAT CAT C 400
(N.B. Where primer sequences have been taken from references the PCR conditions have been further optimised compared to the original reference.)
* Housekeeping genes.
Table 2 List of antibodies used for immunohistochemical studies.
Antibody Pre-Treatment Dilution Incubation Cat No. Source Control Tissue
Glutathione S-Transferase pi GST-π (polyclonal) None 1:150 30 min RT PU249-UP BioGenex (Distributor: Menarini Diagnostics, Wokingham, UK) Breast Ca
P-glycoprotein (MDR-1) (Clone JSB-1) Pressure Cook 2 min pH 6.0 1:100 Overnight 4°C NCL-JSB1 Novo Castra Newcastle-upon-Tyne, UK Kidney
Multidrug Resistant-Related Protein (MRP) (Clone MRPm6, specific for MRP-1) Pressure Cook 2 min pH 7.0 1:30 30 min RT MAB4122 Chemicon International Chandlers Ford, UK Kidney
Thymidylate Synthase TS (Clone TS 106) Pressure Cook 2 min pH 6.0 1:20 30 min RT MS471P Neo Markers (Distributor: Lab Vision, Newmarket, UK) Colon Ca
When sections required microwaving a Matsui MIIOM microwave was used at 800 W power. Pressure cooking was performed with a Tefal Clipso Pressure Cooker using 70 P power.
Table 3 Relative expression of mRNA levels in esophageal samples obtained from patients before and after chemotherapy (median values). The last 2 columns on the right represent values for the 12 paired biopsies. The p values have been calculated using non parametric statistics, and in detail the Mann Whitney U test for unpaired samples, and the Wilcoxon matched pairs test for paired samples.
All esophageal samples Paired esophageal samples
Target gene Pre-chemo n = 13 Post-chemo n = 6 p Pre-chemo n = 6 Post-chemo n = 6 p
DPD 0.71 1.09 0.4117 0.68 1.09 0.3750
GST-π 4.39 0.88 0.1969 1.61 1.85 0.8438
MDR1 0.07 0.37 <0.0001 0.11 0.43 0.0313
MRP1 4.81 10.60 0.1624 4.58 10.77 0.4375
MT II 36.35 23.43 0.5846 12.49 24.12 0.8438
MVP 6.61 8.57 0.6544 6.39 9.11 0.5625
TP 4.21 2.58 0.2326 3.56 3.01 >0.999
TS 1.11 2.43 0.1851 1.27 2.27 0.4375
Table 4 Median expression (range in brackets) of protein levels in paired esophageal samples (n = 16) obtained from patients before and after chemotherapy. Slides were assessed using the H-score. A H-score of 100 or more was regarded as positive and below 100 was regarded as negative.
Antibody Pre-chemotherapy Post-chemotherapy
GST-π 35 (0–300) 200 (70–300)
P-gP 45 (0–300) 100 (0–300)
MRP 100 (0–200) 100 (10–200)
TS 120 (0–300) 100 (0–300)
Table 5 Relative expression of mRNA levels in tumor samples after ex vivo exposure to doxorubicin. The IC50 concentrations for the samples tested are shown for each drug (median and range). The p values have been calculated using non-parametric statistics, in detail the Wilcoxon matched pairs test. On statistical advice, we chose not to use a Bonferroni's correction, but it should be noted that some technically statistically significant results could have arisen by chance.
Breast samples Ovarian samples
Target gene n Control Doxorubicin IC50 = 0.828 μM (0.569–1.16) p n Control Doxorubicin IC50 = 1.34 μM (0.310–17.7) p
BCRP 14 0.039 0.044 0.8077 12 0.010 0.027 0.042
ERCC1 14 0.711 0.826 0.2412 11 0.440 0.506 0.4922
MDR1 14 0.069 0.056 0.0494 12 0.005 0.005 0.2036
MRP1 14 3.886 5.692 <0.0001 12 1.890 2.321 0.083
MVP 14 4.990 6.925 0.0023 12 2.251 2.687 0.042
TOPO I 14 5.464 5.128 0.0419 12 6.112 3.360 0.001
TOPO IIα 14 0.546 0.017 <0.0001 12 1.491 0.089 0.0015
TOPO IIβ 14 6.617 6.461 0.0785 12 5.907 5.252 0.0322
Table 6 Relative expression of mRNA levels in tumor samples after ex vivo exposure to topotecan and irinotecan. The IC50 concentrations for the samples tested are shown for each drug (median and range). The p values have been calculated using non-parametric statistics, in detail the Wilcoxon matched pairs test. On statistical advice, we chose not to use a Bonferroni's correction, but it should be noted that some technically statistically significant results could have arisen by chance.
Ovarian samples Colorectal samples
Target gene n Control Topotecan IC50 = 0.754 μM (0.164–4.82) p n Control Irinotecan IC50 = 55.4 μM (29.5–100) p
BCRP 9 0.0156 0.0125 >0.999 7 0.0354 0.0515 0.6875
COX-2 10 0.253 0.475 0.7695 7 105.17 38.49 0.0156
EGFR 10 0.470 0.163 0.0039 - - - -
ERCC1 10 0.252 0.399 0.002 7 1.403 2.469 0.0156
MDR1 9 0.0090 0.0034 0.5703 7 0.2253 0.2938 >0.999
MLH1 10 0.263 0.2268 0.0371 - - - -
MRP1 - - - - 7 5.132 3.775 0.0156
TOPO I 10 4.490 2.448 0.0371 7 1.468 1.119 0.1094
TOPO IIα 10 1.093 0.0670 0.002 6 0.2847 0.0733 0.2188
TOPO IIβ - - - - 7 0.7405 0.5058 0.0313
Table 7 Relative expression of mRNA levels in tumor samples after ex vivo exposure to 5 FU. The IC50 concentrations for the samples tested are shown for each drug (median and range). The p values have been calculated using non-parametric statistics, in detail the Wilcoxon matched pairs test. On statistical advice, we chose not to use a Bonferroni's correction, but it should be noted that some technically statistically significant results could have arisen by chance.
Breast samples IC50 = 200 μM (86.5–363) Colorectal samples IC50 = 112 μM (13.8–311)
Target gene n Control 5-FU p n Control 5-FU p
DPD 13 1.289 0.494 0.0012 10 0.503 0.448 0.4922
TP 13 8.876 10.928 0.6848 10 4.616 5.963 0.084
TS 13 1.136 3.819 0.0005 10 2.157 7.954 0.002
Table 8 Relative expression of mRNA levels in tumor samples after ex vivo exposure to cisplatin. The IC50 concentrations for the samples tested are shown for each drug (median and range). The p values have been calculated using non-parametric statistics, in detail the Wilcoxon matched pairs test. On statistical advice, we chose not to use a Bonferroni's correction, but it should be noted that some technically statistically significant results could have arisen by chance.
Breast samples IC50 = 18.4 μM (8.9–30.9) Ovarian samples IC50 = 17.8 μM (6.3–25.6)
Target gene n Control Cisplatin p n Control Cisplatin p
ATP7B 13 0.256 0.177 0.0327 7 0.268 0.111 0.4375
ERCC1 13 0.933 1.382 0.0266 7 0.542 0.758 0.0156
MGMT 13 1.203 1.275 0.8926 7 0.794 0.944 0.1094
MLH1 13 0.758 1.000 0.1465 7 0.512 0.683 0.3750
MSH2 13 1.289 1.047 0.0479 7 2.579 2.732 0.9375
MSH6 13 2.351 2.194 0.1909 7 2.997 2.194 0.1094
MTII 13 44.221 53.817 0.5417 7 12.996 11.445 0.8125
==== Refs
Andreotti PE Linder D Hartmann DM Cree IA Pazzagli M Bruckner HW TCA-100 tumour chemosensitivity assay: differences in sensitivity between cultured tumour cell lines and clinical studies J Biolumin Chemilumin 1994 9 373 378 7879653
Cree IA Kurbacher CM Untch M Sutherland LA Hunter EM Subedi AM James EA Dewar JA Preece PE Andreotti PE Correlation of the clinical response to chemotherapy in breast cancer with ex vivo chemosensitivity Anti-Cancer Drugs 1996 7 630 635 8913430
Matsumoto Y Takano H Fojo T Cellular adaptation to drug exposure: evolution of the drug-resistant phenotype Cancer Res 1997 57 5086 5092 9371507
Egger G Liang G Aparicio A Jones PA Epigenetics in human disease and prospects for epigenetic therapy Nature 2004 429 457 463 15164071
Abolhoda A Wilson AE Ross H Danenberg PV Burt M Scotto KW Rapid activation of MDR1 gene expression in human metastatic sarcoma after in vivo exposure to doxorubicin Clin Cancer Res 1999 5 3352 3356 10589744
Hu XF Slater A Kantharidis P Rischin D Juneja S Rossi R Lee G Parkin JD Zalcberg JR Altered multidrug resistance phenotype caused by anthracycline analogues and cytosine arabinoside in myeloid leukemia Blood 1999 93 4086 4095 10361105
Stein U Jurchott K Schlafke M Hohenberger P Expression of multidrug resistance genes MVP, MDR1, and MRP1 determined sequentially before, during, and after hyperthermic isolated limb perfusion of soft tissue sarcoma and melanoma patients J Clin Oncol 2002 20 3282 3292 12149303
Tada Y Wada M Migita T Nagayama J Hinoshita E Mochida Y Maehara Y Tsuneyoshi M Kuwano M Naito S Increased expression of multidrug resistance-associated proteins in bladder cancer during clinical course and drug resistance to doxorubicin Int J Cancer 2002 98 630 635 11920626
Andreotti PE Cree IA Kurbacher CM Hartmann DM Linder D Harel G Gleiberman I Caruso PA Ricks SH Untch M Chemosensitivity testing of human tumors using a microplate adenosine triphosphate luminescence assay: clinical correlation for cisplatin resistance of ovarian carcinoma Cancer Research 1995 55 5276 5282 7585588
Wilmanns C Fan D O'Brian CA Bucana CD Fidler IJ Orthotopic and ectopic organ environments differentially influence the sensitivity of murine colon carcinoma cells to doxorubicin and 5-fluorouracil Int J Cancer 1992 52 98 104 1500231
Hunter EM Sutherland LA Cree IA Subedi AM Hartmann D Linder D Andreotti PE The influence of storage on cytotoxic drug activity in an ATP-based chemosensitivity assay Anticancer Drugs 1994 5 171 176 8049500
Vandesompele J De Preter K Pattyn F Poppe B Van Roy N De Paepe A Speleman F Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes Genome Biology 2002 3 research0034.0031 research0034.0011 12184808
Applied Biosystems ABI PRISM 7700 Sequence Detection System user Bulletin #2 Relative Quantitation of Gene Expression (P/N 4303859B) 1997 10/2001 11 15
Cree IA LaRossa RA Luminescence-Based cell viability testing Bioluminescence Methods and Protocols 1998 Totowa, NJ: Humana Press Inc 169 177
Kurbacher CM Cree IA Brenne U Bruckner HW Kurbacher JA Mallmann P Andreotti PE Krebs D Heterogeneity of in vitro chemosensitivity in perioperative breast cancer cells to mitoxantrone versus doxorubicin evaluated by a microplate ATP bioluminescence assay Breast Cancer Res Treat 1996 41 161 70 8944334
Gottesman MM Fojo T Bates SE Multidrug resistance in cancer: role of ATP-dependent transporters Nat Rev Cancer 2002 2 48 58 11902585
Borst P Evers R Kool M Wijnholds J A family of drug transporters: the multidrug resistance-associated proteins J Natl Cancer Inst 2000 92 1295 1302 10944550
Longley DB Harkin DP Johnston PG 5-fluorouracil: mechanisms of action and clinical strategies Nat Rev Cancer 2003 3 330 338 12724731
Chu E Koeller DM Johnston PG Zinn S Allegra CJ Regulation of thymidylate synthase in human colon cancer cells treated with 5-fluorouracil and interferon-gamma Mol Pharmacol 1993 43 527 533 8474431
Wang JC DNA topoisomerases Annu Rev Biochem 1996 65 635 692 8811192
Foglesong PD Reckord C Swink S Doxorubicin inhibits human DNA topoisomerase I Cancer Chemother Pharmacol 1992 30 123 125 1318169
Dabholkar M Vionnet J Bostick-Bruton F Yu JJ Reed E Messenger RNA levels of XPAC and ERCC1 in ovarian cancer tissue correlate with response to platinum-based chemotherapy J Clin Invest 1994 94 703 708 8040325
Samimi G Fink D Varki NM Husain A Hoskins WJ Alberts DS Howell SB Analysis of MLH1 and MSH2 expression in ovarian cancer before and after platinum drug-based chemotherapy Clinical Cancer Research: an Official Journal of the American Association for Cancer Research 2000 6 1415 1421 10778972
D'Atri S Graziani G Lacal PM Nistico V Gilberti S Faraoni I Watson AJ Bonmassar E Margison GP Attenuation of O(6)-methylguanine-DNA methyltransferase activity and mRNA levels by cisplatin and temozolomide in jurkat cells J Pharmacol Exp Ther 2000 294 664 671 10900246
Beck WT Morgan SE Mo YY Bhat UG Tumor cell resistance to DNA topoisomerase II inhibitors: new developments Drug Resist Updat 1999 2 382 389 11498354
Johnston PG Drake JC Trepel J Allegra CJ Immunological quantitation of thymidylate synthase using the monoclonal antibody TS 106 in 5-fluorouracil-sensitive and -resistant human cancer cell lines Cancer Res 1992 52 4306 4312 1643628
Copur S Aiba K Drake JC Allegra CJ Chu E Thymidylate synthase gene amplification in human colon cancer cell lines resistant to 5-fluorouracil Biochem Pharmacol 1995 49 1419 1426 7763285
Welsh SJ Titley J Brunton L Valenti M Monaghan P Jackman AL Aherne GW Comparison of thymidylate synthase (TS) protein up-regulation after exposure to TS inhibitors in normal and tumor cell lines and tissues Clin Cancer Res 2000 6 2538 2546 10873110
Swain SM Lippman ME Egan EF Drake JC Steinberg SM Allegra CJ Fluorouracil and high-dose leucovorin in previously treated patients with metastatic breast cancer J Clin Oncol 1989 7 890 899 2661735
Peters GJ van der Wilt CL van Triest B Codacci-Pisanelli G Johnston PG van Groeningen CJ Pinedo HM Thymidylate synthase and drug resistance Eur J Cancer 1995 31A 1299 1305 7577040
Salonga D Danenberg KD Johnson M Metzger R Groshen S Tsao-Wei DD Lenz HJ Leichman CG Leichman L Diasio RB Danenberg PV Colorectal tumors responding to 5-fluorouracil have low gene expression levels of dihydropyrimidine dehydrogenase, thymidylate synthase, and thymidine phosphorylase Clin Cancer Res 2000 6 1322 1327 10778957
Ishikawa Y Kubota T Otani Y Watanabe M Teramoto T Kumai K Takechi T Okabe H Fukushima M Kitajima M Dihydropyrimidine dehydrogenase and messenger RNA levels in gastric cancer: possible predictor for sensitivity to 5-fluorouracil Jpn J Cancer Res 2000 91 105 112 10744051
Ichikawa W Uetake H Shirota Y Yamada H Nishi N Nihei Z Sugihara K Hirayama R Combination of dihydropyrimidine dehydrogenase and thymidylate synthase gene expressions in primary tumors as predictive parameters for the efficacy of fluoropyrimidine-based chemotherapy for metastatic colorectal cancer Clin Cancer Res 2003 9 786 791 12576451
Kornmann M Schwabe W Sander S Kron M Strater J Polat S Kettner E Weiser HF Baumann W Schramm H Hausler P Ott K Behnke D Staib L Beger HG Link KH Thymidylate synthase and dihydropyrimidine dehydrogenase mRNA expression levels: predictors for survival in colorectal cancer patients receiving adjuvant 5-fluorouracil Clin Cancer Res 2003 9 4116 4124 14519634
Buchholz TA Stivers DN Stec J Ayers M Clark E Bolt A Sahin AA Symmans WF Hess KR Kuerer HM Valero V Hortobagyi GN Pusztai L Global gene expression changes during neoadjuvant chemotherapy for human breast cancer Cancer J 2002 8 461 468 12500855
Clarke PA George ML Easdale S Cunningham D Swift RI Hill ME Tait DM Workman P Molecular pharmacology of cancer therapy in human colorectal cancer by gene expression profiling Cancer Res 2003 63 6855 6863 14583483
Kurata T Tamura K Kaneda H Nogami T Uejima H Asai Go G Nakagawa K Fukuoka M Effect of re-treatment with gefitinib ('Iressa', ZD1839) after acquisition of resistance Ann Oncol 2004 15 173 174 14679138
Moniotte S Vaerman JL Kockx MM Larrouy D Langin D Noirhomme P Balligand JL Real-time RT-PCR for the detection of beta-adrenoceptor messenger RNAs in small human endomyocardial biopsies J Mol Cell Cardiol 2001 33 2121 2133 11735259
van den Heuvel-Eibrink MM Wiemer EA Prins A Meijerink JP Vossebeld PJ van der Holt B Pieters R Sonneveld P Increased expression of the breast cancer resistance protein (BCRP) in relapsed or refractory acute myeloid leukemia (AML) Leukemia 2002 16 833 839 11986944
Bieche I Onody P Laurendeau I Olivi M Vidaud D Lidereau R Vidaud M Real-Time Reverse Transcription-PCR Assay for Future Management of ERBB2-based Clinical Applications Clin Chem 1999 45 1148 1156 10430778
Faneyte IF Kristel PM Maliepaard M Scheffer GL Scheper RJ Schellens JH van de Vijver MJ Expression of the breast cancer resistance protein in breast cancer Clin Cancer Res 2002 8 1068 1074 11948115
Blanquicett C Gillespie GY Nabors LB Miller CR Bharara S Buchsbaum DJ Diasio RB Johnson MR Induction of thymidine phosphorylase in both irradiated and shielded, contralateral human U87MG glioma xenografts: implications for a dual modality treatment using capecitabine and irradiation Mol Cancer Ther 2002 1 1139 1145 12481438
Lord RV Brabender J Gandara D Alberola V Camps C Domine M Cardenal F Sanchez JM Gumerlock PH Taron M Sanchez JJ Danenberg KD Danenberg PV Rosell R Low ERCC1 expression correlates with prolonged survival after cisplatin plus gemcitabine chemotherapy in non-small cell lung cancer Clin Cancer Res 2002 8 2286 2291 12114432
Yague E Armesilla AL Harrison G Elliott J Sardini A Higgins CF Raguz S P-glycoprotein (MDR1) expression in leukemic cells is regulated at two distinct steps, mRNA stabilization and translational initiation J Biol Chem 2003 278 10344 10352 12525496
Taipalensuu J Tornblom H Lindberg G Einarsson C Sjoqvist F Melhus H Garberg P Sjostrom B Lundgren B Artursson P Correlation of gene expression of ten drug efflux proteins of the ATP-binding cassette transporter family in normal human jejunum and in human intestinal epithelial Caco-2 cell monolayers J Pharmacol Exp Ther 2001 299 164 170 11561076
|
16026610
|
PMC1199589
|
CC BY
|
2021-01-04 16:03:06
|
no
|
BMC Cancer. 2005 Jul 18; 5:78
|
utf-8
|
BMC Cancer
| 2,005 |
10.1186/1471-2407-5-78
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-781602661010.1186/1471-2407-5-78Research ArticleCancer cell adaptation to chemotherapy Di Nicolantonio Federica [email protected] Stuart J [email protected] Louise A [email protected] Francis G [email protected] Pauline A [email protected] Sanjay [email protected] Augusta [email protected] Sharon [email protected] Palma Silvana [email protected] Penny [email protected] Shaw S [email protected] Simon [email protected] Bernie [email protected] Alan [email protected] Tim [email protected] Jeremy [email protected] Constantinos [email protected] Ian A [email protected] Translational Oncology Research Centre, Department of Histopathology, Queen Alexandra Hospital, Portsmouth PO6 3LY, UK2 Department of Surgery, Queen Alexandra Hospital, Portsmouth PO6 3LY, UK3 Department of Mathematics and Statistics, University of Portsmouth, Buckingham Building, Lion Terrace, Portsmouth PO1 3HE, UK4 Department of Radiotherapy and Oncology, Southend Hospital, Prittlewell Chase, Westcliff-on-Sea, Essex SS0 0RY, UK5 Department of Radiotherapy and Oncology, St Mary's Hospital, Milton Road, Portsmouth PO3 6AD, UK2005 18 7 2005 5 78 78 23 11 2004 18 7 2005 Copyright © 2005 Di Nicolantonio et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Tumor resistance to chemotherapy may be present at the beginning of treatment, develop during treatment, or become apparent on re-treatment of the patient. The mechanisms involved are usually inferred from experiments with cell lines, as studies in tumor-derived cells are difficult. Studies of human tumors show that cells adapt to chemotherapy, but it has been largely assumed that clonal selection leads to the resistance of recurrent tumors.
Methods
Cells derived from 47 tumors of breast, ovarian, esophageal, and colorectal origin and 16 paired esophageal biopsies were exposed to anticancer agents (cisplatin; 5-fluorouracil; epirubicin; doxorubicin; paclitaxel; irinotecan and topotecan) in short-term cell culture (6 days). Real-time quantitative PCR was used to measure up- or down-regulation of 16 different resistance/target genes, and when tissue was available, immunohistochemistry was used to assess the protein levels.
Results
In 8/16 paired esophageal biopsies, there was an increase in the expression of multi-drug resistance gene 1 (MDR1) following epirubicin + cisplatin + 5-fluorouracil (ECF) chemotherapy and this was accompanied by increased expression of the MDR-1 encoded protein, P-gp. Following exposure to doxorubicin in vitro, 13/14 breast carcinomas and 9/12 ovarian carcinomas showed >2-fold down-regulation of topoisomerase IIα (TOPOIIα). Exposure to topotecan in vitro, resulted in >4-fold down-regulation of TOPOIIα in 6/7 colorectal tumors and 8/10 ovarian tumors.
Conclusion
This study suggests that up-regulation of resistance genes or down-regulation in target genes may occur rapidly in human solid tumors, within days of the start of treatment, and that similar changes are present in pre- and post-chemotherapy biopsy material. The molecular processes used by each tumor appear to be linked to the drug used, but there is also heterogeneity between individual tumors, even those with the same histological type, in the pattern and magnitude of response to the same drugs. Adaptation to chemotherapy may explain why prediction of resistance mechanisms is difficult on the basis of tumor type alone or individual markers, and suggests that more complex predictive methods are required to improve the response rates to chemotherapy.
==== Body
Background
Tumor resistance to chemotherapy is a well-known clinical phenomenon that is now yielding its secrets to investigation at the molecular level in biopsy material. Studies in cell lines do not always correlate well with results from tumor tissue [1], which may consist largely of non-neoplastic cells that support and modify the biology of neoplastic cells. Thus it is important to validate the mechanisms important in vitro with the situation in the patient. Nevertheless, cell line studies and immunohistochemical studies of tumors suggest that resistance is a selective process: only those cells that survive a drug-induced insult will re-grow.
We have previously shown development of such resistance to combination chemotherapy in tumor-derived cells from matched biopsies collected from breast cancer patients before and after administration of doxorubicin-containing chemotherapy [2]. In this study we show similar results in patients with esophageal cancer from biopsies obtained prior to and several months after chemotherapy. Two cycles of the combination of epirubicin, cisplatin and 5-FU (ECF) are given to these patients prior to resection, allowing studies to be performed with paired samples before and after chemotherapy. We have used real-time quantitative RT-PCR (qRT-PCR) and immunohistochemistry (IHC) to assess targets known to be of importance to resistance to these agents.
The mechanisms involved in resistance to chemotherapy usually involve up-regulation of resistance mechanisms, or down-regulation of target genes. Examples of the former include drug efflux pump molecules such as multi-drug resistance gene 1/P-glycoprotein (MDR1/P-gp), while the latter include topoisomerases (TOPOs), targets of drugs such as etoposide and doxorubicin. Many papers attest to the importance of clonal selection in this process: it is for instance possible to expose cell lines to low concentrations of drugs and, over time, to produce highly resistant sub-clones [3]. However, there is another potential mechanism that does not require clonal selection: cells may be able to adapt by regulation of expression of resistance or target molecules individually if they survive the initial exposure to the drug. This could be a more rapid process and would require changes in molecular expression, possibly due to epigenetic change, rather than genetic mechanisms such as mutation [4]. As a result, resistance may therefore arise rapidly following treatment with chemotherapy.
Recent studies have shown that the expression of MDR1/P-gp is up-regulated within hours of anti-cancer drug treatment in vivo in patient samples [5-8], although this effect was not observed in all patients. We therefore wished to examine how quickly and in how many cases these resistance molecules were up-regulated in tumor-derived cells from several tumor types. We have used selective short-term cell culture (6 days) to examine the changes in expression that occur following exposure to chemotherapy compared to medium-only control cells from the same samples. Our short-term culture system employs a serum-free medium and polypropylene 96 'U' well microplates. This inhibits the proliferation and survival of normal cells and allows selective survival of a neoplastic cell population [9]. The short incubation period also limits the possibility of selection of clones or sub-populations in vitro. However, the presence of non-neoplastic cells for most of the incubation period allows the interaction between stromal cells and neoplastic cells, a factor that appears to be important to maintain the chemosensitivity profile [10].
Methods
Patients and tissue samples for in vitro studies
Tumor derived cells were obtained from 17 breast cancer patients (16 primaries; 1 pre-treated with mitoxantrone and paclitaxel), 13 ovarian cancer patients (all pre-treated with a cisplatin-based regimen), 10 colorectal cancer patients (all primaries) and 7 esophageal cancer patients (3 untreated; 4 treated with ECF). Cells were grown for 6 days in serum-free medium with or without drugs, before RNA extraction and further PCR analysis.
Patients and tissue samples for in vivo study
Thirty-four esophageal adenocarcinoma biopsies, thirty-two of which were paired samples were obtained from patients (17M:1F; median age 57, range 42–81) before and after administration of 2 cycles of ECF chemotherapy. After enzymatic digestion, tumor derived cells were centrifuged over Ficoll (Sigma Chemical Co, Poole, UK, Cat. No. 1077-1) to remove blood contaminating cells, washed in PBS, and stored in RNA later (Ambion, Huntingdon, UK) at -80°C until further molecular analysis was performed. Tissue sections from these samples were stained for GST-π, MRP1, P-gp and TS. All tumor samples were removed as part of patient treatment, with consent for tissue donation and local research ethics committee approval for use of the tissue surplus to diagnostic requirements for cellular and molecular assays. Chemosensitivity data were available for 9 patients with recurrent ovarian cancer, before treatment and on relapse, though in only one case was material available from both samples for quantitative RT-PCR.
Drugs
Cisplatin, 5-fluorouracil (5-FU), epirubicin, doxorubicin, irinotecan, paclitaxel (Taxol®) and topotecan (Hycamptin®) were obtained from the pharmacy at Queen Alexandra Hospital (Portsmouth, UK). Cisplatin, 5-FU, irinotecan and paclitaxel were stored at room temperature, while all other drugs were stored at -20°C, as previously reported [11]. Test drug concentrations (TDC) were 10.0 μM for cisplatin, 345 μM for 5-FU, 148 μM for irinotecan, 15.9 μM for paclitaxel, 2.5 μM for doxorubicin, 0.862 μM for epirubicin, and 1.64 μM for topotecan. Combinations were made up by adding two or three drugs concurrently at their 200% TDC at the beginning of the ATP-cell viability assay and diluted in a constant ratio: sequential studies were not performed.
Short-term cell culture
Briefly, tumor tissue or fluid was taken by a histopathologist or surgeon under sterile conditions, and transported to the laboratory in cell culture medium of Dulbecco's modified Eagle's medium (DMEM; Sigma Cat No. D5671) with antibiotics (100 U/ml penicillin and 100 μg/ml streptomycin, Sigma, Cat No. P0781) at 4°C. Cells were obtained from solid tumors by enzymatic dissociation, usually 0.75 mg/ml collagenase (Sigma Cat No. C-8051) overnight. Viable tumor-derived cells were purified by density centrifugation (Histopaque 1077-1, Sigma), washed, counted and resuspended to 100,000 cells/ml in case of effusions or 200,000 cells/ml for solid biopsies. In the meantime 96-well polypropylene microplates (Corning-Costar, High Wycombe, UK; Cat No. 3790) were prepared with each drug/combination at six doubling dilutions in triplicate from 200% TDC to 6.25% TDC, according to Andreotti et al. [9]. Approximately 10,000–20,000 cells/well were added to the plates to a final volume of 200 μl/well. The plates were then incubated at 37°C in 5% CO2 for 6 days, after which the degree of cell inhibition was assessed by measurement of the remaining ATP in comparison with negative control (no drug, MO) and positive control (maximum inhibitor, MI) rows of 12 wells each. Prior to cell lysis with an ATP-extracting reagent, an aliquot of 150 μl of cell suspension were removed from each well, centrifuged, washed with phosphate buffered saline (PBS) and stored at -80°C in a GTIC-containing solution (lysis buffer RA1, Macherey-Nagel, Düren, Germany; Cat. No.740961) until further molecular analysis was carried out. The RNA was subsequently extracted from aliquots that had been exposed to a drug concentration capable of inhibiting cell growth by 40–60%. ATP was extracted from the remaining 50 μl cell suspension and measured by light output in a microplate luminometer (Berthold Diagnostic Systems GmbH, Pforzheim, Germany) following addition of luciferin-luciferase.
RNA extraction
Cells obtained after enzymatic dissociation from endoscopic esophageal biopsies or short term cell culture were either resuspended in RNA later (Ambion, Huntingdon, UK; Cat No. 7020) or lysed with buffer RA1 and stored at -80°C until RNA extraction. Total RNA was extracted from at least 50,000 cells with a commercially available kit (NucleoSpin® RNA II mini, Macherey-Nagel; Cat No. 740955) according to the manufacturer's instructions. The protocol included a DNase digestion step to prevent carry-over of genomic DNA in further analysis.
qRT-PCR
A two-step protocol was employed. Firstly, total RNA was reverse-transcribed using the Promega reverse transcription system (Promega, Southampton, UK; Cat No. A3500) including 8 μl RNA, 0.5 μg random primers, 20 units of recombinant RNasin® ribonuclease inhibitor and 15 units of reverse transcriptase AMV (Promega, Cat No. M9004) to each 20 μl reaction. The resulting c-DNA was amplified by qPCR on a Biorad iCycler instrument (BioRad Laboratories, Hemel Hampstead, UK). The constituents of each PCR reaction (25 μl) were 1 μl of cDNA (or H2O), 200–500 nM of each primer (Table 1), 200 μM each dATP, dCTP, dGTP, 400 μM dUTP, 3.0–5.0 mM MgCl2, 0.125 units AMPErase® UNG, 0.625 units of AmpliTaq Gold DNA polymerase and 1 × SYBR Green PCR buffer (all reagents were from Applied Biosystems, Warrington, UK). Product amplification was performed up to 45 PCR cycles, after uracil removal (2 min at 50°C) and polymerase activation (10 min at 95°C). Each two-step PCR cycle comprised denaturing (15 s at 95°C), annealing, and extending (1 min at 60°C). At the end of each run a final melt curve cycle (cooling to 50°C and then increasing stepwise 1°C to 95°C) was performed to exclude the presence of primer-dimer artefacts. At least 3 housekeeping genes were used for each experiment chosen from the following: glyceraldehyde-3 phosphate dehydrogenase (GAPDH), hypoxanthine phosphoribosyltransferase 1 (HPRT1), human porphobilinogen deaminase (PBGD), succinate dehydrogenase complex-subunit A (SDHA) and TATA box binding protein (TBP). The internal reference genes were selected due to their relative low abundance in normal tissue [12]. The housekeeping genes were amplified parallel to the target genes in separate wells. When possible, primer sequences (Table 1) were chosen to span exon boundaries to the following targets: breast cancer resistance protein (BCRP), dihydropyrimidine dehydrogenase (DPD), epidermal growth factor receptor (EGFR), excision repair cross-complementing 1 (ERCC1) gene, glutathione-S-transferase isoform π (GST-π), multi-drug resistance gene 1 (MDR1), MLH1, multi-drug resistance related protein 1 (MRP1), multi-drug resistance related protein 2 (MRP2), metallothionein (MT), major vault protein (MVP), thymidine phosphorylase (TP), thymidylate synthase (TS), topoisomerase I (TOPO I), topoisomerase IIα and β (TOPO IIα and TOPO IIβ). Each amplicon was amplified in a separate reaction to prevent competition among multiple sets of primers. A positive control (pooled c-DNA from a variety of human tumors, including breast, ovarian, colorectal and esophageal carcinoma) and negative controls with no template and RT-negative as template were added in every experiment. All assays were run in triplicate. Validation experiments were run to show that the efficiencies of the target and reference genes amplifications were approximately equal, and in the range 95–105%. The PCR cycle number that generated the first fluorescence signal above a threshold (threshold cycle, Ct; 10 standard deviations above the mean fluorescence generated during the baseline cycles) was determined, and a comparative Ct method was then used to measure relative gene expression [13]. The following formula was used to calculate the relative amount of the transcript in the sample: 2-ΔΔCt, where ΔCt is the difference in Ct between the gene of interest and the mean of the at least two reference genes, and ΔΔCt = ΔCt of drug non-exposed cells – ΔCt of drug-exposed cells, for the ex-vivo experiments, or ΔΔCt = ΔCt of pre-chemotherapy sample – ΔCt of post-chemotherapy sample, for matched tumor biopsies.
Immunohistochemistry
The monoclonal antibodies (Abs) for GST-π, MRP1, Pg-p, and TP (Table 2) were detected using the Chemicon IHC Select™ – Immuno Peroxidase secondary detection system (Chemicon International, Chandlers Ford, Southampton, UK, Cat# Det-HP1000) and stained with 3,3' diaminobenzidine (DAB; HD Supplies, Aylesbury, UK, Cat. 4170) on a Dako Autostainer instrument (Dako, Ely, UK). Positive and negative controls were included with each run. Volumes of 50 μl avidin/ml goat serum and 50 μl biotin/ml primary Ab were used to block endogenous avidin binding. The slides were counterstained with Gills Haematoxylin, dehydrated and cleared using the Leica© XL slide staining machine and mounted in Styrolite® mounting medium (BDH, Poole, Dorset, UK; Cat No. 361704Y).
Data analysis
The luminometer readings obtained from the ATP-TCA were entered into an Excel 2000 spreadsheet (Microsoft) which calculated the percentage of cell inhibition for each drug concentration according to the previous published formula: 1 - [(Test - MI)/(MO - MI)]*100 [9]. For each drug-response curve, the 50% inhibitory concentration (IC50) and the 90% inhibitory concentration (IC90) were also calculated as previously described [14].
Assessment of slides was done using the H-score. Staining intensity (none, 0 points; weak, 1 point; moderate, 2 points; strong, 3 points) and percentage of positive tumor cells were multiplied to achieve a score between 0 and 300. A H-score of 100 or more was regarded as positive and results less than 100 were regarded as negative. The correlation coefficients were calculated by the method of the least squares, and the correlation between the IC90 and IC50 values and immunohistochemistry indices was assessed using univariate linear regression (Statsdirect, Sale, UK).
Non-parametric statistical methods were used. The calculated and descriptive data were entered into an Access 2000 database (Microsoft) and analysed using a Wilcoxon two-tailed paired rank sum test for paired data or the Mann-Whitney U test for unpaired data, as appropriate (Statsdirect). IC50 and IC90 values for each compound were correlated to the relative mRNA levels of target genes using Spearman's rank correlation coefficient and multivariate analysis. On statistical advice, we chose not to use a Bonferroni's correction, but it should be noted that some technically statistically significant results could have arisen by chance.
Results
Chemotherapy-induced changes in mRNA levels in biopsy material
Figures 1a–e illustrate changes to IC50 μM values in pre- and post chemotherapy ovarian tumor-derived cells tested with a variety of chemotherapeutic agents. There is a general increase in resistance to doxorubicin when anthracycline-based regimens (mitoxantrone + paclitaxel or liposomal doxorubicin) are administered to ovarian cancer patients (Fig. 1a). We observed little cross-resistance for mitoxantrone, topotecan, paclitaxel, and cisplatin following anthracycline-based chemotherapy (Fig. 1b–e), [15], although some patients do show alteration of sensitivity to other agents. Similar changes can be seen in breast cancer biopsies before and after administration of a chemotherapy regimen [2]. The increase in resistance to doxorubicin shown in Figure 1a was accompanied by changes in drug resistance gene expression. Figure 1f highlights such up-regulation of MDR1 and breast cancer resistance protein (BCRP) observed in ovarian-tumor derived cells from a single patient.
Membrane proteins such as MDR1/P-gp, multi-drug resistance protein 1 (MRP1) and major vault protein (MVP) have been demonstrated to confer resistance to epirubicin by pumping the cytotoxic drug out of cells [16]. Paired esophageal samples were obtained from the same six patients both before and after chemotherapy; in all 6 biopsy pairs (Table 3), there was an increase of the mRNA levels of MDR1 following ECF chemotherapy (Fig. 2a). In a larger series of 20 esophageal samples (including those that did not have a paired biopsy), divided in two groups according to the patients' exposure to treatment, there was a 5-fold increase in MDR1 expression in the post-chemotherapy group (median expression index 0.07 pre- and 0.37 post-chemotherapy, p < 0.0001, Mann-Whitney non-parametric unpaired U test). The paired samples from patients before and after ECF chemotherapy also showed a trend towards increased expression of MRP1 and MVP (Table 3), with three tumors showing concomitant up-regulation of both genes. Furthermore, the increased expression of MRP1 paralleled that of glutathione-S-transferase isoform π (GST-π) in 2 of these 3 samples, consistent with the mechanism of detoxification of MRP1 that involves transport of glutathione-conjugated molecules [17].
We then determined the relative mRNA levels of enzymes that have previously been correlated with 5-FU sensitivity: dihydropyrimidine dehydrogenase (DPD), thymidine phosphorylase (TP) and thymidylate synthase (TS) [18]. No significant difference was found for these 3 genes when we compared their expression pre- and post-chemotherapy in unpaired samples (Table 3). However, we found a modest increase of TS expression in 5/6 paired samples (Fig. 2b), 3 of which also showed a concomitant decrease in TP levels. One sample showed a paradoxical decrease in TS, indicating some heterogeneity. Although the sample size is small, the trend of increased TS levels in ECF exposed tumors is consistent with a number of previous reports that demonstrated an acute induction of TS expression in cell lines, animal models and human tumors following 5-FU treatment [18,19].
It should be noted that these qRT-PCR results were all obtained from samples that included normal cells present in the tumor as well as neoplastic cells, and could be affected by changes in normal cells as well as malignant cells.
Chemotherapy-induced changes in protein levels in biopsy material
IHC was performed on 16 paired esophageal biopsies: qRT-PCR data were obtained for four of these pairs. Overall, we were able to confirm an increased expression of the MDR-1 encoded protein, P-gp (Table 4), in 8/16 paired tumor samples obtained from patients who had been exposed to ECF chemotherapy. TS positivity (Table 4) was found in 10/16 samples obtained before treatment and 10/16 post-chemotherapy specimens; in this instance the modest induction of TS seen at the mRNA level did not reflect an increase in protein level (data not shown). We found GST-π (Table 4) positivity in 5/16 pre-chemotherapy biopsies. In the post-chemotherapy specimens we detected strong expression of GST-π in all but one of the samples. In one case that had been negative at diagnosis we measured a marginal increase of mRNA levels with qRT-PCR. A total of 13/16 pre- and 12/16 post-chemotherapy biopsies were found to be positive for MRP1 (Table 4). Overall, these data show little correlation between the mRNA and the degree of expression of protein levels determined by IHC. Confounding factors may be the presence of non-cancerous cells as mentioned above, but also the nature of IHC, which is at best a semi-quantitative technique.
Changes in short-term cell culture
Of the 47 solid tumor samples studied in short-term cultures, 7 were esophago-gastric, 17 were breast carcinomas, 13 were ovarian carcinomas and 10 were colorectal tumors. A total of 93 experiments were performed, as in most cases the same sample was treated with 2 or more drugs.
Doxorubicin effects in short-term cell culture
In the first instance, we decided to study the effects of anthracycline exposure for 6 days on tumour-derived cells. Our in vitro experiments showed a significant down-regulation of TOPO IIα; its median levels decreased from 0.546 (range 0.005–1.782) to 0.017 (range 0.0004–0.595) units in a group of 14 breast samples (p < 0.0001, Wilcoxon matched pairs test) and from 1.491 (range 0.080–6.884) to 0.089 (range 0.003–3.605) units in the ovarian cancer subgroup (p < 0.0015, Wilcoxon matched pairs test). The down-regulation of TOPO IIα levels was greater than 2-fold in 13/14 breast samples and in 9/12 ovarian samples (Fig. 3a). It has been suggested that cancer cells can concomitantly down-regulate TOPO IIα and up-regulate TOPO I in response to treatment with topoisomerase II inhibitors [20]. In our series we noted a general trend of diminished levels of TOPO I, particularly in the ovarian subgroup (Fig 3b). It is possible that doxorubicin exposure resulted in quiescence of cells which were not killed, and which would not therefore require higher levels of TOPO I, or that doxorubicin may also partly inhibit TOPO I, as suggested by a previous study [21].
We observed a significant up-regulation of MRP-1 (p < 0.0001, Wilcoxon matched pairs test; Fig. 3c) and of MVP (p < 0.0023, Wilcoxon; Table 5) in the breast cancer subgroup. The same effect was not found in the ovarian cancer group, in which the levels of MRP-1 and MVP increased over 0.5 fold in only 4/12 and 3/12 samples, respectively (Table 5). No significant changes were found in the expression of MDR-1 and BCRP, although it should be noted that there was substantial heterogeneity among individual tumors (Table 5). For example, we measured a greater than 2-fold increase of MDR-1 levels in 2/12 ovarian samples (none of whom had received a MDR-1 pumped chemotherapeutic agent before), and a significant up-regulation of BCRP in 3/12 ovarian samples.
Topoisomerase I inhibitors effects in short-term cell culture
Subsequently we looked at the effect of 2-camptothecin derivatives, irinotecan and topotecan, on colorectal and ovarian tumor cells, respectively (Table 6). As both compounds act by TOPO I inhibition, we firstly measured the levels of their target enzyme. We noted a trend towards down-regulation of TOPO I in treated cells: exposure to irinotecan decreased TOPO I levels >2-fold in 3/7 colorectal tumors, while topotecan caused down-regulation in 4/10 ovarian samples (Fig. 3d). The decrease of TOPO I was accompanied by a concomitant reduction of TOPO IIα expression, which was particularly pronounced (>4-fold) in 6/7 colorectal tumors and 8/10 ovarian tumors (Fig. 3e).
No significant changes were observed in the expression of the drug efflux molecules, MDR-1, BCRP and MRP-1 (Table 6), though, as in the case of doxorubicin, considerable heterogeneity was noted. We observed an increase of BCRP levels after irinotecan or topotecan exposure in 2/7 colorectal samples and 3/9 ovarian samples.
Amongst the genes implicated in DNA repair (Table 6), the modest down-regulation of MLH1 by topotecan exposure was not found statistically significant, although it was noted in 7/10 ovarian samples. Up-regulation of ERCC1 expression was found in all 10 ovarian cancer samples exposed to topotecan (p < 0.002, Wilcoxon), and in all 7 colorectal specimens treated with irinotecan, although in this group the increase was modest (p = 0.016, Wilcoxon).
Finally, we looked at EGFR as a marker of tumor growth and progression. After topotecan exposure we found a decrease (more than 2-fold) in the expression of this growth factor in 9/10 ovarian samples (Table 6), with the median levels decreasing from 0.470 to 0.163 units (p < 0.0039, Wilcoxon).
5-FU effects in short-term cell culture
Preliminary data obtained from 2 esophageal biopsies indicated an increase of TS levels following 5-FU treatment. Subsequent in vitro experiments performed on material derived from 10 colorectal tumor samples and 13 breast tumor samples confirmed the expected increase of TS levels (Table 7) in the samples exposed to 5-FU (p < 0.0001, Wilcoxon; Fig. 3f). The results indicated a general trend towards DPD down-regulation in the cells that had been exposed to 5-FU in vitro, though this effect was modest in most cases and we observed a greater than 2-fold decrease in only 7/13 breast samples and 2/10 colorectal samples.
Cisplatin effects in short-term cell culture
We examined the mRNA expression levels of genes previously highlighted as important in cisplatin resistance in tumor-derived cells from 13 breast and 7 ovarian tumors (Table 8). Up-regulation of the nucleoside excision repair (NER) gene ERCC1 has previously been shown to increase the removal of platinum affected DNA [22]. We observed a significant increase in ERCC1 levels in both tumor-derived breast (p = 0.0266, Wilcoxon matched pairs test) and ovarian cells (p = 0.0156, Wilcoxon matched pairs test) (Fig 3g). We also observed significant down regulation of a second DNA repair gene, MSH2 in tumor-derived breast cells (p = 0.0479, Wilcoxon matched pairs test), but not in tumor-derived ovarian cells. Samimi et al., [23] have previously demonstrated that following cisplatin therapy, there is selection for cells expressing lower hMLH1 and hMSH2. However, no significant changes were observed in the other mismatch repair genes, MLH1 and MSH6 in either tumor type (Table 8).
There was no significant change in MGMT in tumor-derived breast or ovarian cells. However, it has previously been shown that MGMT mRNA levels begin to recover after 24 hours in the absence of drug [24]. There was a paradoxical decrease in the copper export pump ATP7B following cisplatin exposure (Table 8), in tumor-derived breast cells (p = 0.0327, Wilcoxon matched pairs test), but not in ovarian cells. No significant changes were observed in the heavy-metal binding protein MTII, in either tumor type.
ECF effects in short-term cell culture
We were able to study tumor-derived cells from 7 esophageal cancer patients, 4 of which had already been given ECF in vivo. The results mirror those obtained from paired esophageal biopsies: we noticed increased expression of TS in the cells that had been exposed to ECF (Fig 3h). However, the expected up-regulation of MDR1 was only detected in 4/7 samples (data not shown), and may therefore be occurring in non-neoplastic cells that do not normally express MDR1, rather than in the neoplastic cells, which we found to express this molecule to the same degree pre- and post-treatment.
Correlation of in vitro cytotoxicity with molecular expression
Lastly, for 5-FU, doxorubicin and irinotecan, IC90 and IC50 data were obtained by measuring the ATP levels in a small aliquot of cell suspension at the end of the incubation period. This allowed a comparison to be made between drug sensitivity and the expression of putative resistance genes in the cells that had been exposed to chemotherapeutic agents. Using multivariate analysis, the IC90 of 5-FU was correlated with the median change of mRNA levels of both TS and DPD measured in 5-FU treated tumor-derived cells compared to control cells (expressed as 2-ΔΔCt) (R2 = 0.872704; p = 0.0006 for DPD; p < 0.0001 for TS). In addition, expression of ERCC1 mRNA correlated with the IC50 values determined for doxorubicin in 11 breast samples (R = 0.7204, p < 0.0124). No other correlations were noted between sensitivity to the drugs and gene expression levels.
Discussion
Our results suggest that rapid adaptation to chemotherapy may result in a resistant phenotype. This is mediated by down- or up-regulation of genes that are usually correlated to the mechanism of action of the individual chemotherapeutic agent or relevant resistance mechanisms. Our data suggest that short-term cell culture of tumor-derived cells with drugs could provide a suitable model for studying resistance mechanisms. The mechanisms observed appeared to be more specific to the drug used than to the tumor type: constitutive resistance probably reflects pre-chemotherapy expression of resistance mechanisms (e.g. drug efflux molecule expression in esophageal carcinoma). Acquired resistance develops rapidly and is likely to reflect changes in gene regulation rather than mutation-dependent selection of clones. Clonal selection may be important in some rapidly growing tumors and cell lines grown in serum-containing media, but is unlikely to be the major factor in solid tumors which have relatively low doubling rates. While mutation-mediated resistance can be much more profound than that observed here, our data suggest that the functional effects may still be sufficient to render the patient's tumor resistant to treatment within one cycle of chemotherapy.
Changes in resistance and target molecules
A large proportion of the published studies on resistance to chemotherapy have investigated the development of resistance using cell lines generated in the lab after prolonged and step-wise exposure to anti-cancer drugs. These in vitro models are not necessarily representative of the in vivo situation, when patients are usually administered one cycle of chemotherapy every 3–4 weeks. There are few studies in clinical samples. Our approach allows us to expose the tumor cells to single drugs under carefully controlled conditions, even if this would be an inappropriate drug for that particular patient. We are then able to look concomitantly at cytotoxicity, and molecular markers of resistance in the same experiment.
Anthracyclines have a mechanism of action that includes TOPO IIα inhibition via DNA intercalation. Resistance to anthracyclines is thought to be mediated by a number of different mechanisms, which include mutation or alteration of its target enzyme, TOPO IIα, and up-regulation of drug efflux proteins, such as BCRP, MRP1, MVP and MDR-1 [25]. Our data show that many of the cell line data are correct: in most solid tumors, it appears that anthracycline and topotecan exposure do lead to decreased topoisomerase II and I expression respectively, while inducing the expression of drug efflux pump molecules. The reason for the alteration in topoisomerase II expression following topotecan exposure is not clear, but topotecan does affect cell proliferation, and any reduction in proliferation would indirectly affect the expression of topoisomerase II alpha induced during S phase. However, our results also suggest that the heterogeneity of chemosensitivity between tumors is reflected by heterogeneity of molecular determinants of resistance/sensitivity.
The increased TS levels in ECF and 5-FU exposed cells is consistent with a number of previous reports. Gene amplification of TS with consequent increases in TS mRNA and protein has been observed in cell lines that are resistant to 5-FU and fluorodeoxyuridine (FUDR) [26,27]. Treatment with 5-FU has been shown to acutely induce TS expression in cell lines, animal models and human tumors [19,28-30]. In general there is strong evidence that the expression of DPD and TS in GI cancers is predictive of response to 5-FU. It should be noted that the concomitant measurement of both these markers markedly enhanced the ability to predict tumor response to 5-FU-based chemotherapy in a number of studies [31-34].
Gene profiling and drug response
There are few studies comparing gene expression before and after chemotherapy. A few studies have reported microarray data in biopsy material taken before and after (or even during) chemotherapy. These studies are of great interest, though normal cell effects cannot be excluded from the results. Buchholz et al. [35] employed cDNA microarray to measure gene expression changes during chemotherapy in 5 patients with breast cancer. Clarke et al. [36] studied gene expression changes in 18 rectal cancer patients undergoing therapy with Mitomycin C or 5-FU. This study reported a number of genes implicated in protein synthesis and RNA metabolism to be significantly decreased during drug treatment. These studies are not directly comparable: tumors of different types respond better to different drugs, and differences in their adaptation to these drugs are therefore expected on the basis of their innate sensitivity or resistance.
The potential for positive selection: molecular chess
It is common to show a cross-over effect with clinical trials of treatments with differing mechanisms of action, in which patients treated with one type of chemotherapy show sensitivity to the alternative regimen following failure of the one to which they were allocated. The recognition that selection of a molecular phenotype by exposure to one anti-cancer agent may leads to the expression of molecular targets for other drugs raises the possibility that it might be possible to enhance sensitivity to second-line or maintenance therapy by careful selection of patients for first-line therapy [37]. This approach would provide patients with a "backstop" for their first-line chemotherapy. One can envisage a series of interlocking treatments using drugs with specific molecular targets, monitored by molecular assays, which would allow the oncologist to employ a form of molecular chess to defeat the tumor. This approach might overcome the inherent heterogeneity, which is likely to underlie the variable results obtained from sequential chemotherapy to date. Assessment of this process in tumors could provide predictive assays allowing the oncologist to tailor therapy to the patient and avoid the development of resistance within the tumor.
Conclusion
In summary, this study suggests that up-regulation of resistance genes or down-regulation in target genes may occur rapidly in human solid tumors, within days of the start of treatment, and that similar changes are present in pre- and post-chemotherapy biopsy material. The molecular processes used by each tumor appear to be linked to the drug used, but there is heterogeneity between individual tumors, even those with the same histological type, in the pattern and magnitude of response to the same drugs. Adaptation to chemotherapy may explain why prediction of resistance mechanisms is difficult on the basis of tumor type alone or individual markers, and suggests that more complex predictive methods are required to improve the response rates to chemotherapy.
List of abbreviations
Gene names were abbreviated as follows: glyceraldehyde-3 phosphate dehydrogenase (GAPDH), hypoxanthine phosphoribosyltransferase 1 (HPRT1), human porphobilinogen deaminase (PBGD), succinate dehydrogenase complex-subunit A (SDHA) and TATA box binding protein (TBP), breast cancer resistance protein (BCRP), dihydropyrimidine dehydrogenase (DPD), excision repair cross-complementing 1 (ERCC1) gene, epidermal growth factor receptor (EGFR), glutathione-S-transferase isoform π (GST-π), multi-drug resistance gene 1 (MDR1), mutL homologue 1 (MLH1), multi-drug resistance related protein 1 (MRP1), multi-drug resistance related protein 2 (MRP2), metallothionein II (MT II), major vault protein (MVP), thymidine phosphorylase (TP), thymidylate synthase (TS), topoisomerase I (TOPO I), topoisomerase IIα and β (TOPO II).
Competing interests
IAC is a director of CanTech Ltd. The remaining authors declare that they have no competing interests.
Authors' contributions
IAC, FDN and SJM conceived and designed the study. FDN, SJM, LAK, PAW, SS, AF and SG participated in the short-term cell culture studies. FDN, SJM, LAK and FGG also carried out the qRT-PCR studies. PJ carried out the immunohisotochemical studies and IAC and SDP carried out histological analysis. IAC, FDN, SJM and BH participated in the statistical analysis. All authors participated in the data analysis, drafting of the manuscript and read and approved the final version.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This project was funded by Portsmouth Hospitals NHS Trust, CanTech Ltd, the BBSRC (Ref 31/ABY14513), the Royal Navy (SJM), the European Commission (grant number BMH4-CT98-9522; FDN), and was supported by a donation from Schering Plough Ltd (LAK). We are grateful to the NHS predictive oncology programme and all the patients, oncologists and surgeons who submitted material for chemosensitivity testing to make this study possible. We thank Christine Seddon for assistance with data entry.
Figures and Tables
Figure 1 Changes due to chemotherapy in biopsies taken before and after chemotherapy. Cytotoxicity of a) doxorubicin; b) mitoxantrone; c) topotecan; d) paclitaxel and e) cisplatin (expressed as IC50 μM) in paired samples obtained from ovarian cancer patients before and after they were treated with an anthracycline containing regimen (paclitaxel+mitoxantrone n = 7; liposomal doxorubicin n = 2). Each line represents an individual patient. (f) Gene expression changes pre- and post-treatment (paclitaxel plus mitoxantrone chemotherapy) in one patient were analysed by qRT-PCR following in vitro exposure to doxorubicin.
Figure 2 Changes in relative gene expression in 6 paired esophageal tumor biopsies. a) Changes in MDR1 relative gene expression and b) changes in TS relative gene expression. Each dot represents the relative mRNA level for an individual tumor, measured before and after ECF chemotherapy.
Figure 3 Changes in relative expression of putative chemoresistance genes in tumor-derived cells. Each dot represents the relative mRNA level for an individual sample, measured after in vitro drug exposure compared with untreated control cells. (a) TOPO IIα expression in breast (black lines) and ovarian (red lines) tumor cells after doxorubicin exposure. (b) TOPO I expression in breast (black lines) and ovarian (red lines) tumor cells after doxorubicin exposure. (c) MRP1 expression in breast (black lines) and ovarian (red lines) tumor cells after doxorubicin exposure. (d) TOPO I expression in ovarian tumor cells after topotecan exposure. (e) TOPO IIα expression in ovarian tumor cells after topotecan exposure. (f) TS expression in breast (red lines) and colorectal (black lines) tumor cells after 5-FU exposure. (g) ERCC1 expression in breast (black lines) and ovarian (red lines) tumor cells after cisplatin exposure. (h) TS expression in esophageal tumor cells after ECF exposure. The numerical data for these graphs is summarized in Tables 5-8.
Table 1 List of primers for qRT-PCR. Sequence of primers (forward and reverse) used for qRT-PCR experiments. GenBank accession numbers for each gene are indicated in brackets. The primers were designed using an old version of the software Primer 3.0, available at the following website:
Name Sequence 5'-3' (forward and reverse primer) Product length Primer nM Mg2+ mM Reference
GAPDH* (NM_002046) GAA GGT GAA GGT CGG AGT C 226 200 4 -
GAA GAT GGT GAT GGG ATT TC 200
HPRT1* (NM_000194) TCA GGC AGT ATA ATC CAA AGA TGG T 84 400 4 [38]
AGT CTG GCT TAT ATC CAA CAC TTC G 400
PBGD* (NM_000190) CTG CAC GAT CCC GAG ACT CT 98 400 4 [39]
GCT GTA TGC ACG GCT ACT GG 400
SDHA* (NM_004168) TGG GAA CAA GAG GGC ATC TG 400 4 [12]
CCA CCA CTG CAT CAA ATT CAT G 400
TBP* (X54993) CAC GAA CCA CGG CAC TGA TT 89 400 4 [40]
TTT TCT TGC TGC CAG TCT GGA C 400
BCRP (AF098951) CAC AAC CAT TGC ATC TTG GC 74 400 4 [41]
GCT GCA AAG CCG TAA ATC CA 400
DPD (NM_000110) CCA AAG GCA GTA AAG CAG GAA 66 400 4 [42]
TCA CGA CTC CCC GTA TCG A 400
EGFR (NM_005228) TGG TCA AGT GCT GGA TGA TAG A 156 400 4 -
GGT AGA AGT TGG AGT CTG TAG GA 400
ERCC-1 (NM_001983) GGG AAT TTG GCG ACG TAA TTC 71 400 3 [43]
GCG GAG GCT GAG GAA CAG 400
GST-π (NM_000852) CGG AGA CCT CAC CCT GTA 169 400 5 -
CGC CTC ATA GTT GGT GTA GA 400
MDR1 (AF016535) TGG TTC AGG TGG CTC TGG AT 72 300 4 [44]
CTG TAG ACA AAC GAT GAG CTA TCA CA 300
MLH-1 (NM_000249) GGC ACA GCA TCA AAC CAA GT 147 400 4 -
GCA AGC ATG GCA AGG TCA A 400
MRP1 (L05628) CAA TGC TGT GAT GGC GAT G 70 400 4 [39]
GAT CCG ATT GTC TTT GCT CTT CA 400
MRP2 (NM_000392) TGC AGC CTC CAT AAC CAT GAG 80 400 3 [45]
GAT GCC TGC CAT TGG ACC TA 400
MT-II (NM_005953) GAT CCC AAC TGC TCC TGC 127 400 4 -
ACT TGG CAC AGC CCA CAG 400
MVP (NM_017458) CAG CTG GCC ATC GAG ATC A 68 400 4 [39]
TCC AGT CTC TGA GCC TCA TGC 400
TP (NM_001953) CCT TGG ATA AGC TGG AGT CT 107 400 4 -
CCT ACT CTG ACC CAC GAT AC 400
TS (NM_001071) CCA GAG ATC GGG AGA CAT GG 66 400 4 -
TAC GTG AGC AGG GCG TAG CT 400
TOPO I (J03250) CTC CAC AAC GAT TCC CAG AT 149 400 3 -
TTA TGT TCA CTG TTG CTA TGC TT 400
TOPO IIα (NM_001067) GTA ATT TTG ATG TCC CTC CAC GA 223 400 3 -
TCA AGG TCT GAC ACG ACA CTT 400
TOPO IIβ (NM_001068) GCA GCC GAA AGA CCT AAA TA 85 400 3 -
AAT CAT TAT TGT CAT CAT CAT CAT C 400
(N.B. Where primer sequences have been taken from references the PCR conditions have been further optimised compared to the original reference.)
* Housekeeping genes.
Table 2 List of antibodies used for immunohistochemical studies.
Antibody Pre-Treatment Dilution Incubation Cat No. Source Control Tissue
Glutathione S-Transferase pi GST-π (polyclonal) None 1:150 30 min RT PU249-UP BioGenex (Distributor: Menarini Diagnostics, Wokingham, UK) Breast Ca
P-glycoprotein (MDR-1) (Clone JSB-1) Pressure Cook 2 min pH 6.0 1:100 Overnight 4°C NCL-JSB1 Novo Castra Newcastle-upon-Tyne, UK Kidney
Multidrug Resistant-Related Protein (MRP) (Clone MRPm6, specific for MRP-1) Pressure Cook 2 min pH 7.0 1:30 30 min RT MAB4122 Chemicon International Chandlers Ford, UK Kidney
Thymidylate Synthase TS (Clone TS 106) Pressure Cook 2 min pH 6.0 1:20 30 min RT MS471P Neo Markers (Distributor: Lab Vision, Newmarket, UK) Colon Ca
When sections required microwaving a Matsui MIIOM microwave was used at 800 W power. Pressure cooking was performed with a Tefal Clipso Pressure Cooker using 70 P power.
Table 3 Relative expression of mRNA levels in esophageal samples obtained from patients before and after chemotherapy (median values). The last 2 columns on the right represent values for the 12 paired biopsies. The p values have been calculated using non parametric statistics, and in detail the Mann Whitney U test for unpaired samples, and the Wilcoxon matched pairs test for paired samples.
All esophageal samples Paired esophageal samples
Target gene Pre-chemo n = 13 Post-chemo n = 6 p Pre-chemo n = 6 Post-chemo n = 6 p
DPD 0.71 1.09 0.4117 0.68 1.09 0.3750
GST-π 4.39 0.88 0.1969 1.61 1.85 0.8438
MDR1 0.07 0.37 <0.0001 0.11 0.43 0.0313
MRP1 4.81 10.60 0.1624 4.58 10.77 0.4375
MT II 36.35 23.43 0.5846 12.49 24.12 0.8438
MVP 6.61 8.57 0.6544 6.39 9.11 0.5625
TP 4.21 2.58 0.2326 3.56 3.01 >0.999
TS 1.11 2.43 0.1851 1.27 2.27 0.4375
Table 4 Median expression (range in brackets) of protein levels in paired esophageal samples (n = 16) obtained from patients before and after chemotherapy. Slides were assessed using the H-score. A H-score of 100 or more was regarded as positive and below 100 was regarded as negative.
Antibody Pre-chemotherapy Post-chemotherapy
GST-π 35 (0–300) 200 (70–300)
P-gP 45 (0–300) 100 (0–300)
MRP 100 (0–200) 100 (10–200)
TS 120 (0–300) 100 (0–300)
Table 5 Relative expression of mRNA levels in tumor samples after ex vivo exposure to doxorubicin. The IC50 concentrations for the samples tested are shown for each drug (median and range). The p values have been calculated using non-parametric statistics, in detail the Wilcoxon matched pairs test. On statistical advice, we chose not to use a Bonferroni's correction, but it should be noted that some technically statistically significant results could have arisen by chance.
Breast samples Ovarian samples
Target gene n Control Doxorubicin IC50 = 0.828 μM (0.569–1.16) p n Control Doxorubicin IC50 = 1.34 μM (0.310–17.7) p
BCRP 14 0.039 0.044 0.8077 12 0.010 0.027 0.042
ERCC1 14 0.711 0.826 0.2412 11 0.440 0.506 0.4922
MDR1 14 0.069 0.056 0.0494 12 0.005 0.005 0.2036
MRP1 14 3.886 5.692 <0.0001 12 1.890 2.321 0.083
MVP 14 4.990 6.925 0.0023 12 2.251 2.687 0.042
TOPO I 14 5.464 5.128 0.0419 12 6.112 3.360 0.001
TOPO IIα 14 0.546 0.017 <0.0001 12 1.491 0.089 0.0015
TOPO IIβ 14 6.617 6.461 0.0785 12 5.907 5.252 0.0322
Table 6 Relative expression of mRNA levels in tumor samples after ex vivo exposure to topotecan and irinotecan. The IC50 concentrations for the samples tested are shown for each drug (median and range). The p values have been calculated using non-parametric statistics, in detail the Wilcoxon matched pairs test. On statistical advice, we chose not to use a Bonferroni's correction, but it should be noted that some technically statistically significant results could have arisen by chance.
Ovarian samples Colorectal samples
Target gene n Control Topotecan IC50 = 0.754 μM (0.164–4.82) p n Control Irinotecan IC50 = 55.4 μM (29.5–100) p
BCRP 9 0.0156 0.0125 >0.999 7 0.0354 0.0515 0.6875
COX-2 10 0.253 0.475 0.7695 7 105.17 38.49 0.0156
EGFR 10 0.470 0.163 0.0039 - - - -
ERCC1 10 0.252 0.399 0.002 7 1.403 2.469 0.0156
MDR1 9 0.0090 0.0034 0.5703 7 0.2253 0.2938 >0.999
MLH1 10 0.263 0.2268 0.0371 - - - -
MRP1 - - - - 7 5.132 3.775 0.0156
TOPO I 10 4.490 2.448 0.0371 7 1.468 1.119 0.1094
TOPO IIα 10 1.093 0.0670 0.002 6 0.2847 0.0733 0.2188
TOPO IIβ - - - - 7 0.7405 0.5058 0.0313
Table 7 Relative expression of mRNA levels in tumor samples after ex vivo exposure to 5 FU. The IC50 concentrations for the samples tested are shown for each drug (median and range). The p values have been calculated using non-parametric statistics, in detail the Wilcoxon matched pairs test. On statistical advice, we chose not to use a Bonferroni's correction, but it should be noted that some technically statistically significant results could have arisen by chance.
Breast samples IC50 = 200 μM (86.5–363) Colorectal samples IC50 = 112 μM (13.8–311)
Target gene n Control 5-FU p n Control 5-FU p
DPD 13 1.289 0.494 0.0012 10 0.503 0.448 0.4922
TP 13 8.876 10.928 0.6848 10 4.616 5.963 0.084
TS 13 1.136 3.819 0.0005 10 2.157 7.954 0.002
Table 8 Relative expression of mRNA levels in tumor samples after ex vivo exposure to cisplatin. The IC50 concentrations for the samples tested are shown for each drug (median and range). The p values have been calculated using non-parametric statistics, in detail the Wilcoxon matched pairs test. On statistical advice, we chose not to use a Bonferroni's correction, but it should be noted that some technically statistically significant results could have arisen by chance.
Breast samples IC50 = 18.4 μM (8.9–30.9) Ovarian samples IC50 = 17.8 μM (6.3–25.6)
Target gene n Control Cisplatin p n Control Cisplatin p
ATP7B 13 0.256 0.177 0.0327 7 0.268 0.111 0.4375
ERCC1 13 0.933 1.382 0.0266 7 0.542 0.758 0.0156
MGMT 13 1.203 1.275 0.8926 7 0.794 0.944 0.1094
MLH1 13 0.758 1.000 0.1465 7 0.512 0.683 0.3750
MSH2 13 1.289 1.047 0.0479 7 2.579 2.732 0.9375
MSH6 13 2.351 2.194 0.1909 7 2.997 2.194 0.1094
MTII 13 44.221 53.817 0.5417 7 12.996 11.445 0.8125
==== Refs
Andreotti PE Linder D Hartmann DM Cree IA Pazzagli M Bruckner HW TCA-100 tumour chemosensitivity assay: differences in sensitivity between cultured tumour cell lines and clinical studies J Biolumin Chemilumin 1994 9 373 378 7879653
Cree IA Kurbacher CM Untch M Sutherland LA Hunter EM Subedi AM James EA Dewar JA Preece PE Andreotti PE Correlation of the clinical response to chemotherapy in breast cancer with ex vivo chemosensitivity Anti-Cancer Drugs 1996 7 630 635 8913430
Matsumoto Y Takano H Fojo T Cellular adaptation to drug exposure: evolution of the drug-resistant phenotype Cancer Res 1997 57 5086 5092 9371507
Egger G Liang G Aparicio A Jones PA Epigenetics in human disease and prospects for epigenetic therapy Nature 2004 429 457 463 15164071
Abolhoda A Wilson AE Ross H Danenberg PV Burt M Scotto KW Rapid activation of MDR1 gene expression in human metastatic sarcoma after in vivo exposure to doxorubicin Clin Cancer Res 1999 5 3352 3356 10589744
Hu XF Slater A Kantharidis P Rischin D Juneja S Rossi R Lee G Parkin JD Zalcberg JR Altered multidrug resistance phenotype caused by anthracycline analogues and cytosine arabinoside in myeloid leukemia Blood 1999 93 4086 4095 10361105
Stein U Jurchott K Schlafke M Hohenberger P Expression of multidrug resistance genes MVP, MDR1, and MRP1 determined sequentially before, during, and after hyperthermic isolated limb perfusion of soft tissue sarcoma and melanoma patients J Clin Oncol 2002 20 3282 3292 12149303
Tada Y Wada M Migita T Nagayama J Hinoshita E Mochida Y Maehara Y Tsuneyoshi M Kuwano M Naito S Increased expression of multidrug resistance-associated proteins in bladder cancer during clinical course and drug resistance to doxorubicin Int J Cancer 2002 98 630 635 11920626
Andreotti PE Cree IA Kurbacher CM Hartmann DM Linder D Harel G Gleiberman I Caruso PA Ricks SH Untch M Chemosensitivity testing of human tumors using a microplate adenosine triphosphate luminescence assay: clinical correlation for cisplatin resistance of ovarian carcinoma Cancer Research 1995 55 5276 5282 7585588
Wilmanns C Fan D O'Brian CA Bucana CD Fidler IJ Orthotopic and ectopic organ environments differentially influence the sensitivity of murine colon carcinoma cells to doxorubicin and 5-fluorouracil Int J Cancer 1992 52 98 104 1500231
Hunter EM Sutherland LA Cree IA Subedi AM Hartmann D Linder D Andreotti PE The influence of storage on cytotoxic drug activity in an ATP-based chemosensitivity assay Anticancer Drugs 1994 5 171 176 8049500
Vandesompele J De Preter K Pattyn F Poppe B Van Roy N De Paepe A Speleman F Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes Genome Biology 2002 3 research0034.0031 research0034.0011 12184808
Applied Biosystems ABI PRISM 7700 Sequence Detection System user Bulletin #2 Relative Quantitation of Gene Expression (P/N 4303859B) 1997 10/2001 11 15
Cree IA LaRossa RA Luminescence-Based cell viability testing Bioluminescence Methods and Protocols 1998 Totowa, NJ: Humana Press Inc 169 177
Kurbacher CM Cree IA Brenne U Bruckner HW Kurbacher JA Mallmann P Andreotti PE Krebs D Heterogeneity of in vitro chemosensitivity in perioperative breast cancer cells to mitoxantrone versus doxorubicin evaluated by a microplate ATP bioluminescence assay Breast Cancer Res Treat 1996 41 161 70 8944334
Gottesman MM Fojo T Bates SE Multidrug resistance in cancer: role of ATP-dependent transporters Nat Rev Cancer 2002 2 48 58 11902585
Borst P Evers R Kool M Wijnholds J A family of drug transporters: the multidrug resistance-associated proteins J Natl Cancer Inst 2000 92 1295 1302 10944550
Longley DB Harkin DP Johnston PG 5-fluorouracil: mechanisms of action and clinical strategies Nat Rev Cancer 2003 3 330 338 12724731
Chu E Koeller DM Johnston PG Zinn S Allegra CJ Regulation of thymidylate synthase in human colon cancer cells treated with 5-fluorouracil and interferon-gamma Mol Pharmacol 1993 43 527 533 8474431
Wang JC DNA topoisomerases Annu Rev Biochem 1996 65 635 692 8811192
Foglesong PD Reckord C Swink S Doxorubicin inhibits human DNA topoisomerase I Cancer Chemother Pharmacol 1992 30 123 125 1318169
Dabholkar M Vionnet J Bostick-Bruton F Yu JJ Reed E Messenger RNA levels of XPAC and ERCC1 in ovarian cancer tissue correlate with response to platinum-based chemotherapy J Clin Invest 1994 94 703 708 8040325
Samimi G Fink D Varki NM Husain A Hoskins WJ Alberts DS Howell SB Analysis of MLH1 and MSH2 expression in ovarian cancer before and after platinum drug-based chemotherapy Clinical Cancer Research: an Official Journal of the American Association for Cancer Research 2000 6 1415 1421 10778972
D'Atri S Graziani G Lacal PM Nistico V Gilberti S Faraoni I Watson AJ Bonmassar E Margison GP Attenuation of O(6)-methylguanine-DNA methyltransferase activity and mRNA levels by cisplatin and temozolomide in jurkat cells J Pharmacol Exp Ther 2000 294 664 671 10900246
Beck WT Morgan SE Mo YY Bhat UG Tumor cell resistance to DNA topoisomerase II inhibitors: new developments Drug Resist Updat 1999 2 382 389 11498354
Johnston PG Drake JC Trepel J Allegra CJ Immunological quantitation of thymidylate synthase using the monoclonal antibody TS 106 in 5-fluorouracil-sensitive and -resistant human cancer cell lines Cancer Res 1992 52 4306 4312 1643628
Copur S Aiba K Drake JC Allegra CJ Chu E Thymidylate synthase gene amplification in human colon cancer cell lines resistant to 5-fluorouracil Biochem Pharmacol 1995 49 1419 1426 7763285
Welsh SJ Titley J Brunton L Valenti M Monaghan P Jackman AL Aherne GW Comparison of thymidylate synthase (TS) protein up-regulation after exposure to TS inhibitors in normal and tumor cell lines and tissues Clin Cancer Res 2000 6 2538 2546 10873110
Swain SM Lippman ME Egan EF Drake JC Steinberg SM Allegra CJ Fluorouracil and high-dose leucovorin in previously treated patients with metastatic breast cancer J Clin Oncol 1989 7 890 899 2661735
Peters GJ van der Wilt CL van Triest B Codacci-Pisanelli G Johnston PG van Groeningen CJ Pinedo HM Thymidylate synthase and drug resistance Eur J Cancer 1995 31A 1299 1305 7577040
Salonga D Danenberg KD Johnson M Metzger R Groshen S Tsao-Wei DD Lenz HJ Leichman CG Leichman L Diasio RB Danenberg PV Colorectal tumors responding to 5-fluorouracil have low gene expression levels of dihydropyrimidine dehydrogenase, thymidylate synthase, and thymidine phosphorylase Clin Cancer Res 2000 6 1322 1327 10778957
Ishikawa Y Kubota T Otani Y Watanabe M Teramoto T Kumai K Takechi T Okabe H Fukushima M Kitajima M Dihydropyrimidine dehydrogenase and messenger RNA levels in gastric cancer: possible predictor for sensitivity to 5-fluorouracil Jpn J Cancer Res 2000 91 105 112 10744051
Ichikawa W Uetake H Shirota Y Yamada H Nishi N Nihei Z Sugihara K Hirayama R Combination of dihydropyrimidine dehydrogenase and thymidylate synthase gene expressions in primary tumors as predictive parameters for the efficacy of fluoropyrimidine-based chemotherapy for metastatic colorectal cancer Clin Cancer Res 2003 9 786 791 12576451
Kornmann M Schwabe W Sander S Kron M Strater J Polat S Kettner E Weiser HF Baumann W Schramm H Hausler P Ott K Behnke D Staib L Beger HG Link KH Thymidylate synthase and dihydropyrimidine dehydrogenase mRNA expression levels: predictors for survival in colorectal cancer patients receiving adjuvant 5-fluorouracil Clin Cancer Res 2003 9 4116 4124 14519634
Buchholz TA Stivers DN Stec J Ayers M Clark E Bolt A Sahin AA Symmans WF Hess KR Kuerer HM Valero V Hortobagyi GN Pusztai L Global gene expression changes during neoadjuvant chemotherapy for human breast cancer Cancer J 2002 8 461 468 12500855
Clarke PA George ML Easdale S Cunningham D Swift RI Hill ME Tait DM Workman P Molecular pharmacology of cancer therapy in human colorectal cancer by gene expression profiling Cancer Res 2003 63 6855 6863 14583483
Kurata T Tamura K Kaneda H Nogami T Uejima H Asai Go G Nakagawa K Fukuoka M Effect of re-treatment with gefitinib ('Iressa', ZD1839) after acquisition of resistance Ann Oncol 2004 15 173 174 14679138
Moniotte S Vaerman JL Kockx MM Larrouy D Langin D Noirhomme P Balligand JL Real-time RT-PCR for the detection of beta-adrenoceptor messenger RNAs in small human endomyocardial biopsies J Mol Cell Cardiol 2001 33 2121 2133 11735259
van den Heuvel-Eibrink MM Wiemer EA Prins A Meijerink JP Vossebeld PJ van der Holt B Pieters R Sonneveld P Increased expression of the breast cancer resistance protein (BCRP) in relapsed or refractory acute myeloid leukemia (AML) Leukemia 2002 16 833 839 11986944
Bieche I Onody P Laurendeau I Olivi M Vidaud D Lidereau R Vidaud M Real-Time Reverse Transcription-PCR Assay for Future Management of ERBB2-based Clinical Applications Clin Chem 1999 45 1148 1156 10430778
Faneyte IF Kristel PM Maliepaard M Scheffer GL Scheper RJ Schellens JH van de Vijver MJ Expression of the breast cancer resistance protein in breast cancer Clin Cancer Res 2002 8 1068 1074 11948115
Blanquicett C Gillespie GY Nabors LB Miller CR Bharara S Buchsbaum DJ Diasio RB Johnson MR Induction of thymidine phosphorylase in both irradiated and shielded, contralateral human U87MG glioma xenografts: implications for a dual modality treatment using capecitabine and irradiation Mol Cancer Ther 2002 1 1139 1145 12481438
Lord RV Brabender J Gandara D Alberola V Camps C Domine M Cardenal F Sanchez JM Gumerlock PH Taron M Sanchez JJ Danenberg KD Danenberg PV Rosell R Low ERCC1 expression correlates with prolonged survival after cisplatin plus gemcitabine chemotherapy in non-small cell lung cancer Clin Cancer Res 2002 8 2286 2291 12114432
Yague E Armesilla AL Harrison G Elliott J Sardini A Higgins CF Raguz S P-glycoprotein (MDR1) expression in leukemic cells is regulated at two distinct steps, mRNA stabilization and translational initiation J Biol Chem 2003 278 10344 10352 12525496
Taipalensuu J Tornblom H Lindberg G Einarsson C Sjoqvist F Melhus H Garberg P Sjostrom B Lundgren B Artursson P Correlation of gene expression of ten drug efflux proteins of the ATP-binding cassette transporter family in normal human jejunum and in human intestinal epithelial Caco-2 cell monolayers J Pharmacol Exp Ther 2001 299 164 170 11561076
|
16091145
|
PMC1199590
|
CC BY
|
2021-01-04 16:30:07
|
no
|
BMC Cardiovasc Disord. 2005 Aug 9; 5:23
|
latin-1
|
BMC Cardiovasc Disord
| 2,005 |
10.1186/1471-2261-5-23
|
oa_comm
|
==== Front
BMC Dev BiolBMC Developmental Biology1471-213XBioMed Central London 1471-213X-5-141603364810.1186/1471-213X-5-14Research ArticleA genome-wide in situ hybridization map of RNA-binding proteins reveals anatomically restricted expression in the developing mouse brain McKee Adrienne E [email protected] Emmanuel [email protected] Charlene [email protected] Shervin [email protected] Charles D [email protected] Pamela A [email protected] Department of Systems Biology, Harvard Medical School, Boston, MA 02115 USA2 Department of Cancer Biology, The Dana-Farber Cancer Institute, Boston, MA 02115 USA3 URBC-FUNDP, 61 rue de Bruxelles, 5000 Namur, Belgium4 Department of Microbiology and Molecular Genetics, Harvard Medical School, Boston, MA 02115 USA2005 20 7 2005 5 14 14 6 5 2005 20 7 2005 Copyright © 2005 McKee et al; licensee BioMed Central Ltd.2005McKee et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
In eukaryotic cells, RNA-binding proteins (RBPs) contribute to gene expression by regulating the form, abundance, and stability of both coding and non-coding RNA. In the vertebrate brain, RBPs account for many distinctive features of RNA processing such as activity-dependent transcript localization and localized protein synthesis. Several RBPs with activities that are important for the proper function of adult brain have been identified, but how many RBPs exist and where these genes are expressed in the developing brain is uncharacterized.
Results
Here we describe a comprehensive catalogue of the unique RBPs encoded in the mouse genome and provide an online database of RBP expression in developing brain. We identified 380 putative RBPs in the mouse genome. Using in situ hybridization, we visualized the expression of 323 of these RBP genes in the brains of developing mice at embryonic day 13.5, when critical fate choice decisions are made and at P0, when major structural components of the adult brain are apparent. We demonstrate i) that 16 of the 323 RBPs examined show neural-specific expression at the stages we examined, and ii) that a far larger subset (221) shows regionally restricted expression in the brain. Of the regionally restricted RBPs, we describe one group that is preferentially expressed in the E13.5 ventricular areas and a second group that shows spatially restricted expression in post-mitotic regions of the embryonic brain. Additionally, we find a subset of RBPs that share the same complex pattern of expression, in proliferating regions of the embryonic and postnatal NS and peripheral tissues.
Conclusion
Our data show that, in contrast to their proposed ubiquitous involvement in gene regulation, most RBPs are not uniformly expressed. Here we demonstrate the region-specific expression of RBPs in proliferating vs. post-mitotic brain regions as well as cell-type-specific RBP expression. We identify uncharacterized RBPs that exhibit neural-specific expression as well as novel RBPs that show expression in non-neural tissues. The data presented here and in an online database provide a visual filter for the functional analysis of individual RBPs.
==== Body
Background
The ordered production and differentiation of cell types that occurs during nervous system (NS) development relies upon tightly regulated gene expression. In neural cells, spatial and temporal gene regulation occurs through both transcriptional and post-transcriptional mechanisms. While the transcriptional networks that direct neural cell fate and govern cell shape, position, and connectivity have been well studied [1-3], the post-transcriptional influences on neural development and gene expression are less well understood.
At the core of post-transcriptional gene regulation are RNA-binding proteins (RBPs). Proteins containing canonical RNA-binding domains (RBDs) are involved in numerous steps of nuclear and cytoplasmic RNA processing [4]. Through mRNA capping, splicing, editing, polyadenylation and nonsense-mediated decay, RBPs modulate the diversity of transcribed genes [4-6]. RBPs also affect the processing of non-coding RNAs [7]. Specific RBPs additionally enable asymmetric RNA distribution and translational regulation [8-10], two phenomena that are critical for affecting localized protein synthesis [11,12].
The importance of post-transcriptional processing in NS gene regulation is underscored by functional examples of specific RBPs [13,14]. For instance, the neuronal-specific factor Nova-1 regulates splicing of pre-mRNAs that encode components of inhibitory synapses [15]. Mice lacking Nova-1 die postnatally due to aberrant regulation of apoptotic neuronal death [16]. As a second example, RBPs encoded by the quaking and Musashi loci promote glial cell fate [17] and CNS stem cell self-renewal [18] by stabilizing transcripts involved in cell differentiation. Thirdly, the fragile X mental retardation protein, members of the ELAV/Hu protein family, and the Staufen proteins are involved in targeting and translational regulation of dendritic transcripts [19-21]. Additionally, the finding that long-term memory requires de novo protein synthesis highlights the significance of post-transcriptional processes in neural function [22,23].
Despite our knowledge of several key RBPs, much of the understanding of RBPs in the brain comes from studies of adult animals or neural cell lines. Thus, how the functional class of RBPs contributes to the positioning, growth, and diversification of cells in the developing brain is not well understood. One step towards increasing our understanding RBPs is to resolve where they are expressed. Here, we utilize the approach of in situ hybridization mapping [24-26] to investigate the expression of 323 RBPs within the developing mouse brain. Two stages of development were characterized, embryonic day 13.5 (E13.5), when critical cellular fate choice decisions are made and postnatal day 0 (P0), when the major structural components of the brain are apparent. We find that, in contrast to their proposed ubiquitous involvement in gene regulation, most RBPs are not uniformly expressed. The majority of RBPs profiled demonstrates spatially restricted expression in the brain or in other peripheral tissues examined. The data presented here and in an online database afford a visual filter for the functional analysis of individual RBPs in the developing mammalian NS.
Results
Mouse RBPs were identified according to gene sequence
The RNA recognition motif (RRM), the hnRNP K-homology (KH) domain, and the double-stranded RNA-binding domain (dsRM) are evolutionarily conserved, well-characterized domains known to bind either single or double-stranded RNA [27-29]. Sequence similarity searches and structural analyses of these domains have led to the ability to predict other RBPs based on primary coding sequence [29]. To identify unique genomic loci that encode putative RBPs in the mouse genome, we analyzed existing public [30,31] and private [32] databases for sequences containing one or more RBD. Candidates were classified as RBPs only if their predicted protein sequence contained a Protein Families Database (Pfam)-defined RBD [31].
We identified 290 genes harboring one or more RRM, KH, or dsRM sequences. We also identified 32 genes encoding other domains shown to interact with RNA, including the zinc knuckle, G-patch, PIWI, DEAD box RNA helicase, and TUDOR domains. Finally, as the absence of a canonical RBD does not preclude interaction with RNA, we sought 58 additional genes known or predicted to be associated with RNA processing. In total, this collection contains 380 putative RBPs. Additional file 1 lists the number of genes, per RBD, identified and analyzed by in situ hybridization. A list of all genes and primer sequences is given in Additional file 2.
RBP expression in the developing mouse brain was analyzed by in situ hybridization
To localize RBP expression, we preformed in situ hybridization on whole head tissue sections of E13.5 embryos and P0 mice. We designed gene-specific primers to produce 400–700 bp probes for 340 candidate RBPs. These primer sets were used to perform PCR on cDNA prepared from embryonic or P0 mouse brains. A small number of probes were obtained from mouse intestine, liver, kidney, or testes cDNA. 323 genes (95%) showed positive PCR products (data not shown). Following subcloning, anti-sense digoxygenin-labeled riboprobes were prepared and hybridized against coronal head and transverse upper-body sections (to include the brain and spinal cord, respectively). Digital images of the entire in situ hybridization set have been deposited in the Mahoney RNA-Binding Protein Expression Database [33].
RBPs exhibit restricted expression in the developing mouse brain
Several neural-specific RBPs have been identified, yet how many others demonstrate this degree of specificity is unknown. Of the genes examined we found 16 RBPs (listed in Additional file 2) that exhibit NS-restricted expression in the tissues analyzed. Among this list are known examples of neuronal-specific RBPs including Nova-1 [34], the ELAV/Hu proteins B, C, and D [35], and Ataxin 2 binding protein 1 (A2bp1) [36] but additionally include putative RBPs for which expression has not been reported. With the exception of one gene that was only detected at E13.5, all (15/16) of these RBPs appear brain or NS-specific at both developmental stages in the tissues analyzed. Overall, these RBP encoding genes are not limited in expression to one brain region but are found in multiple brain or NS structures.
RBPs show spatially restricted expression in anatomically distinct brain regions
We find that greater than half of the RBPs profiled exhibit spatially restricted expression. Of the 323 genes examined, 221 demonstrate localized, enriched expression in one or more discrete brain regions in addition to detectable expression in non-NS tissues. We divided the E13.5 and P0 CNS into five and eight general areas for annotation, respectively: the E13.5 precortical area, the striatum (and other basal ganglia), the periventrical areas, hindbrain, and spinal cord, as well as the P0 cortex, striatum, hippocampus, thalamus, hypothalamus, midbrain, hindbrain, and spinal cord. The presence or absence of expression for each RBP was analyzed visually at each location and is annotated in Additional file 3. Very few of the 221 RBPs with spatially restricted expression patterns were expressed in only one brain region, however most (73%) showed restricted expression at both developmental stages (Additional file 3).
We observe multiple RBPs that demonstrate region-specific expression in the E13.5 ventricular areas. Shown in Figure 1 are representative RBP genes that are transcribed in mitotically-active cells in the neuroepithelia of the developing telencephalon. Among the RBPs expressed in this region occupied by neural progenitor cells, we find examples of mRNA export factors in addition to putative splicing factors and transcriptional regulators (Fig. 1). In all instances, expression in the embryonic lateral ventricular zone is accompanied by expression in the periventricular areas of the 3rd and 4th E13.5 ventricles and often by heightened expression in the P0 subventricular zone [33]. Notably, we observed this pattern of expression for the dsRM-containing Musashi proteins [33]. Our results are consistent with the documented expression of Msi1 and Msi2 [37,38].
Figure 1 RBP expression in proliferative zones of the E13.5 mouse forebrain. In situ hybridization patterns for four RBPs on sections through the forebrain of E13.5 mice. Labels indicate Locuslink gene names. All images show the same magnification.
Multiple RBPs show restricted expression in post-mitotic regions of embryonic brain. Presented in Figure 2 are examples of four putative RBPs that demonstrate region-specific expression in areas containing post-mitotic neurons. Transcripts of the genes encoding the RRM protein Brunol6 and the predicted zinc-knuckle protein 1500031H04Rik appear pan-neuronal at both developmental stages (Fig. 2A, 2B and [33]). Expression of the RRM-containing RIKEN gene 4930565A21 is most pronounced in the ventral telencephalon, while D11Bwg0517e is found in the precortical layer, the thalamic area and hindbrain (Fig. 2C, 2D and [33]). Among the genes that occupy post-mitotic regions of the developing brain we additionally observe members of the ELAV/Hu family as well as other RBPs that have well-documented neuronal expression [34,35].
Figure 2 RBP expression in post-mitotic areas of the E13.5 mouse forebrain. In situ hybridization patterns for four RBPs on sections through the forebrain of E13.5 mice. Labels indicate Locuslink gene names. bg, basal ganglia; hy, hypothalamus; nc, neocortex. All images show the same magnification.
RBPs demonstrate cell-type specific expression in the P0 mouse retina
As our in situ hybridization analyses were performed on sections through whole head, we were able to visualize RBP expression in the developing retina. The vertebrate retina provides a distinctive system for studying CNS development as its seven major neural cell types are readily distinguished from one another by their morphology and laminar position [39]. Shown in Figure 3 are examples of the diversity of RBP expression in the P0 retina. The RRM-containing A2bp1 is expressed in the retinal ganglion cell layer (GCL), which contains primarily retinal ganglion cells and a small number of displaced amacrine cells (Fig 3A, 3B). The KH-domain encoding gene poly(rC) binding protein 3 (Pcbp3) shows dramatically enriched expression in the inner nuclear layer (INL) (Fig. 3C and 3D), possibly indicating localization to the bipolar neuron cell bodies that occupy the scleral portion of the INL. Notably, both A2bp1 and Pcbp3 show restricted expression in post-mitotic regions of the E13.5 and P0 brain [24,36]. Transcripts of the RRM-encoding scaffold attachment factor B (Safb) and of the three-RRM containing SPOC gene Rbm15 are expressed in the outer neuroblastic layer of the retina (Fig. 3E–H). Safb, but not Rbm15, is additionally expressed in the GCL, possibly in the Müller glia. Both Safb and Rbm15 show enriched expression in neuroepithelia of the ventricular zone (Fig. 1 and [33]).
Figure 3 Diversity of RBP expression in major cellular subtypes of the P0 retina. In situ hybridization for four representative RBPs that exhibit laminar-specific expression in the P0 mouse retina. Labels indicate Locuslink gene names. A, B) A2bp1, C, D) Pcbp3, E, F) Safb, G, H) Rbm15. Panels A, C, E, and G show the same magnification. Panels B, D, F, and H show the same magnification. gcl, granule cell layer; inl, inner nuclear layer, onbl; outer neuroblastic layer.
A systems-based view of RBP expression
Gene regulation by RBPs is believed to occur through coordinated, combinatorial interactions with RNA. During the course of this study we identified multiple RBPs that are coordinately expressed in the brain and other tissues. We find 48 genes (listed in Additional file 4) that show elevated expression in proliferating areas of the embryonic and postnatal brain as well as in postnatal nasal epithelia, teeth, and thymus. Presented in Figure 4 are expression data for snRNP E and Son, two representative examples of this "synexpression group" of genes that share a similar, complex pattern of expression. Further examples are shown in Additional file 5. This same expression distribution has been observed for the polypyrimidine tract-binding protein, PTBP1, and our data are consistent with previous findings [40]. Notably, the protein products of many of the genes listed are understood to interact either physically or genetically.
Figure 4 Representative examples of RBP synexpression in E13.5 and P0 mouse tissues. snRNP E and Son are transcribed in the perventricular areas of the E13.5 brain (A, E), in the P0 subventricular area of the lateral ventricle (B, F), in the external granule layer of the P0 cerebellum (C, G), as well as in postnatal developing teeth (D, H).
RBPs show restricted expression in non-NS tissues
As our analyses were performed on whole head and upper thoracic tissues, our data provide detailed information about RBP expression in developing cranial facial tissues. We identified putative RBPs that display tissue-restricted expression in non-NS structures (listed in Additional file 3). Figure 5 presents in situ hybridization results for two RRM-encoding transcripts that show highly restricted expression in different epithelial tissues. The Riken gene 2210008M09 is transcribed in epithelia covering the facial skeleton (Fig. 5A, 5B), while the gene BC013481 is expressed in the choroid plexus (Fig. 5C) and in the lining of the intestine and placenta (Fig. 5D, 5E).
Figure 5 In situ hybridization profiling uncovers the non-neural, restricted expression of novel RBPs. Data from ISH performed on (A, C) coronal E13.5 and on (B, D, E) E15 sagittal sections are presented for RRM-encoding RBPS. A, B) The Riken gene 2210008M09 is transcribed in epithelia covering the facial skeleton. C-E) BC013481 is detected in the choroid plexus, in the intestinal lining, and in the lining of the placenta. Panels C-E show the same magnification.
Discussion
Neural cells utilize multiple forms of post-transcriptional gene regulation. While RBPs are believed to be potent modulators of post-transcriptional processes, little is known about how this functional class is expressed in the developing brain. As a first step towards increasing our knowledge of RBPs we chose to investigate the spatial and temporal expression of genes that encode motifs known to interact with RNA. We find a small set of RBPs that show neural-specific expression in the tissues analyzed. An even greater number of RBP genes however demonstrate spatially restricted expression in distinct regions of the developing brain.
Within the CNS, most of the RBPs examined show non-uniform, heightened expression in anatomically discrete structures. Tissue differences in the expression levels of individual genes could indicate distinctive protein requirements among cell types, beyond that of tissue-specific RBPs [41]. There is precedent for differential requirements of individual RBPs, as tissue-specific RNA splicing is achieved partly through combinatorial, stoichiometric differences among splicing factors within various cells [42]. It is from this local enrichment within different cell types or tissues that we can begin to hypothesize as to the functional significance of individual genes as well as to the importance of groups of similarly expressed RBPs.
Our study has identified RBPs that display spatially restricted expression in distinct regions of the developing mouse brain. One set of RBPs (Fig. 1) is found in the E13.5 ventricular areas. A second set demonstrates spatially restricted expression in post-mitotic regions of E13.5 brain (Fig. 2). Based on their pattern of expression, these RBPs may have roles in neural proliferation, cell fate choice and cell migration, or in neuronal function, respectively. We also identified novel RBPs that are expressed in tissues of mesodermal and endodermal origin (Fig. 5). The highly restricted expression of these genes may indicate an explicit role for these RBPs in their respective epithelia. Additionally, the cell-type specificity RBPs found in the P0 retina (Fig. 3) illustrates the diversity of RBP expression. The specialized expression of these RBPs may be indicative of a dedicated function in the specified tissues.
By visual inspection of in situ hybridization data, we find a subset of RBPs that are coordinately expressed in multiple tissue types. These genes display heightened expression in the periventricular areas of the E13.5 brain and spinal cord as well as marked expression in the external granule layer of the P0 cerebellum, the lateral subventricular zones, and in teeth, nasal epithelia, and thymus (Fig. 4, Additional file 5, [33]). While not excluded from post-mitotic tissues, these RBPs are predominately expressed in structures that are undergoing cell division.
Notably, the term 'synexpression group' has been used to describe collections of genes that function in a common process and share a similar complex spatial expression pattern in multiple tissues [43]. Among the synexpression group identified here we find examples of RBPs that are known to interact either physically or genetically (Additional file 4). For example, PTBP1 binds the splicing factors PSF [44] and hnRNP L [45] while SF2/ASF and hnRNP A1 select for 5' exon or exclusion or inclusion, respectively [46]. Our data provide visual support to a growing body of evidence that functionally-related transcripts are post-transcriptionally co-regulated [47].
Although the significance of certain splicing and mRNA export factor enrichment in proliferating regions is not known, data from multiple studies point to a role for RBPs in cell proliferation. During hippocampal development expression levels of RBPs were found to be high and then to dramatically decrease, as neurons transition from a proliferating to a post-mitotic state [48]. A number of RBPs were also identified as highly expressed in a molecular characterization of gastric epithelial progenitor cells [49,50]. Furthermore, protein levels of hnRNPs and snRNPs were found to be down-regulated upon stimulated growth inhibition of myeloid cells [51]. Therefore, it is likely that a role for RBPs during cell proliferation and cell fate determination exists in multiple tissue types.
Conclusion
In summary, the data presented here provide new insight into how a distinct functional gene class is expressed in the developing NS. We find that RBPs demonstrate region-specific as well as cell-type specific expression. In addition, we find that specific, proliferating regions of the embryonic and postnatal NS and peripheral tissues are similar in the expression of certain RBPs. These data serve as a starting point for functional investigations into the roles of RBPs in neural development and physiology.
Methods
In silico RBP identification
Putative RBP gene sequences were identified by homology-based whole genome screening using public and private databases: Celera Panther Families, Protein Families Database (Pfam), and Genbank [30-32]. Classification as an RBP was based on the presence of one or more RRM, KH, or dsRMs, as defined by Pfam databases [31]. Databases were also mined for zinc-knuckle, G-patch, PIWI, DEAD-box helicase and Tudor domain-containing sequences and for known factors involved in mRNA splicing, editing, transport, and stability. Genes with multiple RNA-binding domains were assigned to a single subfamily. Unique gene identity was verified by LocusID numbers. As of March 1, 2004, a total of 357 unique genes were identified from these sources. An additional 26 RRM, KH, and dsRM proteins have been identified as of March 7, 2005.
PCR primer design
PCR primer pairs were designed for each identified RNA-binding protein locus. PCR primer sequences were designed with approximately 60% GC content, spanning 400–700 base pairs of primarily the gene's coding sequence. Additional primer pairs were designed for targets that did not initially yield PCR products.
Cloning
Total RNA was obtained from E13.5, P0, or adult C57/BL6 mouse brains (Charles River Laboratories) by Trizol extraction (Invitrogen). Reverse transcription was performed using Superscript II reverse transcriptase and oligo-dT (Invitrogen). PCR was performed with cDNA templates using 40 cycles, 60–65°C annealing temperature, and Platinum Taq (Invitrogen) as polymerase. For a few genes, PCR was performed with cDNA templates prepared from adult brain, kidney, gut, liver, or testis tissues. Positive PCR products were cloned into TA cloning vectors (Invitrogen) and verified by restriction digest or DNA sequencing.
Probe synthesis
Gene fragments from verified plasmids were amplified by PCR using plasmid specific primers. Digoxigenin-labeled RNA probes were made, using PCR products as template and T7 or SP6 RNA polymerases (Roche). cRNA probes were ethanol precipitated and quantified by spectrophotometry.
Tissue preparation
E13.5 embryos were directly fixed overnight in 4% paraformaldehyde (0.1M PBS). P0 mice were transcardially perfused with 4% paraformaldehyde (0.1M PBS) and postfixed overnight at 4°C. After fixation, embryos and P0 mice were transferred to 20% sucrose overnight. The head, neck, and trunk were embedded separately in OCT (Tissue-Tek) on dry ice and stored at -80°C. Serial cryostat sections (14 μm) were cut and mounted on Superfrost Plus slides (Fisher). Ten and twenty adjacent sets of sections were prepared from E13.5 embryos and P0 mice, respectively, and were stored at -20°C until use.
Section in situ hybridization
In situ hybridization was performed according to Gray et al. [25]. Following pretreatment (Proteinase K), slides were pre-hybridized for 1h at 65°C in hybridization solution (50% formamide (Ambion), 5X SSC, 0.3 mg/ml yeast tRNA (Sigma), 100 μg/ml heparin (Sigma), 1X Denhardt's (Sigma), 0.1% tween, 5 mM EDTA). P0 and E13.5 brain sections were hybridized overnight with labeled RNA probe(0.8–1.2 μg/ml) at 65°C, washed in 2X SSC at 67°C, incubated with RNase A (1 μg/ml, 2X SSC) at 37°C, washed in 0.2X SSC at 65°C, blocked in PBS with 10% lamb sera, and incubated in alkaline phosphatase labeled anti-DIG antibody (Roche) (1:2000, 10% sera) overnight. Sections were washed and color was visualized using NBT and BCIP in alkaline phosphatase buffer (100 mM Tris pH 9.5, 50 mM MgCl2, 100 mM NaCl, 0.1% tween-20) containing 75 μg/ml NBT (BioRad), 600 μg/ml BCIP (Roche). Staining was stopped after visual inspection. Sections were washed, fixed in 4% paraformaldehyde, and coverslipped in glycerol [25].
Image acquisition and RBP expression database
Images were acquired and analyzed as described [25]. Images were either scanned using a Nikon Coolscan 8000 slide scanner (4000 DPI) or digitally acquired using a Leica digital camera. Image levels have been modified in Photoshop (Adobe) for clarity. Full resolution scanned images were compressed using JPEG compression, quality 10, and have been deposited in the Mahoney RNA-Binding Protein Expression Database [33].
Authors' contributions
AEM prepared tissue samples, performed data analysis and drafted the manuscript. EM performed data analysis and both EM and SR generated reagents, tissue samples, digitized the raw data, and helped build the website. CS contributed to the design of the study and prepared tissue samples. CDS and PAS conceived of the study, participated in its design and coordination and helped prepare the manuscript. All authors read and approved of the manuscript.
Supplementary Material
Additional File 1
RNA-binding proteins identified in silico and profiled by in situ hybridization. List of annotated RNA-binding domains and the number of family members that were identified in silico and analyzed by in situ hybridization.
Click here for file
Additional File 2
List of 380 genes identified as putative RBPs in the mouse genome and analyzed in this study. Columns indicate LocusID, gene name, type of RBD, primer sequences used to isolate the target cDNA, the size of the cDNA fragment, the presence call by PCR from E13.5 and P0 brain cDNA, cloning status ('c' indicates cloned, 'u' indicates uncloned, 'small' indicates that the target gene had less than 400 bp of unique sequence, 'na' indicates that cloning was not attempted), the RNA polymerase used to generate the anti-sense riboprobe, the tissue from which the cDNA was isolated (if not from E13.5 or P0 mouse brain), and whether the gene was analyzed by in situ hybridization ('x' indicates yes).
Click here for file
Additional File 3
Complete list of gene expression patterns for all in situ hybridizations performed. Of the 323 RBPs examined, 221 showed restricted expression patterns in the brain. The remaining genes either show restricted expression in non-neural tissues, ubiquitous expression that is difficult to distinguish from background, or no expression. Caution is needed in interpreting the results. First, non-expression could be due to the sensitivity limit of non-radioactive in situ hybridization. Second, the background level of individual probes may differ. Third, some probes with high background hybridization may mask the real expression of the transcript. Fourth, we cannot rule out the possibility that some probes may show variable levels of background hybridization in different brain areas, resulting in a false positive signal. Columns A-D describe the LocusID, gene name, type of RBD, and number (internal Mahoney reference number). Columns E and, L (E13.5, P0 "Informativity"): "1" for restricted expression in the nervous system and "0" for either ubiquitous expression that is difficult to distinguish from background or no expression. As noted in Gray et al [25], some of the genes in the "0" category show uneven signals in different brain regions and are also annotated in the subsequent columns. Columns F and M (E13.5, P0 "Specificity"): "1" for restricted expression in neural tissues only, "2" for restricted expression in neural tissue with distinguishable expression in non-neural tissue, "3" for ubiquitous or no expression, and "4" for expression in non-neural tissues only. Columns G-K and N-U (E13.5, P0 "Expression"): "2" for expression, "1" for ubiquitous expression or background, "0" for no expression.
Click here for file
Additional File 4
RNA-binding proteins belonging to a synexpression group. Complete list of RBPs that demonstrate a similar complex pattern of expression. Columns A-D describe the LocusID, gene name, type of RBD, and number (internal Mahoney reference number).
Click here for file
Additional File 5
Examples of RBP synexpression in E13.5 and P0 mouse tissues. Additional examples of RBPs that share a similar pattern of expression. Shown are in situ hybridization results of expression in the periventricular areas of the E13.5 brain (A, E, I, M, Q), in the subventricular area of the P0 lateral ventricle (B, F, J, N, R), in the external granule layer of the P0 cerebellum (C, G, K, O, S), as well as in postnatal developing teeth (D, H, L P, T). A-D) Refbp1, E-H) hnRNP A1, I-L) PTBP1, M-P) Sfpq, Q-R) Hnrpl. Panels A, B, E, F, I, J, M, N, Q, R show the same magnification. Panels C, D, G, H, K, L, O, P, S, T show the same magnification.
Click here for file
Acknowledgements
We are grateful to Drs. Qiufu Ma and John Alberta for critical review of this manuscript and for assistance in this work. We thank Eric Tsung, Zhaohui Cai, and Matthew McCormack for designing the website. This work has been supported by the Bernard A. and Wendy J. Goldhirsh Foundation for Brain Tumor Research and by the Charles A. Dana Foundation. AEM is supported by an institutional training grant from the National Cancer Institute (T32CA09361). EM is funded as a FNRS Researcher through the Belgian National Research Fund and by the D. Collen Research Foundation VZW and BAEF. CS received support from the American Cancer Society (PF-02-128-01-MBC).
==== Refs
Ross SE Greenberg ME Stiles CD Basic helix-loop-helix factors in cortical development Neuron 2003 39 13 25 12848929 10.1016/S0896-6273(03)00365-9
Wilson SW Houart C Early steps in the development of the forebrain Dev Cell 2004 6 167 181 14960272 10.1016/S1534-5807(04)00027-9
Bally-Cuif L Hammerschmidt M Induction and patterning of neuronal development, and its connection to cell cycle control Curr Opin Neurobiol 2003 13 16 25 12593978 10.1016/S0959-4388(03)00015-1
Dreyfuss G Kim VN Kataoka N Messenger-RNA-binding proteins and the messages they carry Nat Rev Mol Cell Biol 2002 3 195 205 11994740 10.1038/nrm760
Lasko P Gene regulation at the RNA layer: RNA binding proteins in intercellular signaling networks Sci STKE 2003 2003 RE6 12709531
Orphanides G Reinberg D A unified theory of gene expression Cell 2002 108 439 451 11909516 10.1016/S0092-8674(02)00655-4
Cullen BR Transcription and processing of human microRNA precursors Mol Cell 2004 16 861 865 15610730 10.1016/j.molcel.2004.12.002
Huang YS Carson JH Barbarese E Richter JD Facilitation of dendritic mRNA transport by CPEB Genes Dev 2003 17 638 653 12629046 10.1101/gad.1053003
Antar LN Bassell GJ Sunrise at the synapse: the FMRP mRNP shaping the synaptic interface Neuron 2003 37 555 558 12597853 10.1016/S0896-6273(03)00090-4
Tang SJ Meulemans D Vazquez L Colaco N Schuman E A role for a rat homolog of staufen in the transport of RNA to neuronal dendrites Neuron 2001 32 463 475 11709157 10.1016/S0896-6273(01)00493-7
Martin KC Local protein synthesis during axon guidance and synaptic plasticity Curr Opin Neurobiol 2004 14 305 310 15194110 10.1016/j.conb.2004.05.009
Huang YS Richter JD Regulation of local mRNA translation Curr Opin Cell Biol 2004 16 308 313 15145356 10.1016/j.ceb.2004.03.002
Agnes F Perron M RNA-binding proteins and neural development: a matter of targets and complexes Neuroreport 2004 15 2567 2570 15570153 10.1097/00001756-200412030-00001
Perrone-Bizzozero N Bolognani F Role of HuD and other RNA-binding proteins in neural development and plasticity J Neurosci Res 2002 68 121 126 11948657 10.1002/jnr.10175
Ule J Jensen KB Ruggiu M Mele A Ule A Darnell RB CLIP identifies Nova-regulated RNA networks in the brain Science 2003 302 1212 1215 14615540 10.1126/science.1090095
Jensen KB Dredge BK Stefani G Zhong R Buckanovich RJ Okano HJ Yang YY Darnell RB Nova-1 regulates neuron-specific alternative splicing and is essential for neuronal viability Neuron 2000 25 359 371 10719891 10.1016/S0896-6273(00)80900-9
Larocque D Galarneau A Liu HN Scott M Almazan G Richard S Protection of p27(Kip1) mRNA by quaking RNA binding proteins promotes oligodendrocyte differentiation Nat Neurosci 2005 8 27 33 15568022 10.1038/nn1359
Sakakibara S Nakamura Y Yoshida T Shibata S Koike M Takano H Ueda S Uchiyama Y Noda T Okano H RNA-binding protein Musashi family: roles for CNS stem cells and a subpopulation of ependymal cells revealed by targeted disruption and antisense ablation Proc Natl Acad Sci U S A 2002 99 15194 15199 12407178 10.1073/pnas.232087499
Pascale A Gusev PA Amadio M Dottorini T Govoni S Alkon DL Quattrone A Increase of the RNA-binding protein HuD and posttranscriptional up-regulation of the GAP-43 gene during spatial memory Proc Natl Acad Sci U S A 2004 101 1217 1222 14745023 10.1073/pnas.0307674100
Jin P Alisch RS Warren ST RNA and microRNAs in fragile X mental retardation Nat Cell Biol 2004 6 1048 1053 15516998 10.1038/ncb1104-1048
Dubnau J Chiang AS Grady L Barditch J Gossweiler S McNeil J Smith P Buldoc F Scott R Certa U Broger C Tully T The staufen/pumilio pathway is involved in Drosophila long-term memory Curr Biol 2003 13 286 296 12593794 10.1016/S0960-9822(03)00064-2
Miller S Yasuda M Coats JK Jones Y Martone ME Mayford M Disruption of dendritic translation of CaMKIIalpha impairs stabilization of synaptic plasticity and memory consolidation Neuron 2002 36 507 519 12408852 10.1016/S0896-6273(02)00978-9
Kang H Schuman EM A requirement for local protein synthesis in neurotrophin-induced hippocampal synaptic plasticity Science 1996 273 1402 1406 8703078
Reymond A Marigo V Yaylaoglu MB Leoni A Ucla C Scamuffa N Caccioppoli C Dermitzakis ET Lyle R Banfi S Eichele G Antonarakis SE Ballabio A Human chromosome 21 gene expression atlas in the mouse Nature 2002 420 582 586 12466854 10.1038/nature01178
Gray PA Fu H Luo P Zhao Q Yu J Ferrari A Tenzen T Yuk DI Tsung EF Cai Z Alberta JA Cheng LP Liu Y Stenman JM Valerius MT Billings N Kim HA Greenberg ME McMahon AP Rowitch DH Stiles CD Ma Q Mouse brain organization revealed through direct genome-scale TF expression analysis Science 2004 306 2255 2257 15618518 10.1126/science.1104935
Gitton Y Dahmane N Baik S Ruiz i Altaba A Neidhardt L Scholze M Herrmann BG Kahlem P Benkahla A Schrinner S Yildirimman R Herwig R Lehrach H Yaspo ML A gene expression map of human chromosome 21 orthologues in the mouse Nature 2002 420 586 590 12466855 10.1038/nature01270
Saunders LR Barber GN The dsRNA binding protein family: critical roles, diverse cellular functions Faseb J 2003 17 961 983 12773480 10.1096/fj.02-0958rev
Nagai K RNA-protein complexes Curr Opin Struct Biol 1996 6 53 61 8696973 10.1016/S0959-440X(96)80095-9
Burd CG Dreyfuss G Conserved structures and diversity of functions of RNA-binding proteins Science 1994 265 615 621 8036511
Wheeler DL Church DM Edgar R Federhen S Helmberg W Madden TL Pontius JU Schuler GD Schriml LM Sequeira E Suzek TO Tatusova TA Wagner L Database resources of the National Center for Biotechnology Information: update Nucleic Acids Res 2004 32 D35 40 14681353 10.1093/nar/gkh073
Sonnhammer EL Eddy SR Birney E Bateman A Durbin R Pfam: multiple sequence alignments and HMM-profiles of protein domains Nucleic Acids Res 1998 26 320 322 9399864 10.1093/nar/26.1.320
Thomas PD Campbell MJ Kejariwal A Mi H Karlak B Daverman R Diemer K Muruganujan A Narechania A PANTHER: a library of protein families and subfamilies indexed by function Genome Res 2003 13 2129 2141 12952881 10.1101/gr.772403
http://mahoney.chip.org/mahoney/RBP
Buckanovich RJ Posner JB Darnell RB Nova, the paraneoplastic Ri antigen, is homologous to an RNA-binding protein and is specifically expressed in the developing motor system Neuron 1993 11 657 672 8398153 10.1016/0896-6273(93)90077-5
Okano HJ Darnell RB A hierarchy of Hu RNA binding proteins in developing and adult neurons J Neurosci 1997 17 3024 3037 9096138
Kiehl TR Shibata H Vo T Huynh DP Pulst SM Identification and expression of a mouse ortholog of A2BP1 Mamm Genome 2001 12 595 601 11471052 10.1007/s00335-001-2056-4
Sakakibara S Okano H Expression of neural RNA-binding proteins in the postnatal CNS: implications of their roles in neuronal and glial cell development J Neurosci 1997 17 8300 8312 9334405
Sakakibara S Nakamura Y Satoh H Okano H Rna-binding protein Musashi2: developmentally regulated expression in neural precursor cells and subpopulations of neurons in mammalian CNS J Neurosci 2001 21 8091 8107 11588182
Blackshaw S Harpavat S Trimarchi J Cai L Huang H Kuo WP Weber G Lee K Fraioli RE Cho SH Yung R Asch E Ohno-Machado L Wong WH Cepko CL Genomic analysis of mouse retinal development PLoS Biol 2004 2 E247 15226823 10.1371/journal.pbio.0020247
Lillevali K Kulla A Ord T Comparative expression analysis of the genes encoding polypyrimidine tract binding protein (PTB) and its neural homologue (brPTB) in prenatal and postnatal mouse brain Mech Dev 2001 101 217 220 11231079 10.1016/S0925-4773(00)00566-9
Zhang W Liu H Han K Grabowski PJ Region-specific alternative splicing in the nervous system: implications for regulation by the RNA-binding protein NAPOR Rna 2002 8 671 685 12022233 10.1017/S1355838202027036
Grabowski PJ Black DL Alternative RNA splicing in the nervous system Prog Neurobiol 2001 65 289 308 11473790 10.1016/S0301-0082(01)00007-7
Niehrs C Pollet N Synexpression groups in eukaryotes Nature 1999 402 483 487 10591207 10.1038/990025
Patton JG Porro EB Galceran J Tempst P Nadal-Ginard B Cloning and characterization of PSF, a novel pre-mRNA splicing factor Genes Dev 1993 7 393 406 8449401
Hahm B Cho OH Kim JE Kim YK Kim JH Oh YL Jang SK Polypyrimidine tract-binding protein interacts with HnRNP L FEBS Lett 1998 425 401 406 9563502 10.1016/S0014-5793(98)00269-5
Eperon IC Makarova OV Mayeda A Munroe SH Caceres JF Hayward DG Krainer AR Selection of alternative 5' splice sites: role of U1 snRNP and models for the antagonistic effects of SF2/ASF and hnRNP A1 Mol Cell Biol 2000 20 8303 8318 11046128 10.1128/MCB.20.22.8303-8318.2000
Hieronymus H Silver PA A systems view of mRNP biology Genes Dev 2004 18 2845 2860 15574591 10.1101/gad.1256904
Mody M Cao Y Cui Z Tay KY Shyong A Shimizu E Pham K Schultz P Welsh D Tsien JZ Genome-wide gene expression profiles of the developing mouse hippocampus Proc Natl Acad Sci U S A 2001 98 8862 8867 11438693 10.1073/pnas.141244998
Mills JC Andersson N Hong CV Stappenbeck TS Gordon JI Molecular characterization of mouse gastric epithelial progenitor cells Proc Natl Acad Sci U S A 2002 99 14819 14824 12409607 10.1073/pnas.192574799
Stappenbeck TS Hooper LV Gordon JI Developmental regulation of intestinal angiogenesis by indigenous microbes via Paneth cells Proc Natl Acad Sci U S A 2002 99 15451 15455 12432102 10.1073/pnas.202604299
Harris MN Ozpolat B Abdi F Gu S Legler A Mawuenyega KG Tirado-Gomez M Lopez-Berestein G Chen X Comparative proteomic analysis of all-trans-retinoic acid treatment reveals systematic posttranscriptional control mechanisms in acute promyelocytic leukemia Blood 2004 104 1314 1323 15142884 10.1182/blood-2004-01-0046
|
16033648
|
PMC1199591
|
CC BY
|
2021-01-04 16:30:32
|
no
|
BMC Dev Biol. 2005 Jul 20; 5:14
|
utf-8
|
BMC Dev Biol
| 2,005 |
10.1186/1471-213X-5-14
|
oa_comm
|
==== Front
BMC Evol BiolBMC Evolutionary Biology1471-2148BioMed Central London 1471-2148-5-421608350810.1186/1471-2148-5-42Research ArticleEvolution of a microbial nitrilase gene family: a comparative and environmental genomics study Podar Mircea [email protected] Jonathan R [email protected] Toby H [email protected] Diversa Corporation, 4955 Directors Place, San Diego, CA 92131 USA2005 6 8 2005 5 42 42 5 5 2005 6 8 2005 Copyright © 2005 Podar et al; licensee BioMed Central Ltd.2005Podar et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Completed genomes and environmental genomic sequences are bringing a significant contribution to understanding the evolution of gene families, microbial metabolism and community eco-physiology. Here, we used comparative genomics and phylogenetic analyses in conjunction with enzymatic data to probe the evolution and functions of a microbial nitrilase gene family. Nitrilases are relatively rare in bacterial genomes, their biological function being unclear.
Results
We examined the genetic neighborhood of the different subfamily genes and discovered conserved gene clusters or operons associated with specific nitrilase clades. The inferred evolutionary transitions that separate nitrilases which belong to different gene clusters correlated with changes in their enzymatic properties. We present evidence that Darwinian adaptation acted during one of those transitions and identified sites in the enzyme that may have been under positive selection.
Conclusion
Changes in the observed biochemical properties of the nitrilases associated with the different gene clusters are consistent with a hypothesis that those enzymes have been recruited to a novel metabolic pathway following gene duplication and neofunctionalization. These results demonstrate the benefits of combining environmental genomic sampling and completed genomes data with evolutionary and biochemical analyses in the study of gene families. They also open new directions for studying the functions of nitrilases and the genes they are associated with.
==== Body
Background
Having colonized virtually every environment, bacteria and archaea have evolved enzymatic solutions for a wide range of metabolic biochemical transformations [1,2]. Studying enzymes derived from organisms inhabiting these environments is important for understanding how microbes adapt, react to and transform the environment. The overwhelming majority of microbial species remain however uncultivated [3]. A variety of functional and sequence-based approaches have been developed for discovering and characterizing genes, operons and even entire genomes directly from the environment, collectively referred to as metagenomics or environmental genomics [4]. The use of environmental genomics has already led to important discoveries such as genes responsible for novel biological functions [5], microbial community metabolic traits [6-8] and dramatic increases in the diversity of various enzyme families [9,10]. Subsequent biochemical and evolutionary analyses can strengthen the biological end ecological inferences even before organisms that carry that genetic information are isolated in culture [11-13]. From a practical perspective, microbial environmental genomics has been a successful approach for the discovery of enzymes for a broad spectrum of biotechnological applications [14-17].
To gain insight into the evolution of function in a gene family that has been extensively sampled by environmental genomic screening and characterized biochemically, we focused on bacterial nitrilases. These enzymes are members of the carbon-nitrogen hydrolase superfamily which catalyze the hydrolysis of a wide range of non-peptide carbon-nitrogen bonds [18-20]. The nitrilase family hydrolyzes nitriles to their corresponding carboxylic acids, releasing ammonia. This reaction is likely involved in detoxification of xenobiotics and nitriles produced as defense chemicals by other microorganisms and plants, as well as in secondary metabolite biosynthetic pathways. Nitrilases appear to be rare in bacteria (out of over 150 sequenced bacterial genomes only 10 contain nitrilase genes). Recently, over 130 nitrilases were identified by functional screening of hundreds of environmental DNA libraries, for use in industrial biocatalysis applications [9]. Those enzymes were characterized biochemically and classified into six subfamilies, four of them with no representatives in known bacterial species. It was found that a number of enzymatic properties (substrate specificity and enantioselectivity) were specific to subfamilies and, in some cases, correlated with the biogeography and ecology of the environmental samples.
The role of gene duplication, natural selection and functional diversification in the evolution of the nitrilase gene family is unknown. The correlation of distinct enzymatic properties with the different genes subfamilies suggest that nitrilases have diverged functionally to accommodate distinct biological roles in microbial communities that occupy various ecological niches. Functional divergence is the result of changes in selection pressure and is often accompanied by associations with novel gene clusters or operons which encode for enzymes with coupled metabolic activities. To begin addressing some of these aspects, we analyzed the genetic neighborhoods of all available nitrilase genes, identified conserved patterns of conserved gene clustering relative to biochemical data and phylogeny and propose a hypothesis on nitrilase evolution involving gene duplications and Darwinian selection.
Results and discussion
The nitrilases from cultivated bacteria belong to clade-specific gene clusters
Bacterial nitrilases (137 environmental sequences and 10 sequences from cultivated species) have been recently classified into six major clades [9] that we refer to as subfamilies. We analyzed more recently released genome sequences and found an additional nine novel nitrilases. Phylogenetic analysis of a sequence dataset consisting of all nitrilase genes from cultivated bacteria shows that 18 sequences belong to subfamilies one and two (Fig. 1). The level of sequence similarity among these 18 enzymes is quite high, ranging from 50–70% pairwise identity in subfamily one to 30–40% in subfamily two. The relationships between the different nitrilases do not reflect the taxonomy of their host organisms. Additionally, for several genera or species that harbor two nitrilases (Pseudomonas, Klebsiella pneumoniae and Burkholderia fungorum), the genes belong to different subfamilies/clades, suggesting ancient gene duplications or acquisition by horizontal gene transfer (HGT). Rhodococcus rhodochrous on the other hand contains two closely related nitrilases, suggesting a more recent gene duplication event. Supporting the possibility of HGT, one of the nitrilase genes we identified by database mining is in the plasmid pLVPK of Klebsiella pneumoniae, which may be transferable to other bacteria. Also, several fungal cyanide hydratase genes form a clade deeply nested within subfamily two of bacterial nitrilases, suggesting HGT acquisition from bacteria, followed by neofunctionalization. The paucity of nitrilase genes in bacterial genomes makes it difficult to evaluate the contribution of the different evolutionary events (duplications, gene loss and HGT) to the observed distribution and the functional significance of the presence of different types of enzymes in related organisms.
Figure 1 Maximum likelihood tree of nitrilases from known bacterial species (accession numbers are in parentheses). Bootstrap support values are indicated for the major groups only. The schematic organization of the gene clusters that contain a nitrilase ORF is shown for species where that sequence information is available.
In bacteria, genes are often organized in clusters (e.g. operons, regulons) that reflect involvement in a common metabolic process or association in a supramolecular complex [21-23]. To determine if nitrilase function could be inferred from the nature of the surrounding genes, we analyzed those genes in the available genomic data. We found that all of the known seven subfamily 1 nitrilase genes (six genomic and one on a plasmid) belong to a conserved and previously undescribed cluster of seven genes, Nit1C (Figure 1 and Figure 2). Six of the coding sequences are on the same DNA strand, separated by few or no intergenic nucleotides and are likely part of an operon/regulon. This hypothesis is supported by analysis using a recent method for operon prediction [24] although we could not identify conserved transcription factor binding sites in the upstream region. The genes in this predicted operon occur in the order (1) hypothetical protein, (2) nitrilase, (3) radical S-adenosyl methionine superfamily member, (4) acetyltransferase, (5) AIR synthase, and (6) hypothetical protein. The seventh gene encodes a predicted flavoprotein, putatively involved in K+ transport and is located either at the beginning of the cluster but on the opposite strand (cyanobacteria Synechocystis sp. PCC6803 and Synechococcus sp. WH8102) or as the last gene of the cluster, in the same orientation as the others (proteobacteria Burkholderia fungorum, Rubrivivax, Photorhabdus luminescens and Klebsiella pneumoniae). In Verrucomicrobium spinosum, the cluster has been rearranged, as ORFs 6 and 7 occur in between ORFs 3 and 4. Yet another variation exists in the betaproteobacteria Burkholderia and Rubrivivax where a glycosyltransferase gene is inserted between ORFs 5 and 6. These slight variations in the cluster architecture correlate to the major taxonomic bacterial groups (Cyanobacteria, Beta- and Gamma proteobacteria). Outside of Nit1C there is no conservation between the different species in terms of genes or metabolic functions encoded by gene clusters. The presence of genes associated with mobile DNA elements (transposases, IS elements) immediately downstream of the Nit1C clusters in Synechocystis and Photorhabdus and the apparent interruption of a large polyketide synthase pathway by the nitrilase cluster in Photorhabdus may indicate HGT or internal chromosomal rearrangements.
Figure 2 Organization of gene clusters around the subfamily 1 nitrilases in sequenced bacterial genomes. The highly conserved gene cluster Nit1C is flanked by unrelated genomic neighbourhoods in the different species. Gene names are based on the available genomic annotation.
In the case of subfamily 2, gene neighborhood information was available for only four of the twelve genes from cultivated bacteria. In Bacillus sp. and Pseudomonas syringae, the nitrilase gene is apparently co-transcribed with a downstream phenylacetaldoxime dehydratase gene and preceded by an araC transcription factor transcribed from the other strand. The other nitrilase genes (from Burkholderia, Bradyrhizobium and Ralstonia) are part of unrelated clusters (Figure 1).
In addition to the nitrilases from completed genomes of cultivated bacteria, we searched for such enzymes in two large environmental sequence datasets: the acid-mine drainage microbial mats [7] and the Sargasso Sea [10] using BLASTP. No nitrilases were found in the acid-mine dataset. In the Sargasso Sea dataset we identified 17 nitrilases that were full-length or long enough to be phylogenetically informative. Three of the genes appear to be eukaryotic while eight bacterial genes are close relatives to nitrilases from Synechoocccus or Burkholderia. The remaining six genes do not appear to have close relatives among known nitrilases and belong to subfamilies 2, 4 and 5 [see Additional file 1]. Finding so few nitrilase genes in such a large dataset suggests that for uncovering the sequence space of a gene family, functional screening of a large number of samples from very different environments is more efficient than deep sequence coverage of one or a few environments.
Nitrilases associated with different types of gene clusters have distinct enzymatic properties
For the nitrilase genes identified from environmental DNA, the identity of the host organism is unknown. However, because those libraries were constructed using fragments of genomic DNA several times larger than the average nitrilase gene length (~1 kb), we also analyzed the the gene neighborhood of the environmental nitrilase. Because of the highly conserved nature of the Nit1C cluster and its occurrence in distant taxa of bacteria, we first focused on mapping its distribution among the environmental nitrilase clones. We found that the Nit1C cluster is strictly confined to a group of subfamily 1 nitrilases that includes the seven genes identified in completed genomes and 14 of the environmental ones. Four of the subfamily 1 nitrilases from the Sargasso Sea dataset had small flanking sequences and we identified the presence of the Nit1C type genes (ORFs 1 or 3), similar to those of their close relatives from Synechococcus and Burkholderia. However, because of their incomplete length, those sequences were not included in further analyses.
The nitrilase genes that belong to the Nit1C cluster are indicated on a maximum likelihood phylogenetic tree calculated using the subfamily 1 genes as well as several outgroup sequences from subfamilies 2 and 3 (Figure 3A). Since the size of the genomic insert in the environmental clones was limited, not all the Nit1C genes were identified; however, we did not find evidence to suggest that the cluster was different in any of the host genomes (Figure 3B). We also identified a more recent evolutionary event that marks the loss of nitrilase association with the Nit1C cluster. After that transition event (TE), nitrilase genes are no longer associated with a highly conserved gene cluster. Instead, they are flanked by genes encoding MarR transcriptional regulators, epimerases, epoxide hydrolases and other ORFs. These latter genes were not so highly conserved in their order as those found in the Nit1C cluster. No cultivated bacteria that contain nitrilases from this group have been found so far.
Figure 3 (A). Protein maximum likelihood tree of subfamily 1 nitrilases. The tree was arbitrarily rooted with sequences from the two most closely related subfamilies 2 and 3. Numbers at nodes represent bootstrap support (not shown if <50). (B). Diagram of the gene clusters that include the nitrilase ORF. For environmental genes, the information was limited by the size of the genomic insert. (C). Histogram representing enzymatic enantioselectivity (R or S) on hydroxyglutaronitrile, based on data from [9](na, not assayed; x, not active).
The sister group of subfamily 1 nitrilases, subfamily 3, consists of only three environmental type genes. We had sufficient flanking sequence to determine the nature of the neighboring genes for only one of the genes (3A1), flanked by two hypothetical ORFs with no identifiable homologs. Therefore, the Nit1C cluster appears to have originated with and is restricted to a subset of subfamily 1 nitrilases. The more distantly related nitrilases from subfamilies 4, 5 and 6 have no apparent associations with a conserved gene cluster (data not shown).
In our previous study [9] we uncovered a number of correlations between the biochemical properties of the environmental microbial nitrilases and their phylogenetic classification. Distinct gains or losses of activity or switches in enantioselectivity coincided with the evolutionary events that led to the formation of the main subfamilies. One of the most interesting findings was a reversal in enantioselectivity (R to S) that occurred in subfamily 1, against the model substrate hydroxyglutaronitrile. To correlate the differences in types of gene clusters with the nitrilase biochemical properties, we graphed the available hydroxyglutaronitrile activity data on the side of the phylogenetic tree (Figure 3C). With one exception (1B15), the enzymes that belong to the Nit1C group are R-enantioselective on hydroxyglutaronitrile. The transition event (TE) marks changes in biochemical properties leading to enantioselectivity reversal. The first enzyme not associated with Nit1C (1A21) was inactive on that substrate, while the next diverging ones (1A20, 1A22, 1A16, 1A17) were R-selective or not enantioselective (low bootstrap values do not support a robust branching order). However, the next statistically supported clade (1A14 and above in the Figure 3A tree) show a reversal of enantioselectivity followed by a steep increase in selectivity to values over 95%.
Analysis of the subfamily 1 nitrilase gene clusters
Having determined that subfamily 1 nitrilases belong to two distinct subgroups based on their associated gene clusters and enzymatic properties, we analyzed the nitrilase neighboring genes for clues to their individual metabolic roles. First in the Nit1C cluster, ORF1 proteins are highly conserved in length (160–163 amino acids) and sequence (>60% identity between any two genes). However, no other homologs were found using standard searching techniques of current databases. Using HMM structural homology modeling (Superfamily 1.63 server) [25], we tentatively assigned the hypothetical protein 1 to the YchN1-like superfamily and fold, whose biochemical activity is unknown. Next in the cluster is the nitrilase gene. The third gene encodes a member of the radical SAM superfamily (Pfam 04055), enzymes that catalyze a wide variety of radical-based reactions through reductive cleavage of S-adenosylmethionine at an iron-sulfur center [26]. The Nit1C SAM genes form a strongly supported clade (~50% average sequence identity), most closely related to bacterial and archaeal genes annotated as biotin synthase-related enzymes (COG2516) [see Additional file 2]. ORF4 in the Nit1C cluster also forms a clade of closely related sequences and belong to the GCN5-related N-acetyltransferase (GNAT) superfamily (Pfam 00583) [27]. These enzymes are involved in antibiotic detoxification as well as in histone acetylation in eukaryotes. The closest homologs to the Nit1C GNAT genes are a number of other acetylases from bacteria like Rhodobacter and Enterococcus [see Additional file 2]. The fifth gene in the cluster encodes members of the large 5'-phosphorybosyl-5-aminoimidazole synthase-related proteins superfamily (AIRS, Pfam 00586). Enzymes in this superfamily are involved in de novo purine biosynthesis, selenophosphate synthesis, or maturation of NifE hydrogenase. These genes form a unique clade, most closely related to a group of archaeal genes encoding phosphoribosylformylglycinamide synthases [see Additional file 2]. The last invariant position in the cluster, ORF6, encodes a protein of approximately 100 amino acids. While the sequence identity between the individual genes surpasses 70%, we could not find any other relatives to these genes by any sequence analysis approach. The seventh ORF of Nit1C is located at either end of the cluster, on either coding strand. This gene is a member of the pyridine nucleotide-disulphide oxidoreductases (Pfam 00070, COG2072), that include flavin-containing monooxygenases and flavoproteins involved in K+ transport. The closest relatives to the Nit1C genes are putative monooxygenases found in several species of Pseudomonas [see Additional file 2]. All Nit1C genes form clusters of closely related sequences within their respective superfamilies, suggesting a common function, possibly in a pathway for detoxification of plant or microbial defense compounds.
Members of the nitrilase clade that split after the transition event are exclusively of environmental origin, with no sequence representatives in characterized bacterial species. Approximately two thirds of the nitrilases in this group are associated with genes encoding a MarR transcriptional regulator, epimerases and epoxide hydrolases. MarR genes (PFam 01047) are transcriptional repressors controlling the expression of the Mar operon, involved in multiple antibiotic resistances [28]. The nitrilase-associated MarR genes form a specific clade, most closely related to genes from Xanthomonas and Desulfitobacterium (30–40% identity) [see Additional file 3] and are always upstream of the nitrilase gene. The location of the epimerase and epoxide hydrolase varies somewhat, the epimerase ORF being usually between the nitrilase and the epoxide hydrolase ORFs. Epimerases are a large class of enzymes that reversibly determine stereochemical inversions of hydroxyl substituents in carbohydrates, participating in numerous metabolic pathways [29,30]. The nitrilase-associated epimerases form a unique clade in which the relationship between the genes parallels that of their associated nitrilases. Their closest relatives are epimerases from species of Streptomyces (~35% identity) [see Additional file 3]. Epoxide hydrolases belong to the large superfamily of alpha-beta fold hydrolases and hydrate chemically reactive epoxides to more stable dihydrodiols. This reaction is of major importance in detoxification of a large number of endogenous epoxide metabolites and xenobiotic compounds in all organisms [31]. The association of all these genes with nitrilases could indicate the requirement for coupled reactions under the transcriptional control of MarR, perhaps involved in detoxifying sugar-based cyanogenic compounds in soils rich in decaying plant material.
Positive selection as a possible driving force for nitrilase functional diversification
The observed changes in associated gene clusters and in enzymatic properties suggest that the hypothetical gene duplication in subfamily 1 was followed by nitrilase recruitment to novel metabolic functions, possibly under selective constraints. A powerful approach to studying changes in the selective pressure in protein encoding genes involves calculation of the nonsynonymous/synonymous substitution rate ratio (ω = dN/dS) (reviewed in [32,33]). A ratio below one indicates negative (purifying) selection, restricting amino acid changes that could interfere with a well-established protein function, while ω = 1 suggests that the gene evolves neutrally. On the other hand, a ratio significantly higher than one may indicate a selective advantage for fixation of amino acid changes. This can be considered evidence of positive selection associated with functional divergence after events such as gene duplications or changes in the environment (e.g. [34,35]).
Using a relative rate test [36], we first investigated the rate variation between the branches flanking the transition event (1A23/1A25 and 1A21). A likelihood ratio test based on a three-taxon tree (consisting of 1A25 and 1A21 as test sequences and 1A29 as outgroup) compared the null hypothesis (equal rates for both branches following the transition event) with an alternative model with unconstrained rates. The null model was rejected (P = 2 × 10-6, df = 1), supporting a 5.6 times faster overall rate for the 1A21 lineage than for 1A25, which has maintained the Nit1C association. A rate increase is predicted when gene duplication is followed by functional divergence and could occur because of positive Darwinian selection or an increase in fixation of neutral mutations as result of relaxation of functional constraints [37-40].
To test if positive selection acted along the nitrilase lineages flanking the cluster transition event, we used a maximum likelihood (ML) approach based on codon substitution models [34]. These models take into account sequence features such as transition-transversion rate biases, codon usage variation and allow testing hypotheses at specific branches in a phylogeny by employing heterogeneous ω values among sites and lineages. Positive selection can also be investigated using a parsimony-based method, there being some controversy on to which of the two methods is more reliable [41-43].
The tree used for ω estimation was obtained based on the nitrilase DNA sequences, focusing on the genes around the transition event (Figure 6A). The first set of likelihood models that we used, site-specific [44], assume variations in the selective pressure across sites but no variations among individual genes. Using these models we determined that purifying selection has a dominant role across subfamily 1 nitrilases (ω = 0.04) (Table 1). This is reflected in the large number of conserved amino acids: 86 invariant (~25% of sites) and 149 conserved at 90% level in this data set. No significant positive selection signal was identified using this category of models. However, since these models average the substitution ratios of individual sites over all lineages, they are known to lack sensitivity in detecting positive selection that acts only along a few lineages (e.g. [44,45].
Table 1 Parameter estimates, likelihood scores and identified selected sites under various models. Branch numbers refer to Figure 4A. Parameters indicating positive selection are in bold. A likelihood ratio test (LRT) is used to compare a pair of nested models: one which accounts for sites with ω > 1 and one which does not (the null model). To accept or reject the ω > 1 hypothesis, twice the log-likelihood difference in the scores is compared with a χ2 distribution with the degrees of freedom equal to the difference in the numbers of parameters between the two models. When ML detects lineages with ω > 1, an empirical Bayes analysis identifies sites under positive selection and calculate posterior probabilities that provide a measure of confidence for that prediction.
Model p l Parameter estimates Positively selected sites Likelihood Ratio Test
M0:one ratio 1 -11903.5 ω = 0.0418 none
Site-specific models
M1:neutral (K = 2) 1 -13195.5 p0 = 0.298, p1 = 0.702 not allowed
M3:discrete (K = 2) 3 -11627.6 p0 = 0.6, p1 = 0.4, ω0 = 0.012, ω1 = 0.098 none
Branch-site models
Branch 1
Model A 3 -13160.0 p0 = 0.3, p1 = 0.70, p2+p3 = 0, ω2 = 0 none
Model B 5 -11627.6 p0 = 0.4, p1 = 0.6, p2+p3 = 0
ω0 = 0.098, ω1 = 0.012, ω2 = 0 none
Branch 2
Model A 3 -13188.7 p0 = 0.296, p1 = 0.688, p2+p3 = 0.016, ω2 = 129.6 Q157 (P = 0.77), Q203 (P = 0.999), T41, Q157, Y184, N200, Q203, R284 (P > 0.9) LRT vs. M1 2Δl = 6.8, P = 0.03, df = 2
Model B 5 -11621.4 p0 = 0.356, p1 = 0.59, p2+p3 = 0.05
ω0 = 0.1, ω1 = 0.0125, ω2 = 9.7 LRT vs. M3 (K = 2) 2Δl = 6.2, P = 0.04, df = 2
To investigate if adaptive evolution acted alongside branches around the transition event, we also used a more recently developed set of maximum likelihood models, which allow the ω ratio to vary among both sites and lineages [46]. These models are more sensitive in detecting positively selected sites along a pre-specified lineage of interest ("foreground" branch) as compared to the rest of the genes ("background" branches). These models were applied to the two lineages that followed the transition event (branches 1 and 2 in Figure 4A). For branch 1, which belongs to the Nit1C nitrilases and served as a negative control, we did not detect any positive selection signal. Branch 2 represents the basal lineage for the group of nitrilase genes that have lost the Nit1C cluster association, potentially having led to nitrilase neofunctionalization. A significant positive selection pressure (ω = 9.7 under model B) was detected for that lineage, the empirical Bayes analysis pointing to residues T41, Q157, Y184, N200, Q203 and R284 as being the selection target. These amino acid positions may represent hot spots for changes in substrate specificity or other nitrilase enzymatic properties. The variation of those aminoacids across the subfamily is shown in Figure 4. Shown also is a site (residue 39) that is invariant before the transition event then changes with that event and becomes again invariant.
Figure 4 (A) Maximum likelihood tree for subfamily 1 nitrilases used to test for positive selection. Branch lengths are scaled to the mean number of substitutions per codon site under model M3. Branches 1 and 2 indicate lineages tested for positive selection signal, following the transition event. The sequences illustrate the variability across the clade at positions identified under positive selection. (B). A three dimensional model of the 1A21 nitrilase dimer. Shown are the catalytic triad (blue) and the residues under positive selection (red). Residue 39, invariant before and after the transition event, is shown in green.
High resolution structures are not yet available for nitrilases. However, the structures of two homologs, the C. elegans NitFhit protein and the Agrobacterium radiobacter N-carbamoyl-D-amino acid amidohydrolase (D-NCAase) have been solved [47,48]. Both proteins form tetramers with two dimer subunits and revealed a novel four layer α-β-β-α fold. It is believed that all members of the nitrilase superfamily share this fold and the catalytic triad Glu-Lys-Cys in the active site. A three dimensional model of 1A21 (the first nitrilase outside the Nit1C group) was derived based on the D-NCAase structure coordinates, and used to map the location of the residues under positive selection at the CTE. Three of those, T41, Q157 and Y184, were found to be buried within the protein, close to the catalytic triad (E44, K126, C160) (Figure 4B). Those residues could be involved in the overall conformation of the active site or may have a direct role in the reaction by interacting with the substrate. The other three positively selected sites, N200, Q203 and R284 cluster on the surface interface between the molecules of the dimer. That interface has been shown in D-NCAase to form a hydrophobic pocket that is responsible for the tight dimer structure. It is known that the quaternary structures of nitrilases and cyanide hydratases can be quite different, ranging in size from monomers and dimers to oligomers containing 10, 14 or more subunits. Substrate binding has also been shown to play a role in the formation of active enzyme oligomers. The three interface residues may play a role in aspects of quaternary structure and substrate specificity associated with the proposed neofunctionalization after the cluster transition event.
Conclusion
In this study, we combined genomic and biochemical analysis of a microbial enzyme family to understand evolutionary events that have shaped the genome organization and metabolism of organisms inhabiting various environments. It has long been known that bacterial genes often cluster based on linked functions. The gene location sometimes correlates with the order of the individual reactions in an enzymatic cascade or facilitate regulatory mechanisms of gene expression. Various models have been proposed to explain the formation, the evolutionary and physiological significance of operons and other gene clusters [23]. Comparative genomic studies have shown that recognition of clusters can assist in functional annotation of novel genes but clusters often they break apart with increasing taxonomic distance [49-53]. The Nit1C cluster that we described is remarkable in that it is highly conserved across several bacterial phyla and is present in organisms that inhabit extremely diverse environments. While limited rearrangements have occurred in Nit1C, the preservation of all seven genes suggests there is selective pressure for maintenance of the entire gene cluster regardless of the genomic dynamics in that neighborhood. The internal rearrangements of Nit1C correlate with high level taxa (cyanobacteria, beta and gamma proteobacteria).
There is no experimental evidence for an involvement of any of the Nit1C genes in a known metabolic transformation. Two of the cluster genes have no close homologs or predictable biochemical activities while the remaining genes, even though have a predictable type of biochemical activity, belong to classes of enzymes that are involved in a wide range of transformations. Predicting function for remote homologs in the absence of experimental data is still a major difficulty in genomics [54,55]. Having a defined cluster of genes such as Nit1C, likely to be functionally connected, sets the ground for future experimental genetic and biochemical investigation in search of its biological function.
Phylogenetically, the nitrilases from the Nit1C cluster appear strictly confined to a basal subset of subfamily 1 genes. More recent diversification of the genes in this subfamily has been accompanied by a change in the type of associated gene clusters and is paralleled by changes in biochemical properties of the nitrilases. While overall, subfamily 1 nitrilases are under strong purifying selection pressure, we detected a significant positive selection signal for the lineage following the transition event and identified several residues under such selection. This supports a hypothesis that a group of nitrilases diverged functionally from the Nit1C-type enzymes, became associated with other metabolic enzymes possibly as part of a novel pathway and advantageous mutations were fixed at specific sites under positive selection. Future studies of bacterial nitrilases and biochemical and genetic characterization of mutations at these residues are needed to better understand the determinants of substrate specificity and the functional differences between the nitrilase subfamilies.
Environmental microbial genomics has demonstrated its utility in studying large scale ecological processes [5,6,11], discovering valuable biocatalysts [15] and reassembling the genomic and metabolic blueprint of natural microbial communities thorough shotgun sequencing [7,8,10]. Vast amounts of sequence data could potentially be used to answer a wide range of questions, although there are open questions regarding experimental design, data analysis and breadth of biological significance [4,56,57]. A broad environmental sampling from worldwide geographical locations coupled with experimental biochemical validation and comparative genomic analysis allowed us to test metabolic and evolutionary hypotheses difficult to approach by using sequence data from only a few environments.
Methods
DNA sequences
The nitrilase sequences discovered from environmental DNA libraries are available from Genbank (AY487426-AY487562). Nitrilase sequences from sequenced bacterial genomes and their corresponding flanking genes were also obtained from GenBank, their names and accession numbers being indicated in the corresponding figures. For Verrucomicrobium spinosum DSM 4136, preliminary sequence data was obtained from the The Institute for Genome Research website [58] and for Burkholderia fungorum and Rubrivivax gelatinosus from the DOE Joint Genome Institute website [59].
Enzymatic activity
The biochemical characterization data used in this study for the environmental nitrilases tested on the non physiological substrate hydroxyglutaronitrile has been published [9].
Sequence analysis and annotation
For the analysis of the ORFs flanking the nitrilase genes in known bacterial genomes we used the sequence coordinates available in the corresponding GenBank files. For the environmental DNA clones containing nitrilase genes we identified and annotated the other open reading frames (ORFs) contiguous with the nitrilase in the genomic insert using standard approaches. The inserts varied in size from 1 to 7 kb and in most cases contained information to identify at least one or more ORFs in addition to the nitrilase gene. Annotation was derived based on available experimental or predicted function or biochemical activity using information associated with those genes in GenBank, PFAM, COG and KEGG databases.
Phylogenetic reconstructions
Amino acid sequences were aligned in BioEdit [60] followed by manual refinement. Sequence alignments are provided [see Additional files 4, 5]. Phylogenetic trees were constructed in PROML (PHYLIP 3.6) [61] using maximum likelihood, JTT amino acid substitution matrix, five global rearrangements with randomized sequence input order and among-site rate variation modeled with an eight rate category discrete approximation to a gamma distribution. The model parameters were estimated using TREE-PUZZLE 5.1. [62]. Branch support was obtained by bootstrapping (100 replicates).
Analysis for positive selection
A DNA sequence alignment for the nitrilase genes was obtained based on the protein alignment and used for phylogenetic reconstructions in PAUP* 4.0 [63] using maximum likelihood and is provided [see Additional file 6]. The model of sequence evolution (GTR+I+G) was selected using Modeltest v.3.06 [64]. To test specific branches for possible rate changes we used Hy-Phy [36]. The topologies for the DNA tree and the protein tree were identical.
The tree topology was used in the program codeml (PAML [65], to estimate dN/dS ratios based on maximum likelihood codon substitution models. Two categories of models were used, site specific [44] as well as branch-site models [46]. Statistical comparisons between the results from different nested models were done using likelihood ratio tests [66].
Molecular modeling
A three-dimensional model for a clade 1 nitrilase (1A21) was obtained based on the structure of the homologous protein N-carbamoyl-D-amino acid amidohydrolase [48], using the Jackal software [67]. Analysis of the model and mapping of amino acid residues involved in catalysis or subject to positive selection was done in PyMol [68].
Authors' contributions
MP participated in the design of the study, performed phylogenetic, comparative genomic and statistical analyses and drafted the manuscript. JE performed sequence analysis and functional annotation. TR participated in the design of the study, performed comparative genomic and gene function analyses. All authors contributed to the writing and approved the final manuscript.
Supplementary Material
Additional file 1
Protein neighbor-joining tree for nitrilase genes from cultivated bacteria and from environmental samples. The environmental sequences are represented by GenBank accession numbers and gene names for those derived from Robertson et al, 2004. The Sargasso Sea sequences are shaded.
Click here for file
Additional file 2
Maximum likelihood phylogenetic trees for genes that belong to the Nit1C clusters identified in known bacterial species, in the context of their respective protein families. Numbers represent bootstrap support (for major clades only). The Nit1C ORF sequences are shaded.
Click here for file
Additional file 3
Maximum likelihood phylogenetic trees for two genes associated with nitrilases after the subfamily 1 cluster transition event, in the context of their respective larger protein families. The nitrilase associated genes are shaded. Numbers represent bootstrap support (for major clades only).
Click here for file
Additional files 4
Alignment of nitrilase amino acid sequences from cultivated bacteria (used to generate the tree in Figure 1)
Click here for file
Additional files 5
Alignment of nitrilase amino acid sequences used to generate the tree in Figure 3.
Click here for file
Additional file 6
Alignment of DNA sequences of nitrilase genes used to test for positive selection and to generate the tree in Figure 4.
Click here for file
Acknowledgements
We thank Jay Short and Michiel Noordewier for their support and guidance, the Diversa Research and Development team, especially, Dan Robertson, Jenny Chaplin and Grace Desantis for leading the nitrilase discovery and characterization projects, David Lomelin and Cosmin Deciu for bioinformatics analysis support and Mark Wall for the three dimensional model of the nitrilase. Special thanks also to Melvin Simon and Phil Hugenholtz for stimulating discussions and suggestions.
==== Refs
Pace NR A molecular view of microbial diversity and the biosphere Science 1997 276 734 740 9115194 10.1126/science.276.5313.734
Rappe MS Giovannoni SJ The uncultured microbial majority Annu Rev Microbiol 2003 57 369 394 14527284 10.1146/annurev.micro.57.030502.090759
Keller M Zengler K Tapping into microbial diversity Nat Rev Microbiol 2004 2 141 150 15040261 10.1038/nrmicro819
Handelsman J Metagenomics: application of genomics to uncultured microorganisms Microbiol Mol Biol Rev 2004 68 669 685 15590779 10.1128/MMBR.68.4.669-685.2004
Beja O Aravind L Koonin EV Suzuki MT Hadd A Nguyen LP Jovanovich SB Gates CM Feldman RA Spudich JL Spudich EN DeLong EF Bacterial rhodopsin: evidence for a new type of phototrophy in the sea Science 2000 289 1902 1906 10988064 10.1126/science.289.5486.1902
Hallam SJ Putnam N Preston CM Detter JC Rokhsar D Richardson PM DeLong EF Reverse methanogenesis: testing the hypothesis with environmental genomics Science 2004 305 1457 1462 15353801 10.1126/science.1100025
Tyson GW Chapman J Hugenholtz P Allen EE Ram RJ Richardson PM Solovyev VV Rubin EM Rokhsar DS Banfield JF Community structure and metabolism through reconstruction of microbial genomes from the environment Nature 2004 428 37 43 14961025 10.1038/nature02340
Tringe SG von Mering C Kobayashi A Salamov AA Chen K Chang HW Podar M Short JM Mathur EJ Detter JC Bork P Hugenholtz P Rubin EM Comparative metagenomics of microbial communities
1 Science 2005 308 554 557 15845853 10.1126/science.1107851
Robertson DE Chaplin JA DeSantis G Podar M Madden M Chi E Richardson T Milan A Miller M Weiner DP Wong K McQuaid J Farwell B Preston LA Tan X Snead MA Keller M Mathur E Kretz PL Burk MJ Short JM Exploring nitrilase sequence space for enantioselective catalysis Appl Environ Microbiol 2004 70 2429 2436 15066841 10.1128/AEM.70.4.2429-2436.2004
Venter JC Remington K Heidelberg JF Halpern AL Rusch D Eisen JA Wu D Paulsen I Nelson KE Nelson W Fouts DE Levy S Knap AH Lomas MW Nealson K White O Peterson J Hoffman J Parsons R Baden-Tillson H Pfannkoch C Rogers YH Smith HO Environmental genome shotgun sequencing of the Sargasso Sea Science 2004 304 66 74 15001713 10.1126/science.1093857
Beja O Spudich EN Spudich JL Leclerc M DeLong EF Proteorhodopsin phototrophy in the ocean Nature 2001 411 786 789 11459054 10.1038/35081051
Bielawski JP Dunn KA Sabehi G Beja O Darwinian adaptation of proteorhodopsin to different light intensities in the marine environment Proc Natl Acad Sci U S A 2004 101 14824 14829 15466697 10.1073/pnas.0403999101
Man D Wang W Sabehi G Aravind L Post AF Massana R Spudich EN Spudich JL Beja O Diversification and spectral tuning in marine proteorhodopsins EMBO J 2003 22 1725 1731 12682005 10.1093/emboj/cdg183
Lorenz P Eck J Metagenomics and industrial applications
1 Nat Rev Microbiol 2005 3 510 516 15931168 10.1038/nrmicro1161
Robertson DE Mathur E Swanson RV Marrs BL Short JM The discovery of new biocatalysts from microbial diversity Society for Industrial Microbiology News 1996 46 3 8
Schloss PD Handelsman J Biotechnological prospects from metagenomics Curr Opin Biotechnol 2003 14 303 310 12849784 10.1016/S0958-1669(03)00067-3
Short JM Recombinant approaches for accessing biodiversity Nat Biotechnol 1997 15 1322 1323 9415872 10.1038/nbt1297-1322
Brenner C Catalysis in the nitrilase superfamily Curr Opin Struct Biol 2002 12 775 782 12504683 10.1016/S0959-440X(02)00387-1
O'Reilly C Turner PD The nitrilase family of CN hydrolysing enzymes - a comparative study J Appl Microbiol 2003 95 1161 1174 14632988 10.1046/j.1365-2672.2003.02123.x
Pace HC Brenner C The nitrilase superfamily: classification, structure and function Genome Biol 2001 2 reviews0001.1–0001.9 11380987 10.1186/gb-2001-2-1-reviews0001
Lathe WCIII Snel B Bork P Gene context conservation of a higher order than operons Trends Biochem Sci 2000 25 474 479 11050428 10.1016/S0968-0004(00)01663-7
Rogozin IB Makarova KS Murvai J Czabarka E Wolf YI Tatusov RL Szekely LA Koonin EV Connected gene neighborhoods in prokaryotic genomes Nucleic Acids Res 2002 30 2212 2223 12000841 10.1093/nar/30.10.2212
Lawrence JG Gene organization: selection, selfishness, and serendipity Annu Rev Microbiol 2003 57 419 440 14527286 10.1146/annurev.micro.57.030502.090816
Price MN Huang KH Alm EJ Arkin AP A novel method for accurate operon predictions in all sequenced prokaryotes
3 Nucleic Acids Res 2005 33 880 892 15701760 10.1093/nar/gki232
Gough J Karplus K Hughey R Chothia C Assignment of homology to genome sequences using a library of hidden Markov models that represent all proteins of known structure J Mol Biol 2001 313 903 919 11697912 10.1006/jmbi.2001.5080
Sofia HJ Chen G Hetzler BG Reyes-Spindola JF Miller NE Radical SAM, a novel protein superfamily linking unresolved steps in familiar biosynthetic pathways with radical mechanisms: functional characterization using new analysis and information visualization methods Nucleic Acids Res 2001 29 1097 1106 11222759 10.1093/nar/29.5.1097
Wybenga-Groot LE Draker K Wright GD Berghuis AM Crystal structure of an aminoglycoside 6'-N-acetyltransferase: defining the GCN5-related N-acetyltransferase superfamily fold Structure Fold Des 1999 7 497 507 10378269 10.1016/S0969-2126(99)80066-5
Sulavik MC Gambino LF Miller PF The MarR repressor of the multiple antibiotic resistance (mar) operon in Escherichia coli: prototypic member of a family of bacterial regulatory proteins involved in sensing phenolic compounds Mol Med 1995 1 436 446 8521301
Allard ST Giraud MF Naismith JH Epimerases: structure, function and mechanism Cell Mol Life Sci 2001 58 1650 1665 11706991
Tanner ME Understanding nature's strategies for enzyme-catalyzed racemization and epimerization Acc Chem Res 2002 35 237 246 11955052 10.1021/ar000056y
Fretland AJ Omiecinski CJ Epoxide hydrolases: biochemistry and molecular biology Chem Biol Interact 2000 129 41 59 11154734 10.1016/S0009-2797(00)00197-6
Yang Z Bielawski JP Statistical methods for detecting molecular adaptation Trends Ecol Evol 2000 15 496 503 11114436 10.1016/S0169-5347(00)01994-7
Yang Z Inference of selection from multiple species alignments Curr Opin Genet Dev 2002 12 688 694 12433583 10.1016/S0959-437X(02)00348-9
Bielawski JP Yang Z Maximum likelihood methods for detecting adaptive evolution after gene duplication J Struct Funct Genomics 2003 3 201 212 12836699 10.1023/A:1022642807731
Zhang J Rosenberg HF Nei M Positive Darwinian selection after gene duplication in primate ribonuclease genes Proc Natl Acad Sci U S A 1998 95 3708 3713 9520431 10.1073/pnas.95.7.3708
Muse SV Gaut BS A likelihood approach for comparing synonymous and nonsynonymous nucleotide substitution rates, with application to the chloroplast genome Mol Biol Evol 1994 11 715 724 7968485
Ohno S Evolution by Gene Duplication 1970 Springer
Dykhuizen D Hartl DL Selective neutrality of 6PGD allozymes in E. coli and the effects of genetic background Genetics 1980 96 801 817 7021316
Rodriguez-Trelles F Tarrio R Ayala FJ Convergent neofunctionalization by positive Darwinian selection after ancient recurrent duplications of the xanthine dehydrogenase gene Proc Natl Acad Sci U S A 2003 100 13413 13417 14576276 10.1073/pnas.1835646100
Zhang J Evolution by gene duplication: an update Trends Ecol Evol 2003 18 292 298 10.1016/S0169-5347(03)00033-8
Suzuki Y Nei M Simulation study of the reliability and robustness of the statistical methods for detecting positive selection at single amino acid sites Mol Biol Evol 2002 19 1865 1869 12411595
Suzuki Y Nei M False positive selection identified by ML-based methods: examples from the Sig1 gene of the diatom Thalassiosira weissflogii and the tax gene of a human T-cell lymphotropic virus Mol Biol Evol 2004 21 914 921 15014169 10.1093/molbev/msh098
Wong WS Yang Z Goldman N Nielsen R Accuracy and power of statistical methods for detecting adaptive evolution in protein coding sequences and for identifying positively selected sites Genetics 2004 168 1041 1051 15514074 10.1534/genetics.104.031153
Yang Z Nielsen R Goldman N Pedersen AM Codon-substitution models for heterogeneous selection pressure at amino acid sites Genetics 2000 155 431 449 10790415
Endo T Ikeo K Gojobori T Large-scale search for genes on which positive selection may operate
5 Mol Biol Evol 1996 13 685 690 8676743
Yang Z Nielsen R Codon-substitution models for detecting molecular adaptation at individual sites along specific lineages Mol Biol Evol 2002 19 908 917 12032247
Pace HC Hodawadekar SC Draganescu A Huang J Bieganowski P Pekarsky Y Croce CM Brenner C Crystal structure of the worm NitFhit Rosetta Stone protein reveals a Nit tetramer binding two Fhit dimers Curr Biol 2000 10 907 917 10959838 10.1016/S0960-9822(00)00621-7
Wang WC Hsu WH Chien FT Chen CY Crystal structure and site-directed mutagenesis studies of N-carbamoyl-D-amino-acid amidohydrolase from Agrobacterium radiobacter reveals a homotetramer and insight into a catalytic cleft J Mol Biol 2001 306 251 261 11237598 10.1006/jmbi.2000.4380
Overbeek R Fonstein M D'Souza M Pusch GD Maltsev N The use of gene clusters to infer functional coupling Proc Natl Acad Sci U S A 1999 96 2896 2901 10077608 10.1073/pnas.96.6.2896
Itoh T Takemoto K Mori H Gojobori T Evolutionary instability of operon structures disclosed by sequence comparisons of complete microbial genomes Mol Biol Evol 1999 16 332 346 10331260
Tan K Moreno-Hagelsieb G Collado-Vides J Stormo GD A comparative genomics approach to prediction of new members of regulons Genome Res 2001 11 566 584 11282972 10.1101/gr.149301
Osterman A Overbeek R Missing genes in metabolic pathways: a comparative genomics approach Curr Opin Chem Biol 2003 7 238 251 12714058 10.1016/S1367-5931(03)00027-9
Tamames J Evolution of gene order conservation in prokaryotes Genome Biol 2001 2 Research0020 11423009 10.1186/gb-2001-2-6-research0020
Makarova KS Koonin EV Comparative genomics of Archaea: how much have we learned in six years, and what's next? Genome Biol 2003 4 115 12914651 10.1186/gb-2003-4-8-115
Babbitt PC Definitions of enzyme function for the structural genomics era Curr Opin Chem Biol 2003 7 230 237 12714057 10.1016/S1367-5931(03)00028-0
DeLong EF Microbial population genomics and ecology: the road ahead Environ Microbiol 2004 6 875 878 15305912 10.1111/j.1462-2920.2004.00668.x
Rodriguez-Valera F Environmental genomics, the big picture FEMS Microbiol Lett 2004 231 153 158 15027428 10.1016/S0378-1097(04)00006-0
The Institute for Genome Research 2005
DOE Joint Genome Institute 2005
Hall T BioEdit 2005
Felsenstein J PHYLIP -- Phylogeny Inference Package (Version 3.2) Cladistics 1989 5 164 166
Schmidt HA Strimmer K Vingron M von Haeseler A TREE-PUZZLE: maximum likelihood phylogenetic analysis using quartets and parallel computing Bioinformatics 2002 18 502 504 11934758 10.1093/bioinformatics/18.3.502
Swofford DL PAUP*: phylogenetic analysis using parsimony (*and other methods). 1998 Sinauer Associates, Sunderland, Mass.
Posada D Crandall KA MODELTEST: testing the model of DNA substitution Bioinformatics 1998 14 817 818 9918953 10.1093/bioinformatics/14.9.817
Yang Z PAML: a program package for phylogenetic analysis by maximum likelihood Comput Appl Biosci 1997 13 555 556 9367129
Yang Z Likelihood ratio tests for detecting positive selection and application to primate lysozyme evolution Mol Biol Evol 1998 15 568 573 9580986
Xiang SZ Jackal A Protein Structure Modeling Package 2005
DeLano WL The PyMOL Molecular Graphics System 2002 DeLano Scientific, San Carlos, CA, USA.
|
16083508
|
PMC1199592
|
CC BY
|
2021-01-04 16:29:16
|
no
|
BMC Evol Biol. 2005 Aug 6; 5:42
|
utf-8
|
BMC Evol Biol
| 2,005 |
10.1186/1471-2148-5-42
|
oa_comm
|
==== Front
BMC GenetBMC Genetics1471-2156BioMed Central London 1471-2156-6-421609552810.1186/1471-2156-6-42Methodology ArticleDevelopment of novel heminested PCR assays based on mitochondrial 16s rRNA gene for identification of seven pecora species Guha Saurav [email protected] VK [email protected] National DNA Analysis Centre, Central Forensic Science Laboratory, 30 Gorachand Road, Kolkata-700014, India2005 11 8 2005 6 42 42 15 12 2004 11 8 2005 Copyright © 2005 Guha and Kashyap; licensee BioMed Central Ltd.2005Guha and Kashyap; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Characterization of molecular markers and the development of better assays for precise and rapid detection of wildlife species are always in demand. This study describes a set of seven novel heminested PCR assays using specific primers designed based on species-specific polymorphism at the mitochondrial 16S rRNA gene for identification of Blackbuck, Goral, Nilgai, Hog deer, Chital, Sambar and Thamin deer.
Results
The designed heminested PCR assays are two consecutive amplifications of the mitochondrial 16S rRNA gene. In the first stage, ~550 bp region of the 16S rRNA gene was amplified by PCR using template DNA and universal primers. In the second stage, a species-specific internal region of the 16S rRNA gene was amplified by PCR using the amplicon of the first PCR along with one universal primer and another species-specific primer as the reverse or forward primer. The amplicon generated after two consecutive amplifications was highly unique to target species. These assays were successfully validated for sensitivity, specificity, and ruggedness under a wide range of conditions.
Conclusion
The validation experiments confirm that the designed heminested PCR assays for identification of the seven species are highly specific, sensitive, reliable and provide a reproducible method allowing analysis of low copy number DNA recovered from decomposed or highly processed tissues. The assays for identification of other species could be devised by extrapolating the principle of designed heminested PCR.
==== Body
Background
Characterization of species-specific molecular markers and designing of species-specific assays for identification of wildlife species are essential to prevent illegal trade of parts and products for better conservation and management of endangered species. Illegal trade of skin, bone, horn, tail-hair, meat, antlers of Pecora family prevalent in India and have considerable trading value worldwide [1]. Laboratories often receive bio-specimens suspected to be of bovid and cervid origin. The protected members of Pecora family extensively hunted in India, e.g. Blackbuck, Goral, Nilgai, Chital, Thamin, Sambar, Hog deer, and Musk deer. Antilope cervicapra (Blackbuck), Naemorhedus goral (Goral) and Boselaphus tragocamelus (Nilgai) constitute three main Bovidae species endemic to India and adjoining countries. The Cervidae family also comprises of a number of species endemic to the Indian subcontinent e.g. Cervus axis (Chital), Cervus unicolor (Sambar), Cervus eldi (Thamin deer) and Axis porcinus (Hog deer). The density and distribution of these species have sharply declined due to loss of habitat through logging, livestock grazing and shift agriculture. All these species are prized for their meat, skin, horn, liver etc. and therefore extensively hunted [2]. All of these species are listed in IUCN (International Union for Conservation of Nature) and/or CITES (Convention on International Trade in Endangered Species of Wild Fauna and Flora) as vulnerable or threatened. In India they are listed in schedule I or III under Wildlife Protection Act, India (1972).
The conventional methods based upon the structural, electrophoretic and immunological characteristics of the species are often used in identification of skin, bone, horn, tail-hair, meat, antlers etc of poached animals [3,4]. However these methods are of limited use in species identification because of low stability and specificity of the markers. The biological materials forwarded for species identification are often of low quality. The analytical methods should be highly specific, sensitive, robust, reproducible and reliable. Presently DNA based techniques are extensively used in species identification [5-11]. Most DNA methods reported involve sequence analysis of mitochondrial DNA (mtDNA) [12-14]. The mitochondrial genome of vertebrates has been extensively used over nuclear DNA for resolving phylogenetic relationships at different evolutionary depth due to its unique properties, including presence of strictly orthologous genes, lack of recombination, and unique substitution rates [15]. These unique features along with high copy number of mitochondrial DNA per cell compared to nuclear DNA makes it a more powerful tool than nuclear DNA for identification of unknown biological materials [16-18]. Most of the designed species-specific identification methods are based upon mitochondrial Cytochrome b (Cytb) gene [19,20] due to the availability of well-documented universal primers for the gene [21]. The other genes also harbor species-specific sequences, are un-translated gene regions, e.g. rRNA genes, where base deletion and insertions can be tolerated without altering protein sequence, structure and function. Mitochondrial rRNA genes are also widely studied for evolutionary studies [22-24] and some recent studies also have proven the usefulness of unique polymorphisms present in mitochondrial rRNA genes in species identification [25-27]. However no report is available for mammalian species identification using species-specific polymorphism of mitochondrail rRNA genes.
DNA based methods employed in species identification comprised of species-specific PCR, RFLP-PCR, RAPD-PCR, SSCP, OLA etc are highly specific and sensitive but do not have inbuilt mechanism to assess DNA quality, quantity and presence of PCR inhibitor due to lack of a control step [7-10]. This control step ensures the quality of DNA amenable to amplify and detect the presence of any inhibitor in PCR reaction. The heminested PCR assay could integrate a independent control step with specific primer in two consecutive PCR reactions with improvement of sensitivity without impairing specificity [28].
In this study, we report a set of seven novel heminested PCR assays based on the amplification of a species-specific internal sequence region of mitochondrial 16s rRNA gene for identification of the species Blackbuck, Goral, Nilgai, Chital, Thamin, Sambar and Hog deer with their respective specific primers. To our knowledge, this is the first attempt to develop heminested PCR assay, exploring species unique polymorphism present in mitochondrial 16s rRNA gene for species identification.
Results and discussion
The DNA based species identification methods often encounter low quantity and/or poor quality DNA template, and presence of PCR inhibitors. To examine these common yet serious problems, a independent control of DNA quality and quantity needs to be integrated in the identification assays. In this study we have designed and validated seven novel heminested mitochondrial 16s rRNA gene-targeted PCR assays for identification of seven species using designed species-specific primers listed in table 1. To design the species-specific primers we have screened the species unique sites or motifs in mitochondrial cytochrome b and 16s rRNA gene of thirty-seven species. Though level of polymorphism was high in cytochrome b gene compare to 16s rRNA gene, but species-specific motifs were higher in 16s rRNA gene among studied species [unpublished data. The species-specific unique sites at 16s rRNA gene were used to design the primers for identification of the studied species.
Table 1 Designed species-specific and selected universal oligonucleotide primers used in this study along with amplicon size.
Species Primer
name Primer
direction Primer
size (bp) Primer
sequence Species – Specific
amplicon size
All 16s Fwd Forward 20 5'-CGCCTGTTTATCAAAAACAT-3'
All 16s Rev Reverse 20 5'- CTCCGGTTTGAACTCAGATC-3 ~550 bp
Antilope. Cervicapra BB16 Reverse 21 5'-ACCTAGTTATTCGCTATCAAG -3' 410 bp
Naemorhedus. goral GL16 Reverse 29 5'-GGCAACTAGTTCAAAAAACCTAGTTATTT -3' 430 bp
Boselaphus. tragocamelus NG16 Reverse 24 5'-AGTTATTCGCTATCTCACTAAACC -3' 408 bp
Cervus. eldi TH16 Reverse 22 5'-CCTAGTTATTTGCTAACACACC -3' 400 bp
Cervus. axis CH16 Reverse 25 5'-TATCCCAGAAGGACAGAATAATTAC -3' 240 bp
Cervus. unicolor SM16 Forward 24 5'-GCTTTAACTACTTAGCCCAAAGAG -3' 350 bp
Axis. porcinus HD16 Reverse 27 5'-CTCACAACAACAAAGGAATCGCTACTC -3' 300 bp
In designed heminested PCR assays ~550 bp fragment were generated with universal outer primers at first cycle in all the studied species (Fig 2 and table 2). These results confirmed the quality of DNA amenable to amplification and nonexistence of any PCR inhibitor within the samples. It also enhanced the number of templates for next cycle of the assay. In the second PCR, using species-specific inner primers paired with another universal primer, all the individuals of a species generated species-specific amplicons (Fig 2 and table 2). The products of both amplifications for each individual were loaded together in to 3 % agarose gel for easy visualization. These species- specific amplicons are not present to any other non-target species having the 16s rRNA amplified region included in the study for testing the specificity of the method. The schematic diagram of species – specific amplicons with species – specific primers based on mitochondrial 16s rRNA gene are presented in fig 1. Though the diagnostic value of Blackbuck, Goral, Nilgai and Thamin deer specific amplicons are limited in 3 % agarose gel due to their almost equal fragment length but they are highly specific with specific primers.
Figure 1 Schematic diagram of the mitochondrial 16s rRNA gene and species-specific amplicons with species – specific primers.
Figure 2 Species-specific profiles of designed heminested PCR assay. M indicates 100 bp ladders. Higher molecular weight amplicons were amplified for all the species with universal 16s rRNA primers. In 2nd PCR only target species have generated species-species amplicons with specific primers. Lane 14 of both the gel contains negative controls.
Table 2 Results from species-specific heminested PCR assay with target and non target species for designed primers
Species Common Name DNA source 1st PCR universal primer 16s rRNA gene 2nd PCR BB 16 2nd PCR GL 16 2nd PCR NG 16 2nd PCR TH 16 2nd PCR CH 16 2nd PCR SM 16 2nd PCR HD 16
Antilope cervicapra a Blackbuck Tissue + + - - - - - -
Antilope cervicapra b Blackbuck Tissue + + - - - - - -
Antilope cervicapra c Blackbuck Blood stain + + - - - - - -
Boselaphus tragocamelus a Nilgai Tissue + - - + - - - -
Boselaphus tragocamelus b Nilgai Tissue + - - + - - - -
Boselaphus tragocamelus c Nilgai Tissue + - - + - - - -
Naemorhedus goral a Goral Blood stain + - + - - - - -
Naemorhedus goral b Goral Tissue + - + - - - - -
Naemorhedus goral c Goral Tissue + - + - - - - -
Axis porcinus a Hog deer Tissue + - - - - - - +
Axis porcinus b Hog deer Tissue + - - - - - - +
Cervus axis a Chital Blood stain + - - - - + - -
Cervus axis b Chital Tissue + - - - - + - -
Cervus axis c Chital Tissue + - - - - + - -
Cervus axis d Chital Tissue + - - - - + - -
Cervus axis e Chital Tissue + - - - - + - -
Cervus eldi a Thamin Tissue + - - - + - - -
Cervus eldi b Thamin Tissue + - - - + - - -
Cervus unicolor a Sambar Tissue + - - - - - + -
Cervus unicolor b Sambar Tissue + - - - - - + -
Cervus duvaucelii Barasingha Tissue + - - - - - - -
Ovis aries Sheep Tissue + - - - - - - -
Bos tarus Cow Tissue + - - - - - - -
Homo sapiens Human Blood + - - - - - - -
Lepidochelys olevacea Turtle Tissue + - - - - - - -
Hemidactylus fluviviridis Lizard Tissue + - - - - - - -
+ positive amplification, - negative amplification
The designed heminested PCR assays with species-specific primers successfully produced specific amplicon for the target species with consecutive dilution of DNA templates ranging from 5 ng to 5 pg (fig 3 and table 3). DNA extracted from various tissues generated identical result for all individual of a species using heminested PCR assay. To ensure the fidelity of the designed heminested PCR, a comparative experiment have been performed between heminested and normal PCR using DNA template isolated from a tissue of chital species. The normal PCR with 61°C annealing temperature using chital specific primer produced two nonspecific bands a long with the target band. On the otherhand, in heminested PCR assay using same annealing temperature only one single band of 240 bp size was produced at the second PCR reaction (fig 5). This experiment clearly reveals the higher fidelity of the designedheminested PCR compared to normal PCR used instudied species identification. The highly degraded DNAsample of a chital individual was also subjected for mitochondrial (16s rRNA) and nuclear (3' UTR of SON gene) gene amplification. The universal primers of 3' UTR of SON gene were not able to amplify the expected fragment of ~175 bp size. While, the universal primers of mitochondrial 16s rRNA gene successfully amplified a fragment of ~550 bp size. This result clearly indicates the usefulness of mitochondrial DNA in the designed assay over the nuclear DNA.
Figure 3 Species-specific profiles with 5 pg and 5 ng template DNA concentration of designed heminested PCR assay. M indicates 100 bp ladders. All the species have shown positive result at 1st and 2nd PCR.
Table 3 Results of validation study of species-specific heminested PCR assays
Validation Parameters No. of samples analysed No. of specific primers analysed No. of samples successfully typed
DNA concentration 5pg 20 7 20
50 pg 20 7 20
500 pg 20 7 20
5 ng 20 7 20
Mixture analysis $ Blackbuck + Nilgai + Goral + Hog deer + Chital + Thamin + Sambar 1 (mixture) 1a 1 (Blackbuck)
Blackbuck + Nilgai + Goral + Hog deer + Chital + Thamin + Sambar 1 (mixture) 1b 1 (Nilgai)
Blackbuck + Nilgai + Goral + Hog deer + Chital + Thamin + Sambar 1 (mixture) 1c 1 (Goral)
Blackbuck + Nilgai + Goral + Hog deer + Chital + Thamin + Sambar 1 (mixture) 1d 1 (Hog deer)
Blackbuck + Nilgai + Goral + Hog deer + Chital + Thamin + Sambar 1 (mixture) 1e 1 (Chital)
Blackbuck + Nilgai + Goral + Hog deer + Chital + Thamin + Sambar 1 (mixture) 1f 1 (Thamin)
Blackbuck + Nilgai + Goral + Hog deer + Chital + Thamin + Sambar 1 (mixture) 1g 1 (Sambar)
Chemical reagents Soap 20 7 20
0.1 N NaOH 20 7 20
5% Acetic acid 20 7 20
1 M NaCL 20 7 20
Gasoline 20 7 20
Exposure to Heat 37°C, 1 week 7* 7 7
37°C, 2 week 7* 7 7
37°C, 3 week 7* 7 7
UV Irradiation (302 nm, 15 cm distance) 1 h 20 7 20
8 h 20 7 20
24 h 20 7 18
* One sample of each species were included for Exposure to Heat validation
$ Each mixture comprised 50 pg of DNA of target species and 500 pg of DNA of each non-target species a – Blackbuck specific primer, b – Nilgai specific primer, c- Goral specific primer, d – Hog deer specific primer, e – Chital specific primer, f – Thamin specific primer, g – Sambar specific primer.
The designed heminested assays also successfully typed the target species from mixture of DNA originating from different species (table 3). The successful amplification of all the tissue samples treated with different chemical reagents confirms the strength of the assay (table 3). The tissue samples were also incubated at 37°C for different time period assuming the damp condition influencing the typing result. But all the samples were amplified successfully after this heat treatment (table 3). The results of analysis tissue samples subjected to UV irradiation indicate a minor influence of this factor on designed assay (table 3). These results confirm substantial stability and robustness of the assays against considerable influence of various chemical and physical factors.
The possibility of carry-over contamination is a common problem in nested or heminested PCR [26]. However these assays have been designed for highly species-specific primers that will prevent any non-specific amplification. Nevertheless, researchers need to setup separate pre- and post-PCR environments and take other precautions to avoid the risk of carry-over contamination.
Conclusion
In conclusion, it is found that the designed heminested PCR assays based on unique species-specific polymorphism at mitochondrial 16s rRNA gene for identification of seven species are highly specific, simple and sensitive technique. Validation studies i.e. sensitivity of the designed assays compare to normal PCR, different DNA concentrations, species cross reactivity, specificity, and stability under various physical and chemical environment clearly reveal the applicability of the assays to less-than-optimal and highly degraded samples. Two consecutive PCRs and agarose gel visualization make the assays rapid as well as easy to interpret. The designed assays are more efficient and cost effective technology compared to other species identification tools, which do not have any inbuilt independent control of DNA quality and quantity. Because of highly species-specific primers, the designed heminested PCR assays provide reliable evidence for wildlife enforcement. These heminested PCR assays based on species-specific polymorphism in mitochondrial 16s rRNA gene can also be adapted for the identification of a wide range of wildlife species.
Methods
Samples and DNA isolation
DNA was extracted from various authenticated tissue samples of the unrelated individuals (N) of Blackbuck (N = 3), Goral (N = 3), Nilgai (N = 3), Thamin (N = 2), Chital (N = 5), Sambar (N = 2) and Hog deer (N = 2) by using standard phenol/chloroform procedure [29]. These samples of known animals species were provided by WII, Dehradun, India. DNA was concentrated and cleaned using Microcon 100 (Millipore) concentrators. Negative controls were included in every extraction. DNA was quantified using spectrophotometric method.
Designing of primers
In heminested PCR the 16s rRNA gene based universal primers; 16s rRNA fwd and 16s rRNA rev [30] were chosen as outer primer and would amplify approximately ~550 bp amplicon. The species-specific internal primers were designed for blackbuck (BB16), goral (GL16), Nilgai (NG16), chital (CH16), thamin (TH16), sambar (SM16) and hog deer (HD16) (table 1). These specific primers were paired with the universal forward primer of first PCR, except for sambar where universal reverse primer used in the second PCR assay. These specific primers in conjunction with universal primers in heminested PCR generated specific amplicons; 410 bp (blackbuck), 430 bp (goral), 408 bp (Nilgai), 400 bp (Thamin), 240 bp (Chital), 350 bp (Sambar) and 300 bp (Hog deer) (table 1 and Fig 1).
Primer specificity
The designed species-specific primers were subjected to BLAST search at NCBI GenBank database to rule out any false similarity with other species. These primers were also tested for amplification specificity using DNA extracted from a panel of non-target species belonging to different classes.
Heminested PCR
DNA samples of individuals of studied species were subjected to heminested PCR assay with outer universal mitochondrial 16s rRNA gene primers at first PCR and with specific primers paired with one universal primer as reverse/ forward at second PCR. The first PCR reaction were carried out in a 25 μl reaction volume with 2 μl of DNA (10 ng – 30 ng/ μl) from tissue samples, 3 μl 10× PCR buffer, 2.5 μl of MgCl2 (15 mM), 2 μl of dNTP (2.5 mM each dNTP), 2 μl of both primers (10 pm/μl), 0.5 μl of Taq polymerase (5 U/μl), under the standardise condition; 94°C for 2 minute pre PCR denaturation, followed by 94°C for 30 seconds, 47°C for 30 seconds, 72°C for 45 seconds, for 30 cycles and a final extension at 72°C for 5 minutes. The PCR products of the first amplification with universal primers were diluted to ~1/20th and 1 μl subjected to second PCR with universal 16s rRNA forward primer and Blackbuck, Goral, Nilgai, Thamin, Chital, and Hog deer specific primers as reverse at six different sets and universal 16s rRNA reverse primer and Sambar specific primer in a single set. The second PCR cycles were carried out almost in identical conditions to the first PCR cycle except the species – specific primer annealing temperature and final MgCl2 concentration. The annealing temperature and final MgCl2 concentration for Blackbuck and Goral specific primers is 56°C and 0.9 mM, Nilgai specific primer is 54°C and 1.5 mM, Chital and Hogdeer specific primer is 61°C and 1.0 mM, Thamin deer is 58°C and 1.0 mM and Sambar is 62°C and 1.0 mM. The amplicons of both the PCR were loaded together for an individual of a species in a single lane and visualized in 3% agarose gel with 100 bp ladder, a negative control and with other non-target species performed the same assay (Fig 2). The results of all the heminested PCR assays are presented at table 2.
Validation study
Sensitivity of these assays were assessed by analysis of consecutive dilution of total DNA extracted from different tissues (muscle, liver, heart, kidney, skin) and blood stains of unrelated individuals of seven species. The DNA dilutions were prepared in a range of 5 pg to 5 ng and subjected to PCR assay with designed specific primers using stringent conditions described above (table 3, fig 3).
Fidelity of the designed heminested PCR assays compare to normal PCR assay was also evaluated by using the DNA template isolated from tissue sample of chital species along with chital specific primer.
To check the suitability of using mitochondrial DNA over nuclear DNA we have extracted DNA from a highly degraded and mutilated chital tissue sample. No high molecular DNA was found and the DNA aliquot was subjected to PCR with mitochondrial 16s rRNA gene universal primers and 3' UTR of the SON gene [31] universal primers (SON Fwd primer 5' – ACATAGCATATGAATACC – 3' and SON Rev primer 5' – GTCTATCTAGGTGTAGCTGA – 3').
Mixture analysis was also performed with all the designed species-specific primers to evaluate the sensitivity of detection from mixture of DNA samples of more than one origin. Seven types of DNA mixtures were prepared and subjected to heminested PCR assays consecutively using seven designed primers (table 3). Each mixture was prepared by aliquoting 50 pg of DNA of target species and 500 pg of DNA of non-target species.
Exposure of samples to different factors
Biospecimens of known species were exposed to different chemical reagents (gasoline, soap, 0.1 N NaOH, 5% Acetic acid, 1 M NaCl) and analyzed after 5 days of storage in ambient temperature. Small pieces of tissue samples were also placed in dishes floating in a water bath settled at 37°C for periods of 1, 2 and 3 week. After the exposure samples were kept frozen in -80°C until extraction. Samples were also treated by UV irradiation (302 nm) from a distance of 15 cm for 1 h, 8 h and 24 h (table 3).
Authors' contributions
SG carried out the experiments of standardisation and validation of heminested PCR assays and significantly contributed in preparation of manuscript.
VKK conceived, designing of experiments, and contributed in preparation of manuscript.
Acknowledgements
We would like to thank to Dr. S.P. Goyal, Wildlife Institute of India for providing biological samples and to Dr. R. Trivedi, CFSL Kolkata for providing laboratory facilities. Special thanks direct to Dr. George Weber for valuable comments and suggestions on manuscript. This research was supported by a grant under IX th plan to Central Forensic Science Laboratory, Kolkata, Ministry of Home Affairs, Govt. of India. The first author is also thankful to DFS for fellowship.
==== Refs
Mukherjee SK Some thoughts on wildlife trade Cheetal 1995 33 30 33
Nowak MR Walker's World of Mammals 1995 Johns Hopkins University Press
Kim H Shelef LA Characterization and identification of raw beef, pork, chicken and turkey meats by electrophoretic patterns of their sarcoplasmis proteins J Food Sci 1986 51 731 741
Hsieh YH Sheu SC Bridgman RC Development of a monoclonal antibody specific to cooked mammalian meats J Food Prot 1998 61 476 487 9709213
Blackett RS Keim P Big Game Species Identification by Deoxyribonucleic Acid (DNA) Probes J Forensic Sci 1992 37 590 596 1500901
Grossman PD Bloch W Brinson E Chang CC Eggerding FA Fung S Iovannisci DA Woo S Winn-Deen ES High-density multiplex detection of nucleic acid sequences: oligonucleotide ligation assay and sequence-coded separation Nucleic Acids Res 1994 22 4527 4534 7526344
Mansfield ES Robertson JM Vainer M Isenberg AR Frazier RR Ferguson K Chow S Harris DW Barker DL Gill PD Budowle B McCord BR Analysis of multiplexed short tandem repeat (STR) systems using capillary array electrophoresis Electrophoresis 1998 19 101 107 9511870 10.1002/elps.1150190118
Rehbein H Kress G Schmidt T Application of PCR-SSCP to species identification of fishery products J Sci Food Agric 1997 74 35 41 10.1002/(SICI)1097-0010(199705)74:1<35::AID-JSFA765>3.0.CO;2-2
Johnson ED Kotowski TM Chimiluminescent detection of RFLP patterns in forensic DNA analysis J Forensic Sci 1996 41 569 578 8754566
Cipriano F Palumbi SR DNA forensics and population structure in North Pacific minke whales. Report to the Scientific Committee, International Whaling Commission 1997 (SC/49/NP17)
Palumbi SR Martin A Romano S McMillan WO Stice L Grabowski G The Simple Fool's Guide to PCR 1996 Kewalo Marine Laboratory and Univ. of Hawaii
Parson W Pegoraro K Niederstätter H Föger M Steinlechner M Species identification by means of the cytochrome b gene Int J Legal Med 2000 114 23 28 11197623 10.1007/s004140000134
Zehner R Zimmermann S Mebs D RFLP and sequence analysis of the cytochrome b gene of selected animals and man: methodology and forensic application Int J Legal Med 1998 111 323 327 9826093 10.1007/s004140050180
Cipriano F Palumbi SR Rapid Genotyping Techniques for Identification of Species and Stock Identity in Fresh, Frozen, Cooked and Canned Whale Products 1999 (SC/51/09)
Allard WM Miyamoto MM Jarecki L Kraus F Tennant RM DNA systematics and evolution of the artiodactyls family Bovidae Proc Natl Acad Sci 1992 89 3972 3976 1570322
Wilson MR DiZinno JA Polanske D Replogle J Budowle B Validation of mitochondrial DNA sequencing for forensic casework analysis Int J Legal Med 1995 108 68 74 8547161 10.1007/BF01369907
Branicki W Kupiec T Pawlowski R Validation of Cytochrome b sequence analysis as method of Species identification J Forensic Sci 2003 48 83 87 12570204
Cheng C Hsieh Y Noguchi T Arakawa O Hwang D Effect of processing on sequence of Cytochrome B gene and its restriction site in the meat of puffer Takifugu rubripes J Food Drug Anal 2001 9 232 237
Wan HQ Fang GS Application of species-specific polymerase chain reaction in the forensic identification of tiger species Forensic Sci Int 2003 131 75 78 12505474 10.1016/S0379-0738(02)00398-5
Hsieh HM Chiang HL Tsai LC Lai SY Huang NE Linacre A Lee JC Cytochrome b gene for species identification of the conservation animals Forensic Sci Int 2001 122 7 18 11587860 10.1016/S0379-0738(01)00403-0
Kocher TD Thomas WK Meyer A Edwards SV Paabo S Villablanca FX Wilson AC Dynamics of mitochondrial DNA evolution in animals: Amplification and sequencing with conserved primers Proc Natl Acad Sci 1989 86 6196 6200 2762322
Allard WM Honeycutt LR Nucleotide sequence variation in the mitochondrial 12s rRNA gene and the phylogeny of African mole-rats (Rodentia: Bathyergidae) Mol Biol Evol 1992 9 27 40 1552839
Whitfield JB Cameron AS Hierarchical analysis of variation in the mitochondrial 16S rRNA gene among Hymenoptera Mol Biol Evol 1998 15 1728 1743 9866207
Arctander P Johansen C Vreto M Phylogeography of three closely related African Bovids Mol Biol Evol 1999 16 1724 1739 10605114
Chapela MJ Sotelo CG Mata PC Perez-martin RI Rehbein H Hold LG Quinteiro J Rey-mendez M Rosa C Santos AT Identification of Cephalopod Species (Ommastrephidae and Loliginidae) in Seafood Products by Forensically Informative Nucleotide Sequencing (FINS) J Food Sci 2002 67 1672 1676
Carrera E Garcia T Cespedes A Gonzalez I Fernandez A Hernandez PE Martin R Salmon and trout analysis by PCR-RFLP for identity authentication J Food Sci 1999 64 410 413
Vences M Thomas M Meijden AVD Chiari Y Vieites DR Comparative performance of the 16S rRNA gene in DNA barcoding of amphibians Frontiers in Zoology 2005 2 5 15771783 10.1186/1742-9994-2-5
Arnheim N Erlich H Polymerase chain reaction strategy Annu Rev Biochem 1992 61 131 156 1379788 10.1146/annurev.bi.61.070192.001023
Sambrook J Fritsch EF Maniatis T Molecular Cloning: a Laboratory Manual 1989 Cold Spring Harbor Laboratory press. Cold Spring Harbor. NY
Palumbi SR Martin A Romano S McMillan WO Stice L Grabowski G The Simple Fool's Guide to PCR, Version 20 1991 Privately published, Univ. Hawaii
Soteriou B Fisher RA Khan IM Kessling AM Archard LC Buluwela L Conserved gene sequences for species identification: PCR analysis of the 3' UTR of the SON gene distinguishes human and other mammalian DNAs Forensic Sci Int 1995 73 171 181 7657183 10.1016/0379-0738(95)01748-8
|
16095528
|
PMC1199593
|
CC BY
|
2021-01-04 16:30:39
|
no
|
BMC Genet. 2005 Aug 11; 6:42
|
utf-8
|
BMC Genet
| 2,005 |
10.1186/1471-2156-6-42
|
oa_comm
|
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-1001604278810.1186/1471-2164-6-100Research ArticlePrediction of the general transcription factors associated with RNA polymerase II in Plasmodium falciparum: conserved features and differences relative to other eukaryotes Callebaut Isabelle [email protected] Karine [email protected] Edwige [email protected] Jean-Paul [email protected] Stanislas [email protected] Centre National de la Recherche Scientifique CNRS UMR7590, Universités Paris 6 et Paris 7, Département de Biologie Structurale, IMPMC, case 115, 4 place Jussieu, 75252 Paris Cedex 05, France2 Centre National de la Recherche Scientifique CNRS UMR 8576, Université des Sciences et Technologies de Lille, Equipe de Parasitologie Moléculaire, Laboratoire de Chimie Biologique, UGSF, Bâtiment C9, 59655 Villeneuve d'Ascq, France2005 23 7 2005 6 100 100 12 3 2005 23 7 2005 Copyright © 2005 Callebaut et al; licensee BioMed Central Ltd.2005Callebaut et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
To date, only a few transcription factors have been identified in the genome of the parasite Plasmodium falciparum, the causative agent of malaria. Moreover, no detailed molecular analysis of its basal transcription machinery, which is otherwise well-conserved in the crown group of eukaryotes, has yet been reported. In this study, we have used a combination of sensitive sequence analysis methods to predict the existence of several parasite encoded general transcription factors associated with RNA polymerase II.
Results
Several orthologs of general transcription factors associated with RNA polymerase II can be predicted among the hypothetical proteins of the P. falciparum genome using the two-dimensional Hydrophobic Cluster Analysis (HCA) together with profile-based search methods (PSI-BLAST). These predicted orthologous genes encoding putative transcription factors include the large subunit of TFIIA and two candidates for its small subunit, the TFIIE β-subunit, which would associate with the previously known TFIIE α-subunit, the TFIIF β-subunit, as well as the p62/TFB1 subunit of the TFIIH core. Within TFIID, the putative orthologs of TAF1, TAF2, TAF7 and TAF10 were also predicted. However, no candidates for TAFs with classical histone fold domain (HFD) were found, suggesting an unusual architecture of TFIID complex of RNA polymerase II in the parasite.
Conclusion
Taken together, these results suggest that more general transcription factors may be present in the P. falciparum proteome than initially thought. The prediction of these orthologous general transcription factors opens the way for further studies dealing with transcriptional regulation in P. falciparum. These alternative and sensitive sequence analysis methods can help to identify candidates for other transcriptional regulatory factors in P. falciparum. They will also facilitate the prediction of biological functions for several orphan proteins from other apicomplexan parasites such as Toxoplasma gondii, Cryptosporidium parvum and Eimeria.
==== Body
Background
Each year 300–500 million people suffer from malaria while 1.5 to 2 million, mostly children, die as a result of the infection (Global Health Council, 2003). The lethal form of human malaria is caused by the infection with the obligate intracellular protozoan parasite Plasmodium falciparum, which displays a developmental life cycle alternating between a vertebrate and an invertebrate host. Infection by the sporozoite form of the parasite occurs after the female Anopheles mosquito's bite. The parasite then enters hepatocytes and multiplies by an asexual division process named schizogony. The resulting merozoites then invade erythrocytes and the parasite goes through a series of morphological changes upon massive rounds of asexual division (ring, trophozoite, schizonte and merozoite). The intermittent fevers, characteristic of malaria infection, are attributed to cycles of erythrocyte invasion, asexual reproduction by schizogony, and release of asexual parasites (merozoites) after rupture of infected red blood cells. For completion of the host-vector cycle, some intra-erythrocytic asexual forms do not undergo schizogony but transform into sexually dimorphic male and female gametocytes upon differentiation. Gametocytes are taken into the mosquito's midgut during a blood meal and complete their sexual development to gametes which will fuse to form a motile zygote named the ookinete. The ookinete grows into an oocyst, dividing into numerous sporozoites that will invade the salivary glands of the mosquito ready for a new cycle of infection [1].
During the complex life cycle of P. falciparum which takes place in both a vertebrate and an invertebrate host, the intracellular development of the different asexual and sexual stages proceeds through a dynamic and multistep process for which the parasite has evolved complex molecular strategies. Several pioneering studies have previously demonstrated that transcriptional regulations are involved in the control of gene expression in the various P. falciparum life cycle forms [2-6]. The recent completion of the full genome sequence of P. falciparum has been useful in studying the global and complex gene expression patterns using microarrays and proteomic approaches. Indeed, these studies suggested that there is a coordinated program of gene expression during the intra-erythrocytic development of the parasite. Microarray data have revealed a sequential expression of transcripts in which messenger RNAs involved in protein synthesis peak at first, followed by metabolism-related genes, then adhesion/invasion genes, and lastly protein kinases [7-10]. Global proteome analysis of sporozoites, merozoites, trophozoites, and gametocytes using tandem mass spectrometry analysis have been used to show that many co-expressed proteins are encoded by genes that are clustered on certain chromosomes [11,12]. These recent studies on gene expression also show that transcription of multiple genes may be achieved by a single developmental induction event resulting in a cascade of gene expressions. This further suggests that only a few specific transcription factors may be required [10]. Nevertheless, it has been established that the gene structure of P. falciparum is similar to that of other eukaryotes [13,14], with the common features including the monocistronically transcribed genes, the presence of 5' and 3' untranslated regions, introns, promoter regions and probably the myriad of transcription factors that are involved in eukaryotic gene expression in general.
Transcription in eukaryotic structural genes requires the assembly of RNA polymerase II (RNAP II) and the general transcription factors (GTFs) on the promoter to form a pre-initiation complex. These basic factors include RNA polymerase II itself and at least six GTFs: TFIIA, TFIIB, TFIID, TFIIE, TFIIF and TFIIH, most of which are themselves multiprotein complexes [15,16]. While the RNA polymerase I, II, III and TATA-binding protein [17-21] have been described in P. falciparum, the elucidation of the mechanisms involved in transcriptional regulation in the parasite is still challenging. For example, the identification of orthologous proteins including the general transcription factors (GTFs) involved in the RNAP II transcription machinery remains elusive. Therefore, the composition and nature of the highly conserved general transcription factors associated with the RNAP II are presently unknown in P. falciparum. In contrast to most eukaryotic genomes, the extensive analysis of the P. falciparum genome has only revealed a few general transcription factors, like TBP and TFIIB [22]. More recently, Coulson et al. [23] have utilized profile-Hidden Markov Models (HMMs) of transcriptional regulators and found a relatively low number of malarial transcription-associated proteins (TAPs) including the general transcription factors associated with RNAP II. Only TFIIB, TFIIEα and a few components of TFIIH were identified in P. falciparum. In addition, no homolog of the RNAP II-associated TFIID complex, which is essential for the basal transcription in eukaryotes, was found, except the TATA-binding protein (TBP) [23]. Therefore, it has been suggested that only a few specific transcription factors may be required for transcription regulation in the parasite. However, the parasite protein levels may also be primarily determined by posttranscriptional mechanisms [9,10,23]. The high proportion of orphans in the Plasmodium genome relative to other organisms (~60% ORFs which have no match with any known sequences [24]) suggests that the paucity of recognizable orthologous GTFs associated with RNAP II in P. falciparum may be explained in a different way. As ORF, gene and function predictions have been performed in a similar way in Plasmodium and in other sequenced genomes (such as various predictive tools trained on the Plasmodium sequences; BLAST with default parameters [24,25]), two hypotheses can be raised. First, it is possible that the parasite proteins have structurally evolved beyond the point where they cannot be identified by simple similarity searches [23,26]. Second, the extraordinary bias toward A+T richness (80%) in nucleotide composition of the parasite, may introduce large changes in both DNA and amino acid sequences which may affect the search procedures. This is particularly striking with an overall high A+T nucleotide content in protein-coding regions, leading to a remarkable bias toward the presence of stretches composed of a few amino acids only. Therefore, it is likely that a substantial number of the unusually high proportion of malarial orphan proteins with no predictable function may actually correspond to «hidden» orthologues. Interestingly, we and others ([22,27]; our unpublished results) have observed that there is a strong selection against low complexity inserts within core secondary elements of secondary structures of P. falciparum proteins. The low complexity sequences are mostly located between two adjacent globular domains and only infrequently invade globular domains.
In the present study, we postulated that the hydrophobic cores of globular domains in functional proteins of P. falciparum should be largely conserved. Consequently, these hydrophobic cores could be identified using appropriate tools involving the analysis of the secondary structure, which is often much more conserved than the primary structure [28]. We have developed and applied a two-dimensional approach of sequence analysis, called Hydrophobic Cluster Analysis (HCA), which has been useful for the prediction of orthologous proteins in different eukaryotic lineages [29,30]. HCA is based on the physico-chemical and topological properties underlying the fold of globular domains. It allows a direct access to the gravity centers of regular secondary structures (RSSs). This information can be used to pick up hidden relationships within non-significant results provided by standard similarity search methods, based on literal approaches. Indeed, the positions of hydrophobic clusters defined using HCA, which distinguish from simple binary patterns, mainly correspond to those of regular secondary structures [31,32]. Importantly, HCA is not sensitive to gaps, even large, the handling of which is one of the main obstacles of conventional sequence comparison methods. The distribution of the secondary structures also indicates the limits of structured domains. This information can help the computational analysis, in particular for P. falciparum sequences for which low complexity regions often disturb standard similarity searches. Using the HCA methodology in combination with standard similarity search methods, we have explored the P. falciparum sequences for the presence of subunits of the basal transcription factors and cofactors associated with RNA polymerase II (RNAP II). Our data suggested that several orphan proteins of P. falciparum can be predicted as general transcription factors involved in the parasite RNAP II transcription machinery.
Results
We have collected protein sequences from the different subunits of the basal transcription factors and cofactors in different genomes. The Homo sapiens and Saccharomyces cerevisiae sequences used are listed in Table 1. These sequences were used as queries for PSI-BLAST searches within the non-redundant database (nr) at NCBI. This search leads to the construction of profiles specific to each protein or protein domain. All P. falciparum sequences predicted in this study are underlined in Table 1 and shown in red in Fig. 1. Several P. falciparum sequences were easily identified as significantly matching with some subunits of known basal transcription factors of the complex RNAP II machinery. It appears that only a few of these putative parasite transcription factors such as TFIIB, TFIIE-α and several components of TFIIH, correspond to those reported elsewhere [23] (non-underlined sequences in Table 1, indicated in blue in Fig. 1). However, in several cases, the two-dimensional analysis provided by HCA led to extend the similarity outside of the limits initially reported by PSI-BLAST (hits indicated with the symbol "+" in Table 1). When no significant similarity was highlighted in the PSI-BLAST data, marginal similarities (Expected values above the threshold value) were investigated using HCA. This led to the prediction of several hypothetical protein or orphans as novel, potential orthologous basal transcription factors and cofactors associated with the RNAP II machinery in P. falciparum (sequences with an asterisk (*) in Table 1, indicated in red in Fig. 1).
Table 1 General transcription factors predicted in Plasmodium falciparum
Factors H. sapiens S. cerevisiae P. falciparum Nuclear signals prediction Expression pattern
NLS NES Micro-array Proteomic data
TFIIA α P52655 P32773 (TOA1) MAL7P1.78 + - - G -
TFIIA β
TFIIA γ P52657 P32774 (TOA2) PFL2435w *+ - - T,Sc -
PFI1630 * + - G -
TFIIB Q00403 P29055 PFA0525w - - All stages -
TFIID TBP P20226 P13393 PFE0305w - - R -
TFIID TAF1 P21675 (TAF250) P46677 (TAF145) PFL1645w + - S,LT S,G
TFIID TAF2 gi:4507347 (TAF150) P23255 (TAF150) MAL7P1.134 + + R,T S
TFIID TAF5 Q15542 (TAF100) P38129 (TAF90) ?
TFIID TAF7 Q15545 (TAF55) Q05021 (TAF67) PFI1425w + + R,S -
TFIID TAF14 P42568 (ENL/AF-9) P35189 (TAF30) ?
TFIID TAF4 O00268 (TAF135) P50105 (TAF48) ?
TFIID TAF12 Q16514 (TAF20) Q03761 (TAF68/61) ?
TFIID TAF6 P49848 (TAF80) P53040 (TAF60) ?
TFIID TAF9 Q16594 (TAF31) Q05027 (TAF17) ?
TFIID TAF11 Q15544 (TAF28) Q04226 (TAF40) ?
TFIID TAF13 Q15543 (TAF18) P11747 (TAF19) ?
TFIID TAF3 gi:13374079 (TAF140) Q12297 (TAF47) ?
TFIID TAF8 gi:31323620 (TAF43) Q03750 (TAF65) ?
TFIID TAF10 Q12962 (TAF30) Q12030 (TAF25) PFE1110w - + R, Sc -
TFIIE α P29083 P36100 MAL7P1.86 + + + Sc -
TFIIE β P29084 P36145 MAL13P1.360 * + - ND -
TFIIF α P35269 (RAP74) P41895 (Tfg1) ?
TFIIF β P13984 (RAP30) P41896(Tfg2) PF11_0458 * - + R,G -
TFIIH core p62/TFB1 P32780 (p62) P32776 (TFB1) MAL3P7.42 *+ (Chr3.phat_258) + + R,T -
TFIIH core p52/TFB2 Q92759 (p52) gi:6325135 (TFB2) PFL2125c + + R,T,Sc -
TFIIH core p44/SSL1 Q13888 (p44) Q04673 (SSL1) MAL13P1.76 + + R,T -
TFIIH core p34/TFB4 Q13889 (p34) Gi:6325313 PF13_0279 - + T -
TFIIH core TFB5 Gi:55665883 Gi:13129164 PF14_0398 - - R, T, G -
TFIIH core XPB/SSL2-RAD25 P19447 (XPB) Q00578 (SSL2/RAD25) PF10_0369 + - G S
TFIIH XPD/RAD3 P18074 (XPD) P06839 (RAD3) PFI1650w + + R,T,G G
TFIIH CAK MAT1/TFB3 P51948 (MAT1) Gi:6320668 (TFB3) PFE0610c + - R,T -
TFIIH CAK Cdk7/KIN28 P50613 (CDK7) P06242 (KIN28) ? $
TFIIH CAK Cyclin H/CCL1 P51946 (cyclin H) P37366 (CCL1) ? $
The general transcription factors which were identified in this report are underlined and shown in bold; * and + indicate similarities which were identified in this report after assessing PSI-BLAST marginal similarities at the sequence 2D level or after extending the comparison at the 2D level outside the limits primarily defined by PSI-BLAST, respectively. The Swiss-Prot accession numbers are given for the human and yeast sequences. When a Swiss-Prot identifier is not available, the genbank identifier (gi) is indicated in italics instead. The references of the human TAF3 and TAF8 sequences can be found in [50] and [100], respectively. $ : see text for comments on these putative homologues of Cdk7/KIN28 and cyclin H/CCL1. The presence of nuclear localization sequences (NLS 102), nuclear export sequences (NES [103]), and the expression of these predicted general transcription factors using microarray and proteomics on different parasite stages ([9–11, 89]) are indicated. G: gametocyte, T: trophozoite, LT: late trophozoite, Sc: schizonte, R: ring, (-): absent, (+): present, (?): cannot be found.
Figure 1 Schematic view of the predicted general transcription factors associated with RNA polymerase II in Plasmodium falciparum. Components which have been predicted in previous studies and in the present analysis, are displayed respectively in blue and in red. The components which have not been predicted from sequence analyses are shown in grey and white. Grey boxes indicate components for which potential candidates exist, but which cannot be discriminated from sequence analysis alone, due to the absence of specific domains. Green boxes indicate the HFD-containing TAF pairs which have not been identified in Plasmodium falciparum.
In all cases with marginal similarities (PSI-BLAST E-values > 0.005), the alignment with the candidate hypothetical protein has low Expected E-values, proximal to the threshold value. These values are lower than those observed for alignments with other P. falciparum hypothetical proteins. However, other potential candidates (which have higher Expected E-value) were carefully checked by HCA for similarities that might be supported at the 2D level.
Except one of the hypothetical proteins, all predicted proteins here as general transcription factors in P. falciparum have corresponding sequences in other Plasmodium species (Table 2), with identity levels above 50%. TAF10 is the only exception as it is apparently not present in P. yoelii (see Discussion). This overall conservation provides support for the prediction of the P. falciparum hypothetical proteins listed in Tables 1 and 2 as potential important components of the basic transcription machinery.
Table 2 General transcription factors predicted in this study in four Plasmodium species
P. falciparum P. yoelii yoelii P. chabaudii P. berghei Reciprocal Search
TFIIA large subunit MAL7P1.78 197 aa PY01022 57% N-ter 133 aa* PC302380.00.0 57% N-ter 133 aa* PB000347.02.0 59% N-ter 133 aa* +
TFIIA small subunit PFL2435w 131 aa chrPyl_02265-4-2031-1630 71% tl 134 aa PC403116.00.0 68% 43 aa (partim) PB101071.00.0 63% 36 aa (partim) B: A.thaliana (gi 1429228)
PFI1630c 184 aa° PY01831 51%tl 200 aa PC000365.00.0 80% tl 105 aa PB001668.02.0 95% tl 200 aa B: G. theta (gi 4583664)
TFIID Taf1 PFL1645w 3896 aa PY03752 65% $ 3182 aa PC000201.00.0 64% $ 1254 aa PB000870.00.0 64% $ 843 aa +
TFIID Taf2 MAL7P1.134 3351 aa PY03343 80% $ 1684 aa PC000872.02.0 80% $ 1353 aa PB000540.02.0 78% $ 926 aa +
TFIID Taf7 PFI1425w 397aa PY04173 58% tl 321 aa PC000532.04.0 56% tl 387 aa PB000149.02.0 54% 325 aa +
TFIID Taf10 PFE1110w 116 aa ? genomic PB108412.00.0 51% tl 93aa B: O.sativa (gi 50726230)
TFIIE α-subunit MAL7P1.86 400 aa PY00824 57% tl 369 aa PC000361.01.0 64% tl 386aa PB000518.01.0 64% tl 381 aa +
TFIIE β-subunit MAL13P1.360 542 aa PY01317 53% $ 2329 aa PC103304.00.0 81% $ 207 aa PB100065.00.0 54% $ 548 aa B: S.cerevisiae (sp P36145)
TFIIF β-subunit PF11_0458 317 aa PY03467 60% tl 310 aa ? PB000215.00.0 60% tl 175 aa* B: C.parvum (gi 46228562)
TFIIH P62 MAL3P7.42 670 aa PY00359 59% tl 674 aa PC000077.04.0 62% tl 682 aa PB000867.00.0 71% tl 343 aa -
TFIIH TFB5 PF14_0398 67aa chrPyl_00238-4-3595-3377 92% tl 73 aa Pc_1897-6-1673-1455 92% tl 73 aa PB000215.03.0 91% tl 67 aa +
The general transcription factors predicted in this study are listed with the corresponding sequences in Plasmodium yoelii yoelii, Plasmodium chabaudii and Plasmodium berghei. Reciprocal searches were performed using each predicted transcription factor, leading to the finding of similarities with the corresponding subunit in the crown group eukaryotes (see comments in the text; + stands for significant, B stands for background (E value >0.005)). The percentages of identity with the P. falciparum sequences are indicated (tl stands for calculated on the total length of the Plasmodium sequences), as well as the lengths of the considered protein sequences. $ : calculated on the similarity regions discussed in the text *: likely uncomplete sequence; °: presently corrected sequence; genomic: found in the genomic sequences, using TBASTN. 2.1.2.
Reciprocal searches were carried out for all the predicted GTF components. In most cases (indicated with a "+" in Table 2), these led to the retrieval, with significant E-values, of the corresponding sequences in other eukaryotes. The reciprocal searches were often conducted using as a probe the similarity region, excluding low complexity regions that are abundant in P. falciparum sequences. However, the profiles deduced from the P. falciparum sequences are generally less informative than those constructed using as probes the human or yeast sequences. As a consequence, such reciprocal searches resulted, in a few cases, in the retrieval of the corresponding sequences in other eukaryotes with marginal, but low expected E-values (just above the threshold E-value). It should be noticed that the sequences of another apicomplexan parasite, Cryptosporidium parvum, often constituted the link between Plasmodium and the crown group eukaryotes.
Additional support for our predictions also comes from other data, such as the prediction of nuclear localisation and nuclear export signals (NLS and NES), as well as the analysis of expression patterns (Table 1). However, it should be mentioned that nuclear factors do not always require the presence of NLS or NES for their targeting into the nucleus. For instance, it has recently been described that the nuclear transport of human TAF10, which lacks both NLS and NES, is mediated by its interacting partners, which contain the nuclear targeting signals [33].
Throughout this study, we decided to designate the putative transcription factor, PfTFIIA for P. falciparum ortholog of higher eukaryote TFIIA. The same nomenclature will be used for the other basal transcription factors and cofactors identified here.
PfTFIIA
The TFIIA proteins form a ternary complex with TBP and DNA. It stabilizes the TBP-DNA binding and promotes the binding of TFIID complex to DNA. Yeast TFIIA is composed of two subunits (TOA1 and TOA2), which can each be divided in two parts, a N-terminal helical region and a beta-strand containing C-terminal region. The N-terminal regions of the two subunits form together a four-helix bundle, whereas the two C-terminal ones fold as a six-stranded beta-barrel contacting TBP-DNA [34,35]. The human TFIIA homologue is made of three polypeptide chainsα/β (large subunit encoded by a single chain, which is post-translationally processed) and γ (small subunit) [36].
The P. falciparum orthologous TFIIA large subunit was easily identified using the yeast TOA1 sequence. We found that the C-terminal part of this first subunit TOA1 sequence (aa 214 to 285) can be aligned with a significant PSI-BLAST E-value (E-value 3 10-4 by iteration 2) with the C-terminal part of the hypothetical protein MAL7P1.78 from P. falciparum (Fig. 2, panel A). Although no similarity was highlighted by PSI-BLAST with the N-terminal part of the proteins, HCA indicates the presence of large helices in the N-terminal part of the malarian protein, which can be aligned with those of the yeast and human sequences (Fig. 2, panel A). This N-terminal sequence similarity could not be detected, even in the background noise, using the first 100 amino acids of either TFIIA α (yeast and human sequences) or MAL7P1.78 as queries in PSI-BLAST searches. Thus, the HCA methodology allowed in this particular case to significantly extend the similarity between the human/yeast and malarian proteins over their whole lengths, suggesting that MAL7P1.78 can be predicted as the PfTFIIA large subunit. The amino acid region separating the N- and C-terminal parts is highly variable between species [37]. In the putative Pf TFIIA large subunit (MAL7P1.78), this region between the N- and C-termini is shown to be smaller than in the yeast and human sequences (Fig. 2, panel A).
Figure 2 Comparison of the HCA plots of TFIIA large subunit (TOA1/TFIIA α/β panel A) and small subunit (TOA2/TFIIA γ panel B) subunits from different species, highlighting the conservation of the hydrophobic core of the domains constituting the two proteins in the hypothetical proteins from Plasmodium falciparum. The sequences are shown on a duplicated α-helical net, in which strong hydrophobic amino acids (VILFMYW) are circled. These form clusters, which mainly correspond to the internal faces of regular secondary structures (α-helices and β-strands). The way to read the sequence and special symbols is indicated in the inset. The regions initially detected by PSI-BLAST (either with significant (TOA1/TFIIA α) or marginal E-value (TOA2/TFIIA γ) are indicated with a dotted line. Cluster similarities are shaded in grey, identities are shown in white on a black background. Despite low level of sequence identity, hydrophobic clusters are well conserved, supporting the presence of a common fold. The deduced 1D alignment is shown at the bottom and at right. The positions of regular secondary structures, as observed from experimental data (pdb 1nh2) are shown up to the HCA plot. Two hypothetical proteins from Plasmodium falciparum share significant similarities with the small subunit of TFIIA (bottom panel) and are thus PfTFIIA small subunit candidates. The similarity with PFI1630c was revealed using its homolog sequence in Plasmodium yoelii (see text). This sequence was missed during our first PSI-BLAST search because of an error in the intron prediction. This novel ortholog has been found through HCA sequence comparison of the translated DNA sequences (the boxed and underlined sequences in the HCA plot and 1D alignment, respectively, correspond to the sequence which was first included in an intron, as predicted automatically from genome data).
Using the sequence of the yeast small subunit TOA2, as query in a PSI-BLAST search, no significant sequence similarity could be found with P. falciparum proteins derived from the whole genome databases at convergence by iteration 2. However, a marginal similarity (E-value of 4.6) was highlighted with the PFL2435w hypothetical sequence, over 83 amino acids (22% identity). This similarity was supported at the 2D level using HCA (Fig. 2, panel B). It covers the N-terminal region as well as the two first strands of the C-terminal region. The third strand can be tentatively identified at the C-terminus of the P. falciparum sequence, when a large insertion is made between the second and third beta-strands. This large insertion likely corresponds to a globular sequence, as assessed by the presence of hydrophobic clusters. A large loop region also exists in this location in the human and yeast sequences, but was not observed in the solved corresponding three-dimensional structures. Another marginal similarity was observed at similar level of Expected-value in the PSI-BLAST results with a hypothetical protein of P. yoelii (PY01831; 24% identity over 82 amino acids, E-value= 0.084). However, this hypothetical protein does not correspond to the PFL2435w homolog. Instead, the PY01831 homolog in P. falciparum corresponds to the PFI1630c hypothetical protein (43% identity). This similarity was however not detected by PSI-BLAST because the PFI1630c sequence was incorrectly predicted (part of the coding region was inappropriately predicted as an intron; more explanations are given in the legend of Fig. 2). The corresponding alignment was also supported at the 2D level using HCA (Fig. 2; panel B). The PFI1630c hypothetical protein contains an N-terminal extension, relative to the human TFIIA γ/yeast TOA2 sequences. This suggests that two genes could exist as functional TFIIA small subunits in Plasmodium falciparum. Multiple genes that encode general transcription factors have already been described for the TATA-box binding protein (TBP) in several species [38-40] and for TFIIA α/β in humans [37].
PfTFIIB
TFIIB, which associates with TFIIA, is the only putative general transcription factor (PFA0525w) that was so far identified during the annotation of P. falciparum genome. It was confirmed by specific HMM searches performed by Coulson et al. [23].
PfTFIID
Evidence for the presence of some P. falciparum TBP-associated factors (TAFs) involved in the multiprotein PfTFIID complex
The TATA-binding protein (TBP) and many TBP-associated factors (TAFs) form the multimeric TFIID complex [41]. While TBP is sufficient for basal transcription in vitro, the TAF subunits of TFIID are essential cofactors for transcriptional activation by providing interaction sites for activators. Yeast TFIID contains 14 TAFs and homologues of many of these TAFs are found in metazoans (Table 1 and [42]). Analysis of the architecture of yeast and metazoan TFIID revealed that more than half of the TAFs contains a histone fold motif (HFD) (Table 1 and [42]). These HFDs specifically assemble into five histone-like pairs.
While TBP was clearly identified in the P. falciparum genome (PFE0305w, Table 1), no orthologous TAFs have been described so far from the genome sequence data of several apicomplexan parasites including several Plasmodium species [24,43], Cryptosporidium parvum [44] and T. gondii . This suggests that this TAF detection failure cannot only be ascribed to the A+T richness of the genome. Indeed, unlike Plasmodium species, the other apicomplexan parasites do not display a bias toward A+T richness. Instead, it is likely that the amino acid sequences of TAFs in apicomplexan parasites reached a point of divergence that hinders their prediction using classical similarity searches. Here, we searched for the presence in P. falciparum of each of these TAFs, including those which contain histone fold motifs.
• PfTAF1 (hTAF250/yTAF145)
In metazoan, the largest TFIID subunit has three enzymatic activities (kinase, histone acetyltransferase (HAT) and ubiquitin-activating and conjugating (ubac) activities) involved in transcriptional regulation (reviewed in [42]). Metazoan TAF250 possess a pair of C-terminal bromo domains, which recognize acetylated histones. In yeast TAF145, which otherwise lacks kinase activity, these bromodomains are not present. Instead, two interacting proteins Bdf1 and Bdf2 provide the missing enzymatic activity and functional domains. We focused our searches on the conserved domain of proteins of the TAF250 family, which is critical for HAT activity [45] (TAF1; aa 549–1290 of human TAF250). We identified, by iteration 2, a significant similarity with P. yoelii PY03752 (E-value 1 10-5), and by iteration 3 with P. falciparum PFL1645w (E value: 3 10-13) (Fig. 3, panel A). These similarities were supported at the 2D level using HCA (data not shown). These two orthologous hypothetical proteins possess one bromo domain in their C-terminal parts. Careful study of their HCA plots led to define the limits of the conserved domain between aa 1420 and 1650 (PFL1645w). Clear hinge regions could also be identified. Using this domain as query in PSI-BLAST led to the identification of all members of the TAF250 family by iteration 4. No other protein from the P. falciparum genome data was identified, suggesting that PFL1645w might be the genuine TAF1 ortholog.
Figure 3 1D alignment of different TAF subunits (TAF1, TAF2 and TAF7) with hypothetical proteins from P. falciparum. Identical and similar amino acids are boxed in black and in grey, respectively. Although restricted to a limited length, the similarity regions highlighted here match the inter-species conserved regions (see text). These similarities are supported at the 2D level using HCA (data not shown).
• PfTAF2 (hTAF150/yTAF150)
TAF150 proteins have a non-specific aminopeptidase domain in their N-terminal parts. We therefore focused our searches on the C-terminal parts of the proteins. Using PSI-BLAST and the yeast TAF150 C-terminal domain (aa 701 to 1407) as query, a significant hit appeared by iteration 2 with the P. yoelii PY03343 hypothetical protein (E-value 0.002), together with those relative to other metazoan TAF150. The identification of P. yoelii orthologous TAF2 has been used to discover the P. falciparum TAF2, which is currently named in the annotated genome as the hypothetical protein MAL7P1.134. This sequence was scored with a significant E-value (3 10-7) by iteration 3 (Fig. 3, panel B). This similarity is supported at the 2D level and concerns the region which is most conserved in the TAF150 C-terminal domain amongst the different species. This suggests that the P. falciparum protein pinpointed here might be the TAF2 ortholog in the parasite.
• PfTAF7 (hTAF55/yTAF67)
TAF7 proteins possess a conserved domain (TAFII55 protein conserved region), located between amino acids 112 and 305 (yeast) or amino acids 12 and 178 (human) [46]. Using this domain as query in PSI-BLAST led to the identification of significant similarities from the second iteration with both P. yoelii PY04173 (E-value 2 10-6) and P. falciparum PFI1425w (E-value 2 10-6) hypothetical proteins. This similarity, limited to the first part of the TAFII55 protein conserved region (PFI1425w aa 161 to 242), is supported at the 2D level (Fig. 3, panel C). This similarity was also retrieved when scanning the Pfam database (pfam04658.5, TAFII55_N). However, the globular domain of the P. falciparum proteins in which the TAFII55-like region is included appears to be larger (aa 148 to ~ 325), and thus might share a similar length to the complete TAFII55 protein conserved region. The region of similarity shared by P. falciparum PFI1425w and other TAF7 was previously shown to be critical for interaction with the bromo domain factor Bdf1 of yeast cells [46].
• PfTAF10 and the apparent lack of TAFs assembling into histone-like pairs in P. falciparum
The histone fold domain (HFD), the core of which is characterized by three alpha-helices, is a fundamental interaction motif involved in heterodimerization of the core histone (H4-H3, H2A-H2B) and their assembly into a nucleosome octamer. This motif is thought to have arisen from the duplication of a minimal helix-extended-helix structure. The two middle helices of the duplicated structure would have fused to form a long, central helix. The histone fold domain can be accompanied by N- or C-terminal extensions, also made of alpha-helices and is found in several non-histone proteins, in addition to core histones [42,47].
Analysis of TFIID has shown that more than half of the TAFs constituting this complex are HFD containing proteins (reviewed in [42]). This led to the first hypothesis of a compact nucleosome-like octamer core in TFIID, which could bind DNA and around which other TAFs could associate [48] (reviewed in [49,50]). This proposal has however to be revisited in light of recent experimental data, highlighting a more complex situation than initially thought. First, irrespective to the nature of the quaternary structure (nucleosome-like octamers, as observed for the TAF4/TAF12 – TAF6/TAF9 assembly [51], or other structures), it has been shown that surface residues of core histones known to make critical contacts with DNA in the nucleosome are generally not conserved in TAF HFDs [52,53]. This suggests an alternative role for HFD in TAFs than DNA binding. Second, immunolabeling electron microscopy experiments have demonstrated that the HFD-containing TAFs are located in three distinct substructures of TFIID, which are assembled by thin linker domains in a molecular clamp architecture [54]. The TAF4/TAF12 – TAF6/TAF9 assembly was shown to co-localize in one of the three lobes of native TFIID [55]. These structural data were supported and enriched by additional mapping of other TFIID functional sites [55].
Our searches for HFD-containing TAFs in the Plasmodium genome did not lead to the identification of any of the five histone-like pairs currently known in other eukaryotic species (TAF6-9, TAF11-13, TAF4-12, TAF3-10 and TAF8-10). These searches were performed using as queries the full-length sequences of yeast and human TAFs, their HFDs, and specific domains accompanying HFDs (e.g. for TAF4, we considered the specific TAF4 domain, including HFD (hTAF135 aa 832 to 1083); the HFD (hTAF135 aa 835 to 950) and the TAFH sequence (hTAF135 TAF homology region, also known as nervy homology region 1 (NHR1); smart00549; aa 590 to 649)).
Only one candidate for TAF10 was retrieved in the Plasmodium genome. The TAF10 subunit in yeast, TAF25, heterodimerizes with TAF3 (yeast TAF47) and TAF8 (yeast TAF65) [50]. It is one of the essential component common to TFIID and SAGA [56]. Yeast TAF25 was predicted to have a HFD, with which it can dimerize with its partner [50]. However, the presence of HFD in yeast TAF25 has not yet been experimentally demonstrated. Searching databases using the yeast TAF25 sequence as a probe led to the identification by convergence of marginal similarities with the P. falciparum PFE1110w hypothetical protein (E-value 1.5; 25% sequence identity over 52 amino acids). These similarities were supported at the 2D level (Fig. 4). A reciprocal search yielded the TAF10 proteins as the first hits above the threshold value.
Figure 4 Comparison of the HCA plots of TAF10 from human (TAF30) and yeast (TAF25) and the P. falciparum hypothetical protein PFE1110w (see figure Fig. 2 for explanation) Cluster similarities are shaded in grey, identities are shown in white on a black background. Vertical bars indicate cluster links. The deduced 1D alignment is shown at the bottom.
Taken together, our data strongly suggest the apparent and unexpected lack of HFD containing TAFs in P. falciparum, except from TAF10. This TAF however remains to be determined as a genuine HFD-containing factor in the parasite and also in the other eukaryotic cells.
Other undetected TAFs within the multiprotein PfTFIID complex
• PfTAF5 (hTAF100/yTAF90)
This protein, interacting with TFIIFβ, contains WD40 repeats. We specifically limited our searches to the WD40 associated region in TFIID subunit (pfam04494; aa 194–340 of hTAF100), but these did not lead to the identification of potential TAF5 candidates. We were thus unable to report the presence of a putative TAF5 ortholog in P. falciparum.
• PfTAF14 (hENL-AF9/yTAF30)
TAF14, named AF9 in human and TAF30 in yeast, is the only non-essential TAF. It is also a component of TFIIF in yeast. TAF14 contains two globular domains, the N-terminal one belonging to the YEATS family of domains, which are found in several proteins involved in chromatin modification and transcriptional regulation [57]. A YEATS protein can be found in the P. falciparum MAL8P1.131 protein, described as a Gas41 homologue. However, the P. falciparum MAL8P1.131 protein does not possess the C-terminal domain common to yTAF30 and hAF9. The similarity between the C-terminal domains of these proteins can be only detected using HCA (Figure 5). However, the clusters typifying the architecture of the C-terminal TAF14 domain are not found in MAL8P1.131 protein, suggesting that this protein may not correspond to a genuine PfTAF14.
Figure 5 Comparison of the HCA plots of the TAF14 subunit in humans (AF9) and in yeast (TFG3) with that of the Plasmodium falciparum MAL8P1.131 hypothetical protein (see Fig. 2 for explanation). Cluster similarities are shaded in grey and identities are shown in white on a black background. A N-terminal YEATS domain [57] is present in the three sequences, whereas HCA detects a common domain in the C-terminal end of only human AF9 and yeast TFG3. The conserved clusters of this C-terminal domain are not detected in the MAL8P1.131 sequence, suggesting that this protein may not correspond to PfTAF14.
PfTFIIE
TFIIE is an heterotetramer composed of a large and a small subunit, referred as α (human)/ TFA1 (yeast) and (human)/ TFA2 (yeast), respectively [58,59]. A clear homologue of the TFIIE α subunit, the MAL7P1.86 hypothetical sequence, was readily identified from the first iteration starting from human TFIIE α sequence (E-value 1 10-5; sequence identity of 26% over the first 162 amino acids of TFIIE α totalizing 439 amino acids). This TFIIE α homologue was also reported by Coulson et al. [23] and has recently been designated in sequence databases as the putative TFIIE α-subunit. The N-terminal sequence pinpointed here is required for activation of basal transcription in vitro through interactions with TBP and Pol II [60,61]. This N-terminal sequence contains an extended winged helix domain [62] followed by a zing finger [63]. Most importantly, the analysis reported here led to extend the similarity between the human and P. falciparum proteins over a small C-terminal globular domain of TFIIE α, as secondary structure similarities, supported by sequence identities, could be clearly identified (Fig. 6, panel A). This similarity was not detected by PSI-BLAST (below and up to the threshold E-value), even if only this small domain instead of the whole protein was used for similarity searches. The small acidic globular domain, for which no known fold could be predicted by threading procedures (3D-PSSM, [64]) (data not shown), is absent in the yeast sequence, the C-terminal of which is likely unstructured [65]. This domain may thus be specific of higher eukaryotes and P. falciparum. This region binds directly to TFIIH and has a stimulatory effect on basal transcription [61].
Figure 6 A) Identification of a small globular domain common to the C-termini of TFIIE α subunits of higher eukaryotes and P. falciparum. Top panel: Comparison of the corresponding HCA plots (see figure Fig. 2 for explanation). Cluster similarities are shaded in grey and identities are shown in white on a black background. The position of the globular domain is boxed. Despite the low level of sequence identity, hydrophobic clusters are well conserved, supporting the presence of a common fold. Bottom panel: HCA-deduced 1D alignment. B) Comparison of the TFIIE β subunits of S. cerevisiae, Cryptococcus neoformans and Plasmodia species (P. yoelii and P. falciparum). Cluster similarities are shaded in grey and identities are shown in white on a black background. The deduced 1D alignment is shown at the bottom.
No P. falciparum orthologue could be found for the second subunit of TFIIE, named TFIIE in a first approach using as queries the human and yeast sequences in PSI-BLAST searches. These two sequences display a low level of sequence identity (less than 30%), especially concentrated in the C-terminal segment. We adopted an iterative strategy, by using distant sequences of the family described in the PSI-BLAST data and thereby discovered potential relationships with Plasmodium proteins. Hence, using the sequence of the hypothetical protein CNBI3180 from Cryptococcus neoformans, which shares 22% of sequence identity with the S. cerevisiae TFA2 over 293 amino acids (E-value by convergence 1 10-28), we found by iteration 3, a significant similarity with the C-terminal fragment of a hypothetical sequence from P. yoelii (PY01317; 15% identity over 219 amino acids, E-value 2 10-38). This relationship was supported at the 2D level (Fig. 6, panel B), especially highlighting well conserved hydrophobic clusters common to distant members of the family (for example, see the hydrophobic cluster highlighted with an asterisk in Fig. 6, panel B). The corresponding sequence in P. falciparum, MAL13P1.360, was found by searching the predicted annotated proteins within PlasmoDB (version 4.3, November 2, 2004; 81% sequence identity with PY01317). The comparison of the whole set of sequences of the TFIIE β subunit family revealed a region of variable length in the middle of the TFIIE β domain. This region is particularly rich in cysteine residues in the proteins of apicomplexan parasites such as P. falciparum, P. yoelii and Cryptosporidium (boxed in Fig. 6, panel B).
PfTFIIF
TFIIF, a tetramer of two subunits, named α (mammalian RAP74/yeast TFG1) and β (mammalian RAP30/yeast TFG2), is intimately associated with the RNA polymerase II enzyme [66]. The TFIIF complex directly binds promoter DNA, TFIIB and the TAF250 subunit of TFIID, and recruits TFIIE and TFIIH to the preinitiation complex [67]
The β-subunit can be divided into two globular regions separated by a central, less structured region. The structures of these two globular domains have been solved in human RAP30 [68,69]. The N-terminal domain is responsible for RAP74 and TFIIB binding [68], and forms with the N-terminal domain of RAP74 a triple barrel dimerization fold [70]. The C-terminal domain, containing a winged-helix [69], binds non-specifically to DNA [71]. Using the human RAP30 and yeast TFG2 sequences as queries in PSI-BLAST, no significant similarities with P. falciparum proteins were detected at convergence by iteration 3. However, marginal similarities were detected above the threshold value with the hypothetical PF11_0458 protein (E values 9.4 and 0.25, respectively; 12% and 19% identity over 326 and 324 amino acids, respectively). These similarities were supported at the 2D level (Fig. 7). Some of the regular secondary structures constituting the core of the globular domains are particularly well conserved, whereas the region separating these domains appears less similar, as observed for other β subunits of different species. However, a good conservation was observed only for the strands beta-2, beta-3 and beta-4 of the N-terminal domain, suggesting that the structure of the parasite protein may locally differ from that of human RAP30.
Figure 7 Comparison of the human TFIIF β subunit (RAP30) with the hypothetical protein PF11_0458 from P. falciparum. Top panel: Comparison of the corresponding HCA plots (see Fig. 2 for explanation). The N- and C-terminal structured domains are boxed, according to the limits defined on the basis of experimental data (pdb 1f3u (chain A) and 2bby, respectively). Cluster similarities are shaded in grey and identities are shown in white on a black background. Putative correspondences in the C-terminus of the first domain are reported with dashed lines. Secondary structures, as observed in the experimental structures, are reported up to the RAP30 sequence. The corresponding 1D alignment is shown in the bottom panel.
RAP74 (α subunit) also possesses two N- and C-terminal globular domains, separated by an unstructured linker sequence. As indicated above, the N-terminal domain, which is responsible for TAF250 and RAP30 binding, forms with the RAP30 N-terminal domain a triple barrel dimerization fold [70]. The C-terminal domain, interacting with TFIIB, FCP1 and DNA, folds as a winged-helix [72]. We thus restricted our queries to these two domains, the limits of which were identified through HCA (from aa 1 to 180 and 450 to end). However, we found no significant similarity with any parasite proteins including several Plasmodium species and Cryptosporidum parvum. We also did not see any marginal similarity, which could be confirmed at the 2D level. Given the single core structure that the two subunits form together, with three interwoven beta- barrels, it seems unlikely that the RAP30 homolog exists alone in apicomplexan parasites. Alternatively, either the RAP74 subunit is too divergent to be detected using the available tools, including HCA, or it does not really exist, suggesting the replacement of a α-β heterodimer by a β-β homodimer instead, given the similar architecture of the two subunits.
PfTFIIH
The general transcription factor TFIIH is the largest and most complex of all. Indeed, it is composed of nine subunits with molecular mass (460 kDa) similar to that of RNA PII, with several subunits having enzymatic activities (reviewed in [67,73-75]). TFIIH has a dual action in both transcription initiation and nucleotide excision repair (NER). It is organized into two structural and functional entities. The first of these, the TFIIH core, includes four polypeptides (named P62, P52, P44 and P34 in human; yeast orthologous sequences are indicated in Table 1) and the xeroderma pigmentosum B (XPB) helicase. The second functional entity, the CDK-Activating Kinase (CAK) complex, is composed of the cyclin-dependent kinase Cdk7, cyclin H and MAT1. The XPD (RAD3) helicase bridges the two complexes, being associated either with the core or CAK. In addition to this, a new subunit of the TFIIH core, TFB5, has recently been discovered, associated in humans with DNA repair-deficient trichothiodystrophy [76,77].
TFIIH is the most thoroughly documented complex in Plasmodium falciparum. Indeed, five out of six core subunits (except from human P62/ yeast TFB1) can easily been identified using profile-based searches. The XPD (RAD3) helicase and components of the CAK complex [23] were also identified (Table 1). However, no direct homologue of Cdk7 and Cyclin H, sharing a high level of sequence identity with the human and yeast counterparts, was identified. Nevertheless, several Cdk7 and Cyclin H putative homologs with lower identity values can be identified. However, their exact nature remains to be determined. Coulson et al. [23] noticed the presence of a cyclin K homolog (PF13_0022), which was shown to be associated with RNAP II [78]. We focused our searches on the missing P62/TFB1 subunit, which possesses two copies of a BSD domain [79] and was recently shown to possess in its N-terminus a domain with a PH fold [80]. Plasmodium sequences marginal similarities were detected by searching databases using the human p62 as query. The sequence alignments display E-values of 0.35 with Plasmodium yoelii PY00359 and 10 with Plasmodium falciparum MAL3P7.42. The corresponding similarities were supported at the 2D level (Fig. 8, panels B and D). The similarities with these Plasmodium sequences were found to be significant when the yeast TFB1 sequence was used instead as query (E-values after convergence by iteration 6 of 2 10-68 and 2 10-59 for the alignments with the PY00359 and MAL3P7.42 sequences, respectively – 14% and 13% identity over 430 and 432 amino acids, respectively). Most of the similarities were focused on the BSD domains (Fig. 8, panels B and D), but longer alignments within the C-terminal part of the sequences were also observed. These C-terminal alignments were also supported at the 2D level, as illustrated in Fig. 8 panel C, and include sequences which are predicted to form long helical structures. Finally, HCA suggests that the N-terminal sequence of P62/TFB1, corresponding to the PH domain [80], might also be aligned with the Plasmodium sequences, although this relationship was not highlighted using PSI-BLAST (Fig. 8, panel A).
Figure 8 Comparison of the human TFIIH P62 subunit with the hypothetical MAL3P7.42 from P. falciparum. Significant similarities were detected for the region including two BSD domains (boxed, panel B). These were supported at the 2D level by comparison of the HCA plots (see Fig. 2 for explanation), where cluster similarities are shaded in grey and identities are shown in white on a black background. Marginal similarities observed in PSI-BLAST are located within the C-terminal parts of the proteins (ranging from ~250 to the C-terminus (human) and from ~aa 370 to the C-terminus (P. falciparum)). These were also supported at the 2D level, as illustrated here on a segment in the most distal C-terminal part of the two proteins (panel C). Upstream of the BSD domain, cluster similarities together with sequence identities can be observed in the most proximal N-terminal part of the protein sequences, corresponding to a pH domain in human p62 (panel A). Secondary structures, as observed in the experimental structure of human p62 (pdb 1pfj; [80]), are reported up to its sequence. The corresponding 1D alignment is shown in panel D.
Discussion
Only a few components of the general transcription machinery have been identified to date in P. falciparum [23]. Of the 33 general transcription factors listed in Table 1, only one third (ten subunits) were predicted from simple similarity searches [24] and previous analysis [23]. This percentage may reflect the poor proportion of gene with automatically predicted function in the complete parasite genome [24]. Hence, the TATA binding protein or TBP is the only known component of the TFIID complex, which has been identified. The multicomplex TFIID remains however essential for accurate and higher transcription levels in eukaryotic cells. Therefore, the paucity of both defined malarial TFIID orthologous components and of general transcriptional factors, contrasts significantly with the situation reported for the crown group eukaryotes, in which TFIID is well conserved even though some differences can be seen for transcription cofactor complexes. Here, the use of the sensitive Hydrophobic Cluster Analysis (HCA) in combination with profile-based search methods suggests that the genome of P. falciparum contains several gene products annotated as hypothetical proteins, which can be predicted as putative general transcription factors (GTF) associated with RNAP II. These include several members of TFIID even if most of the TAF containing histone fold domains (HFDs) remain undetected using the sensitive Hydrophobic Cluster Analysis (HCA). Nine other GTFs were predicted in this way, which brings the total number of predicted subunits of the general transcription machinery to approximately 60% that of the number observed in the crown group eukaryotes.
The lack of detection of several GTFs in the parasite using conventional methods for sequence similarity searches could be ascribed to a higher divergence of these proteins, as well as to the bias introduced in the searches by the overall high A+T nucleotide content. The apparent divergence between Plasmodium GTFs relative to their orthologs in free-living organisms is consistent with the observation previously reported by Coulson et al. [23,81]. These authors have noticed that transcription-associated proteins family taxon specificity appeared to correlate with evolutionary distance and not cellular complexity. On the other hand, Plasmodium sequences are very difficult to analyze due to a particular amino acid bias, reflecting the overall high A+T nucleotide content. This results in unusually high proportions of asparagine and lysine, and to a lesser extent also of isoleucine and tyrosine, which are all encoded by AT-rich codons. This abundance contrasts with a relative paucity in arginine, alanine, proline and glycine, encoded by GC-rich codons (Fig. 9 and [82]). However, part of the most abundant amino acids, asparagine and lysine, is located within the low complexity regions that are often located outside functional domains (K. Prat, J.P. Mornon and I. Callebaut, unpublished results). When we evaluated the presence of the hydrophobic core forming amino acids (class 1, Fig. 9) which are crucial for fold stability, the P. falciparum proteome is similar to the other organisms for the frequencies of phenylalanine, methionine and tryptophane. In addition, there is a nearly perfect balance between valine, leucine and isoleucine, which are chemically related and often interchangeable at the structural level. The last amino acid of the hydrophobic class, tyrosine, is the most coil-forming residue of this class [30] and the increase of its frequency in P. falciparum may actually not affect the general balance of domain hydrophobic cores. The high frequency of lysine, which is on average the most exposed amino acid in globular domains [83-85], might be balanced by those of arginine and alanine. Within the third class (coil-forming residues), the low frequencies of proline and glycine, which have on average a similar behavior relative to α, β and coil states [32], might be together compensated by the considerable high frequency of asparagine residues. Asparagine immediately after glycine, shares with it the ability to adopt left-handed helical local conformations, widely represented in coil regions. Thus, the conservation of the total proportion of hydrophobic amino acids in P. falciparum, and the compensative behavior of other amino acid couples, gives evidence for the conservation of hydrophobic cores of functional domains. This also suggests that an appropriate delineation of these domains through HCA, and their specific use for sequence similarity searches, can lead to the finding of significant relationships within proteins, which otherwise remain orphan if their sequences are analyzed as a whole. The development of an automatic procedure allowing a fast prediction of the structured domain boundaries can help to apply such a strategy at the genome-scale level (K. Prat, J.P. Mornon and I. Callebaut., unpublished results).
Figure 9 Comparison of the amino acid distribution in the proteomes (predicted proteins) of P. falciparum, Homo sapiens, Saccharomyces cerevisiae and Arabidopsis thaliana (5334, 32035, 6699 and 27857 sequences, respectively). Amino acids are grouped with respect to the structural classes defined previously in [30]. The first class defines strong hydrophobic amino acids (V, I, L, F, M, Y, W), that display a similar propensity for yielding regular secondary structures (α-helices and β-strands. The third class includes coil-forming amino acids (G, P, D, N, S), whereas the intermediate class (A, R, C, Q, T, E, K) contains amino acids for which coil and secondary structure forming propensities are similar. The total class I amino acid content is similar in Plasmodium falciparum with respect to other proteomes (see comments in the discussion section). The frequency of cysteine, which is also a frequent component of hydrophobic cores, does not differ in Plasmodium falciparum relatively to other organisms. One can also observe a stable frequency for histidine, which has always a remarkably neutral behavior in the secondary structure propensities.
A first original feature in the P. falciparum predicted GTF sequences is the presence of two genes candidates for the TFIIA small subunit. The presence of two genes has already been described for some GTFs. Indeed, TBP-like proteins are found in A. thaliana, D. melanogaster and H. sapiens [38-40]. Moreover, a functional homolog of TFIIA α/β subunit, which is expressed almost exclusively in testis, has been described [37]. This multiplicity of GTFs is thought to contribute to tissue- and gene-specific regulation. It is therefore possible that the gene candidates for the TFIIA small subunit in P. falciparum are stage-specifically expressed in the parasite life cycle.
Another striking observation is that, among the GTFs associated with RNAP II predicted here, no HFD-containing TAFs could be identified, except for the putative ortholog of TAF10, which was reported by Gangloff and colleagues [50] as containing a potential HFD. However, the potential HFD of TAF10 might correspond to a distant member of the HFD family and remains to be experimentally proven. On the other hand, TAF10 is the only "HFD" protein which is shared by TFIID and SAGA [56]. It is interesting to note the apparent lack of TAF10 candidate in P. yoelii (Table 2), suggesting that this protein might not be essential in all Plasmodium species. This may be consistent with the apparent absence of other HFD-containing TAFs in all Plasmodium species. The apparent absence of canonical HFD TAFs leads to the hypothesis of a higher divergence of proteins of the HFD family in the Plasmodium genome than that of other proteins, beyond the point where they can be identified using homology searches, even the most sensitive of them. Alternatively, if the absence of HFD containing TAFs is confirmed, this will provide evidence for a striking difference in the quaternary structure of TFIID by comparison to the yeast or human complexes. The latter display a similar architecture, formed by three lobes organized into a molecular clamp [54,86,87]. Experimental investigations are needed to further explore this hypothesis. To date, the only Plasmodium proteins with HFD domains listed in the histone database [47] corresponds to classic nucleosomal histones H2A, H2B, H3 and H4. The linker histone H1 is not found [22]. This suggests that the apparent absence of histone fold proteins in Plasmodium is not only restricted to the TAF proteins of TFIID complex.
Conclusion
In conclusion, we have shown that more general transcription factors can be predicted in the genome of P. falciparum than initially thought. It can be anticipated that the HCA method can also be an additional and important tool for the finding of new orthologs amongst the high proportion of hypothetical proteins or orphans in P. falciparum and other apicomplexan parasites such as Cryptosporidium parvum, Eimeria and Toxoplasma gondii. Virtually nothing is known about transcription regulation in these apicomplexan parasites. To our knowledge, this study describes for the first time the prediction of general transcription factors in the genome of P. falciparum using a sensitive predictive method based on secondary structure considerations (HCA). Based on the GTF orthologs predicted here, there are some differences in the composition, and probably in the nature of some multicomplex factors, as illustrated by the possible absence of HFD containing TAFs in the TFIID complex. The identification of novel transcription elements and understanding how the basal transcription differs in the parasite may be exploited to design selective therapeutic agents against P. falciparum. Additionally, further elucidation of mechanisms controlling transcriptional expression in protozoa may provide a unique perspective on how these systems evolved in early eukaryotic cells.
Methods
The non-redundant database (NR; 2 456 374 protein sequences at May 3, 2005) at NCBI (National Center for Biological Information) was searched using the BLASTP program with default parameters [88] (BlastP 2.2.10, Oct 19, 2004; Blosum 62, gap penalties: existence 11, extension 1). Profile searches were conducted using PSI-BLAST, run until convergence with a default profile inclusion expect (E) value threshold of 0.005. Reciprocal searches were carried out for all the predicted GTF components of Plasmodium falciparum (see comments in the Result sections). The PlasmoDB (version 4.3, November 2, 2004) [25,89] was also searched using the same tools (BLASTP 2.1.2). Other databases (Pfam [90], Smart [91]; CDD [92]) were also searched for the presence of known domains.
The two-dimensional Hydrophobic Cluster Analysis (HCA) [29,30] was used to sort at the two-dimension level (2D) the potential sequence and structure relationships. HCA offers the possibility to add to a literal analysis, a lexical one by identifying the regular secondary structures from the consideration of a single sequence. Indeed, the positions of hydrophobic clusters were shown to mainly correspond to the positions of regular secondary structures [31]. These non-intertwined binary patterns, constrained by the consideration of a connectivity distance separating two distinct clusters on the two-dimensional plot (the currently used alpha-helix is associated with a connectivity distance of 4), are much more informative than non constrained ones [32]. Hence, similar structures are often associated with conservation of hydrophobic cluster features, which participate in the protein core, together with sequence similarities. This conservation often helps or allows the alignment procedure for highly divergent sequences (typically within and below the twilight level). This approach has been used to identify new domains (e.g. [93,94]), link orphan sequences to structural and functional families (e.g.[95,96]) or identify and characterize catalytic sites (e.g. [97-99]). Other examples can be found at [100].
Authors' contributions
IC and ST conceived the study and drafted the manuscript. IC carried out the sequence analysis, in which EM and JPM participated. KP performed the statistical analysis of genome sequences.
Acknowledgements
We would like to thank Dr. Steven Ball for critically reading the manuscript and the reviewers for providing helpful comments and suggestions. This work was supported by the CNRS through the interdisciplinary program "Protéomique et Génie des Protéines". The support of the CEA (LRC27V) to I.C., K.P. and J.P.M. is also acknowledged.
==== Refs
Sinden RE Molecular interactions between Plasmodium and its vectors Cell Microbiol 2002 4 713 724 12427094 10.1046/j.1462-5822.2002.00229.x
Lanzer M de Bruin D Ravetch JV A sequence element associated with the Plasmodium falciparum KAHRP gene is the site for developmentally regulated protein-DNA interactions Nucleic Acids Res 1992 20 3051 3056 1620601
Lanzer M de Bruin D Ravetch JV Transcriptional mapping of a 100 kb locus of Plasmodium falciparum identifies a region in which transcription terminates and reinitiates EMBO J 1992 11 1949 1955 1374714
Waters AP The ribosomal RNA genes of Plasmodium Adv Parasitol 1994 34 33 79 7976752
Scherf A Hernandez-Rivas R Buffet P Bottius E Benatar C Pouvelle B Gysin J Lanzer M Antigenic variation in malaria : In situ switching, relaxed and mutually exclusive transcription of var genes during intra-erythrocytic development in Plasmodium falciparum EMBO J 1998 17 5418 5426 9736619 10.1093/emboj/17.18.5418
Dechering KJ Kaan AM Mbacham W Wirth DF Eling W Konings RM Stunnenberg HG Isolation and functional characterization of two distinct sexual-stage-specific promoters of the human malaria parasite Plasmodium falciparum Mol Cell Biol 1999 19 967 978 9891033
Hayward RE DeRisi JL Alfadhli S Kaslow DC Brown PO Rathod PK Shotgun DNA microarray and stage-specific gene expression in Plasmodium falciparum malaria Mol Microbiol 2000 35 6 14 10632873 10.1046/j.1365-2958.2000.01730.x
Mamoun C Gluzman I Hott C MacMillan S Amarakone A Anderson D Carlton J Dame J Chakrabarti D Martin R Brownstein B Goldberg D Co-ordinated programme of gene expression during asewual intraerythrocytic development of the human malaria parasite Plasmodium falciparum revealed by microaaay analysis Mol Microbiol 2001 39 26 36 11123685 10.1046/j.1365-2958.2001.02222.x
Le Roch KG Zhou Y Blair PL Grainger M Moch K Haynes D De la Vega P Holder A Batalov S Carucci DJ Winzeler EA Discovery of gene function by expression profiling of the malaria parasite life cycle Science 2003 301 1503 1508 12893887 10.1126/science.1087025
Bozdech Z Llinas M Pulliam B Wong E Zhu J DeRisi JL The transcriptome of the intraerythrocytic developmental cycle of Plasmodium falciparum PLoS Biol 2003 1 E5 12929205 10.1371/journal.pbio.0000005
Florens L Washburn M Raine D Anthony R Graiger M Haynes D Moch J Muster N Sacii J Tabb D Witner A Wolters D Wu Y Garder M Holder A Sinden R Yates J Carucci D A proteomic view of the Plasmodium falciparum life cycle Nature 2002 419 520 526 12368866 10.1038/nature01107
Lasonder E Ishihama Y Andersen JS Vermunt AMW Pain A Sauerwein RW Eling WMC Hall N Waters A Stunnenberg HG Mann M Analysis of the Plasmodium falciparum proteome by high-accuracy mass spectrometry Nature 2002 419 537 542 12368870 10.1038/nature01111
Lanzer M Wertheimer S De Bruin D Ravetch JV Plasmodium: control of gene expression in malaria parasites J Exp Parasitol 1993 77 121 128 10.1006/expr.1993.1068
Horrocks P Dechering K Lanzer M Control of gene expression in Plasmodium falciparum Mol Biochem Parasitol 1998 95 171 181 9803410 10.1016/S0166-6851(98)00110-8
Hahn S Structure and mechanism of the RNA polymerase II transcription machinery Nat Struct Biol Mol Biol 2004 11 394 402 10.1038/nsmb763
Veenstra GJC Wolffe AP Gene-selective developmental roles of general transcription factors Trends Biochem Sci 2001 25 665 671 11701325 10.1016/S0968-0004(01)01970-3
McAndrew MB Read M Sims PF Hyde JE Characterization of the gene encoding an unusually divergent TATA-binding protein (TBP) from the extremely A+T-rich human malaria parasite Plasmodium falciparum Gene 1993 124 165 171 8444340 10.1016/0378-1119(93)90390-O
Hirtzlin J Farber PM Franklin RM Isolation of a novel Plasmodium falciparum gene encoding a protein homologous to the Tat-binding protein family Eur J Biochem 1994 226 673 680 8001584 10.1111/j.1432-1033.1994.tb20095.x
Fox BA Li WB Tanaka M Inselburg J Bzik DJ Molecular characterization of the largest subunit of Plasmodium falciparum RNA polymerase I Mol Biochem Parasitol 1993 61 37 38 8259131 10.1016/0166-6851(93)90156-R
Li WB Bzik DJ Gu HM Tanaka M Fox BA Inselburg J An enlarged largest subunit of Plasmodium falciparum RNA polymerase II defines conserved and variable RNA polymerase domains Nucleic Acids Res 1989 17 9621 9636 2690004
Li WB Bzik DJ Tanaka M Gu HM Fox BA Inselburg J Characterization of the gene encoding the largest subunit of Plasmodium falciparum RNA polymerase III Mol Biochem Parasitol 1991 46 229 239 1656254 10.1016/0166-6851(91)90047-A
Aravind L Iyer LM Wellems TE Miller LH Plasmodium biology: genomic gleanings Cell 2003 115 771 785 14697197 10.1016/S0092-8674(03)01023-7
Coulson RM Hall N Ouzounis CA Comparative genomics of transcriptional control in the human malaria parasite Plasmodium falciparum Genome Res 2004 14 1548 1554 15256513 10.1101/gr.2218604
Gardner MJ Shallom SJ Carlton JM Salzberg SL Nene V Shoaibi A Ciecko A Lynn J Rizzo M Weaver B Jarrahi B Brenner M Parvizi B Tallon L Moazzez A Granger D Fujii C Hansen C Pederson J Feldblyum T Peterson J Suh B Angiuoli S Pertea M Allen J Selengut J White O Cummings LM Smith HO Adams MD Venter JC Carucci DJ Hoffman SL Fraser CM Genome sequence of the human malaria parasite Plasmodium falciparum Nature 2002 419 498 511 12368864 10.1038/nature01097
Bahl A Brunk B Crabtree J Fraunholz MJ Gajria B Grant GR Ginsburg H Gupta D Kissinger JC Labo P Li L Mailman MD Milgram AJ Pearson DS Roos DS Schug J Stoeckert CJ JrWhetzel P PlasmoDB: the Plasmodium genome resource. A database integrating experimental and computational data Nucleic Acids Res 2003 31 212 215 12519984 10.1093/nar/gkg081
McConkey GA Pinney JW Westhead DR Plueckhahn K Fitzpatrick TB Macheroux P Kappes B Annotating the Plasmodium genome and the enigma of the shikimate pathway Trends Parasitol 2004 20 60 65 14747018 10.1016/j.pt.2003.11.001
Pizzi E Frontali C Low-complexity regions in Plasmodium falciparum proteins Genome Res 2001 11 218 229 11157785 10.1101/gr.GR-1522R
Chothia C Lesk AM The relation between the divergence of sequence and structure in proteins EMBO J 1986 5 823 826 3709526
Gaboriaud C Bissery V Benchetrit T Mornon JP Hydrophobic cluster analysis: an efficient new way to compare and analyse amino acid sequences FEBS Lett 1987 224 149 155 3678489 10.1016/0014-5793(87)80439-8
Callebaut I Labesse G Durand P Poupon A Canard L Chomilier J Henrissat B Mornon JP Deciphering protein sequence information through hydrophobic cluster analysis (HCA):current status and perspectives Cell Mol Life Sci 1997 53 621 645 9351466 10.1007/s000180050082
Woodcock S Mornon JP Henrissat B Detection of secondary structure elements in proteins by hydrophobic cluster analysis Protein Eng 1992 5 629 635 1480617
Hennetin J Le Tuan K Canard L Colloc'h N Mornon JP Callebaut I Non-intertwined binary patterns of hydrophobic/nonhydrophobic amino acids are considerably better markers of regular secondary structures than nonconstrained patterns Proteins 2003 51 236 244 12660992 10.1002/prot.10355
Soutoglou E Demeny MA Scheer E Fienga G Sassone-Corsi P Tora L The nuclear import of TAF10 is regulated by one of its three histone fold domain-containing interaction partners Mol Cell Biol 2005 25 4092 4104 15870280 10.1128/MCB.25.10.4092-4104.2005
Geiger JH Hahn S Lee S Sigler PB Crystal structure of the yeast TFIIA/TBP/DNA complex Science 1996 272 830 836 8629014
Tan S Hunziker Y Sargent DF Richmond TJ Crystal structure of a yeast TFIIA/TBP/DNA complex Nature 1996 381 127 151 8610010 10.1038/381127a0
Bleichenbacher M Tan S Richmond TJ Novel interactions between the components of human and yeast TFIIA/TBP/DNA complexes J Mol Biol 2003 332 783 793 12972251 10.1016/S0022-2836(03)00887-8
Upadhyaya AB Lee SH De Jong J Identification of a general transcription factor TFIIAa/b homolog selectively expressed in testis J Biol Chem 1999 274 18040 18048 10364255 10.1074/jbc.274.25.18040
Gasch A Hoffmann A Horikoshi M Roeder RG Chua NH Arabidopsis thaliana contains two genes for TFIID Nature 1990 346 390 394 2197561 10.1038/346390a0
Crowley TE Hoey T Liu JK Jan YN Jan LY Tjian R A new factor related to TATA-binding protein has highly restricted expression patterns in Drosophila Nature 1993 361 557 561 8429912 10.1038/361557a0
Wieczorek E Brand M Jacq X Tora L Function of TAF(II)-containing complex without TBP in transcription by RNA polymerase II Nature 1998 393 187 191 9603525 10.1038/30283
Chen B-S Hampsey M Transcription activation: unveiling the essential nature of TFIID Current Biol 2002 12 R620 R622 10.1016/S0960-9822(02)01134-X
Gangloff Y-G Romier C Thuault S Werten S Davidson I The histone fold is a key structural motif of transcription factor TFIID Trends Biochem Sci 2001 26 250 257 11295558 10.1016/S0968-0004(00)01741-2
Carlton JM Angiuoli SV Suh BB Kooij TW Pertea M Silva JC Ermolaeva MD Allen JE Selengut JD Koo HL Peterson JD Pop M Kosack DS Shumway MF Bidwell SL Shallom SJ van Aken SE Riedmuller SB Feldblyum TV Cho JK Quackenbush J Sedegah M Shoaibi A Cummings LM Florens L Yates JR Raine JD Sinden RE Harris MA Cunningham DA Preiser PR Bergman LW Vaidya AB van Lin LH Janse CJ Waters AP Smith HO White OR Salzberg SL Venter JC Fraser CM Hoffman SL Gardner MJ Carucci DJ Genome sequence and comparative analysis of the model rodent malaria parasite Plasmodium yoelii yoelii Nature 2002 419 512 519 12368865 10.1038/nature01099
Abrahamsen MS Templeton TJ Enomoto S Abrahante JE Zhu G Lancto CA Deng M Liu C Widmer G Tzipori S Buck GA Xu P Bankier AT Dear PH Konfortov BA Spriggs HF Iyer L Anantharaman V Aravind L Kapur V Complete genome sequence of the apicomplexan, Cryptosporidium parvum Science 2004 304 441 445 15044751 10.1126/science.1094786
Mizzen CA Yang XJ Kokubo T Brownell JE Bannister AJ Owen-Hughes T Workman J Wang L Berger SL Kouzarides T Nakatani Y Allis CD The TAF(II)250 subunit of TFIID has histone acetyltransferase activity Cell 1996 87 1261 1270 8980232 10.1016/S0092-8674(00)81821-8
Matangkasombut O Buratowski RM Swilling NW Buratowski S Bromodomain factor 1 corresponds to a missing piece of yeast TFIID Genes Dev 2000 14 951 962 10783167
Sullivan AS Aravind L Makalowska I Baxevanis AD Landsman D The Histone Database: a comprehensive WWW resource for histones and histone fold-containing proteins Nucleic Acids Res 2000 28 320 322 10592260 10.1093/nar/28.1.320
Hoffmann A Chiang CM Oelgeschlager T Xie X Burley SK Nakatani Y Roeder RG A histone octamer-like structure within TFIID Nature 1996 380 356 9 8598932 10.1038/380356a0
Albright SR Tjian R TAFs revisited: more data reveal new twists and confirm old ideas Genes 2000 242 1 13
Gangloff YG Sanders SL Romier C Kirschner D Weil PA Tora L Davidson I Histone folds mediate selective heterodimerization of yeast TAF(II)25 with TFIID components yTAF(II)47 and yTAF(II)65 and with SAGA component ySPT7 Mol Cell Biol 2001 21 1841 1853 11238921
Selleck W Howley R Fang Q Podolny V Fried MG Buratowski S Tan S A histone fold TAF octamer within the yeast TFIID transcriptional coactivator Nat Struct Biol 2001 8 695 700 11473260 10.1038/90408
Luger K Richmond TJ The histone tails of the nucleosome Curr Opin Genet Dev 1998 8 140 146 9610403 10.1016/S0959-437X(98)80134-2
Werten S Mitschler A Romier C Gangloff YG Thuault S Davidson I Moras D Crystal structure of a subcomplex of human transcription factor TFIID formed by TATA binding protein-associated factors hTAF4 (hTAF(II)135) and hTAF12 (hTAF(II)20) J Biol Chem 2002 277 45502 45509 12237304 10.1074/jbc.M206587200
Leurent C Sanders S Ruhlmann C Mallouh V Weil PA Kirschner DB Tora L Schultz P Mapping histone fold TAFs within yeast TFIID EMBO J 2002 21 3424 3433 12093743 10.1093/emboj/cdf342
Leurent C Sanders SL Demeny MA Garbett KA Ruhlmann C Weil PA Tora L Schultz P Mapping key functional sites within yeast TFIID EMBO J 2004 23 719 727 14765106 10.1038/sj.emboj.7600111
Sanders SL Klebanow ER Weil PA TAF25p, a non-histone-like subunit of TFIID and SAGA complexes, is essential for total mRNA gene transcription in vivo J Biol Chem 1999 274 18847 18850 10383379 10.1074/jbc.274.27.18847
Le Masson I Yu DY Jensen K Chevalier A Courbeyrette R Boulard Y Smith MM Mann C Yaf9, a novel NuA4 histone acetyltransferase subunit, is required for the cellular response to spindle stress in yeast Mol Cell Biol 2003 23 6086 6102 12917332 10.1128/MCB.23.17.6086-6102.2003
Ohkuma Y Sumimoto H Hoffmann A Shimasaki S Horikoshi M Roeder RG Structural motifs and potential sigma homologies in the large subunit of human general transcription factor TFIIE Nature 1991 354 398 401 1956403 10.1038/354398a0
Feaver WJ Henry NL Bushnell DA Sayre MH Brickner JH Gileadi O Kornberg RD Yeast TFIIE. Cloning, expression, and homology to vertebrate proteins J Biol Chem 1994 269 27549 27553 7961670
Maxon ME Goodrich JA Tjian R Transcription factor IIE binds preferentially to RNA polymerase IIa and recruits TFIIH: a model for promoter clearance Genes Dev 1994 8 515 524 7926747
Ohkuma Y Hashimoto S Wang CK Horikoshi M Roeder RG Analysis of the role of TFIIE in basal transcription and TFIIH-mediated carboxy-terminal domain phosphorylation through structure-function studies of TFIIE-alpha Mol Cell Biol 1995 15 4856 4866 7651404
Meinhart A Blobel J Cramer P An extended winged helix domain in general transcription factor E/IIE alpha J Biol Chem 2003 278 48267 48274 13679366 10.1074/jbc.M307874200
Okuda M Tanaka A Arai Y Satoh M Okamura H Nagadoi A Hanaoka F Ohkuma Y Nishimura Y A novel zinc finger structure in the large subunit of human general transcription factor TFIIE J Biol Chem 2004 279 51395 51403 15385556 10.1074/jbc.M404722200
Kelley LA MacCallum RM Sternberg MJ Enhanced genome annotation using structural profiles in the program 3D-PSSM J Mol Biol 2000 299 499 520 10860755 10.1006/jmbi.2000.3741
Kuldell NH Buratowski S Genetic analysis of the large subunit of yeast transcription factor IIE reveals two regions with distinct functions Mol Cell Biol 1997 17 5288 5298 9271406
Flores O Ha I Reinberg D Factors involved in specific transcription by mammalian RNA polymerase II. Purification and subunit composition of transcription factor IIF J Biol Chem 1990 265 5629 5634 2180931
Woychik NA Hampsey M The RNA polymerase II machinery: structure illuminates function Cell 2002 108 453 463 11909517 10.1016/S0092-8674(02)00646-3
Fang SM Burton ZF RNA polymerase II-associated protein (RAP) 74 binds transcription factor (TF) IIB and blocks TFIIB-RAP30 binding J Biol Chem 1996 271 11703 11709 8662660 10.1074/jbc.271.20.11703
Groft CM Uljon SN Wang R Werner MH Structural homology between the Rap30 DNA-binding domain and linker histone H5: implications for preinitiation complex assembly Proc Natl Acad Sci U S A 1998 95 9117 9122 9689043 10.1073/pnas.95.16.9117
Gaiser F Tan S Richmond TJ Novel dimerization fold of RAP30/RAP74 in human TFIIF at 1.7 Å resolution J Mol Biol 2000 302 1119 1127 11183778 10.1006/jmbi.2000.4110
Tan S Garrett KP Conaway RC Conaway JW Cryptic DNA-binding domain in the C terminus of RNA polymerase II general transcription factor RAP30 Proc Natl Acad Sci U S A 1994 91 9808 9812 7937895
Kamada K De Angelis J Roeder RG Burley SK Crystal structure of the C-terminal domain of the RAP74 subunit of human transcription factor IIF Proc Natl Acad Sci U S A 2001 98 3115 3120 11248041 10.1073/pnas.051631098
Schultz P Fribourg S Poterszman A Mallouh V Moras D Egly JM Molecular structure of TFIIH Cell 2000 102 599 607 11007478 10.1016/S0092-8674(00)00082-9
Chang WH Kornberg RD Electron crystal structure of the transcription factor and DNA repair complex, core TFIIH Cell 2000 102 609 613 11007479 10.1016/S0092-8674(00)00083-0
Zurita M Merino C The transcriptional complexity of the TFIIH complex Trends Genet 2003 19 578 584 14550632 10.1016/j.tig.2003.08.005
Ranish JA Hahn S Lu Y Yi EC Li XJ Eng J Aebersold R Identification of TFB5, a new component of general transcription and DNA repair factor IIH Nat Genet 2004 36 707 713 15220919 10.1038/ng1385
Giglia-Mari G Coin F Ranish JA Hoogstraten D Theil A Wijgers N Jaspers NG Raams A Argentini M van der Spek PJ Botta E Stefanini M Egly JM Aebersold R Hoeijmakers JH Vermeulen W A new, tenth subunit of TFIIH is responsible for the DNA repair syndrome trichothiodystrophy group A Nat Genet 2004 36 714 719 15220921 10.1038/ng1387
Edwards MC Wong C Elledge SJ Human cyclin K, a novel RNA polymerase II-associated cyclin possessing both carboxy-terminal domain kinase and Cdk-activating kinase activity Mol Cell Biol 1998 18 4291 4300 9632813
Doerks T Huber S Buchner E Bork P BSD: a novel domain in transcription factors and synapse-associated proteins Trends Biochem Sci 2002 27 168 170 11943536 10.1016/S0968-0004(01)02042-4
Gervais V Lamour V Jawhari A Frindel F Wasielewski E Dubaele S Egly JM Thierry JC Kieffer B Poterszman A TFIIH contains a PH domain involved in DNA nucleotide excision repair Nat Struct Mol Biol 2004 11 616 622 15195146 10.1038/nsmb782
Coulson RM Ouzounis CA The phylogenic diversity of eukaryotic transcription Nucleic Acids Res 2003 31 653 660 12527774 10.1093/nar/gkg156
Bastien O Lespinats S Roy S Metayer K Fertil B Codani JJ Marechal E Analysis of the compositional biases in Plasmodium falciparum genome and proteome using Arabidopsis thaliana as a reference Gene 2004 336 163 173 15246528 10.1016/j.gene.2004.04.029
Soyer A Chomilier J Mornon JP Jullien R Sadoc JF Voronoi tessellation reveals the condensed matter character of folded proteins Phys Rev Lett 2000 85 3532 3535 11030939 10.1103/PhysRevLett.85.3532
Pintar A Carugo O Pongor S Atom depth in protein structure and function Trends Biochem Sci 2003 28 593 597 14607089 10.1016/j.tibs.2003.09.004
Pintar A Carugo O Pongor S Atom depth as a descriptor of the protein interior Biophys J 2003 84 2553 2561 12668463
Andel F Ladurner AG Inouye C Tjian R Nogales E Three-dimensional structure of the human TFIID-IIA-IIB complex Science 1999 286 2153 2156 10591646 10.1126/science.286.5447.2153
Brand M Leurent C Mallouh V Tora L Schultz P Three-dimensional structures of TAFII-containing complexes TFIID and TFTC Science 1999 286 2151 2153 10591645 10.1126/science.286.5447.2151
Altschul SF Madden TL Schaffer AA Zhang J Zhang Z Miller W Lipman DJ Gapped-BLAST and PSI-BLAST : a new generation of protein database search programs Nucl Acids Res 1997 25 3389 3402 9254694 10.1093/nar/25.17.3389
Kissinger JC Brunk BP Crabtree J Fraunholz MJ Gajria B Milgram AJ Pearson DS Schug J Bahl A Diskin S JrGinsburg H Grant GR Gupta D Labo P Li L Mailman MD McWeeney SK Whetzel P Stoeckert CJ Roos DS The Plasmodium genome database Nature 2002 419 490 492 12368860 10.1038/419490a
Bateman A Coin L Durbin R Finn RD Hollich V Griffiths-Jones S Khanna A Marshall M Moxon S Sonnhammer EL Studholme DJ Yeats C Eddy SR The Pfam protein families database Nucleic Acids Res 2004 32 D138 141 14681378 10.1093/nar/gkh121
Letunic I Copley RR Schmidt S Ciccarelli FD Doerks T Schultz J Ponting CP Bork P SMART 4.0: towards genomic data integration Nucleic Acids Res 2004 32 D142 144 14681379 10.1093/nar/gkh088
Marchler-Bauer A Anderson JB Cherukuri PF DeWeese-Scott C Geer LY Gwadz M He S Hurwitz DI Jackson JD Ke Z Lanczycki CJ Liebert CA Liu C Lu F Marchler GH Mullokandov M Shoemaker BA Simonyan V Song JS Thiessen PA Yamashita RA Yin JJ Zhang D Bryant SH CDD: a Conserved Domain Database for protein classification Nucleic Acids Res 2005 33 D192 D196 15608175 10.1093/nar/gki069
Callebaut I Mornon JP From BRCA1 to RAP1 : a widespread BRCT module closely associated with DNA repair FEBS Letters 1997 400 25 30 9000507 10.1016/S0014-5793(96)01312-9
Callebaut I Mornon JP OCRE: a novel domain made of imperfect, aromatic-rich octamer repeats Bioinformatics 2005 21 699 702 15486042 10.1093/bioinformatics/bti065
Callebaut I Mornon JP The V(D)J recombination activating protein RAG2 consists of a six-bladed propeller and a PHD fingerlike domain, as revealed by sequence analysis Cell Mol Life Sci 1998 54 880 891 9760994 10.1007/s000180050216
Girault JA Labesse G Mornon JP Callebaut I The N- termini of FAK and JAKs contains divergent band 4.1 domains Trends Biochem Sci 1999 24 54 57 10098398 10.1016/S0968-0004(98)01331-0
Henrissat B Callebaut I Fabrega S Lehn P Mornon JP Davies G Conserved catalytic machinery and the prediction of a common fold for several families of glycoside hydrolases Proc Natl Acad Sci USA 1995 92 7090 7094 7624375
Callebaut I Moshous D Mornon JP de Villartay JP Metallo-beta-lactamase fold within nucleic acids processing enzymes: the beta-CASP family Nucleic Acids Res 2002 30 3592 3601 12177301 10.1093/nar/gkf470
Callebaut I Curcio-Morelli C Mornon JP Gereben B Buettner C Huang S Castro B Fonseca TL Harney JW Larsen PR Bianco AC The iodothyronine selenodeiodinases are thioredoxin-fold family proteins containing a glycoside hydrolase-clan GH-A-like structure J Biol Chem 2003 276 36887 36896 12847093 10.1074/jbc.M305725200
Guermah M Ge K Chiang CM Roeder RG The TBN protein, which is essential for early embryonic mouse development, is an inducible TAFII implicated in adipogenesis Mol Cell 2003 12 991 1001 14580349 10.1016/S1097-2765(03)00396-4
|
16042788
|
PMC1199594
|
CC BY
|
2021-01-04 16:32:48
|
no
|
BMC Genomics. 2005 Jul 23; 6:100
|
utf-8
|
BMC Genomics
| 2,005 |
10.1186/1471-2164-6-100
|
oa_comm
|
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-1041608350310.1186/1471-2164-6-104Research ArticleA systematic search for new mammalian noncoding RNAs indicates little conserved intergenic transcription Babak Tomas [email protected] Benjamin J [email protected] Timothy R [email protected] Banting and Best Department of Medical Research, 112 College St., Toronto, ON M5G 1L6 Canada2 Department of Medical Genetics and Microbiology, 10 King's College Circle, Toronto, ON M1R 4F9 Canada2005 5 8 2005 6 104 104 22 7 2005 5 8 2005 Copyright © 2005 Babak et al; licensee BioMed Central Ltd.2005Babak et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Systematic identification and functional characterization of novel types of noncoding (nc)RNA in genomes is more difficult than it is for protein coding mRNAs, since ncRNAs typically do not possess sequence features such as splicing or translation signals, or long open reading frames. Recent "tiling" microarray studies have reported that a surprisingly larger proportion of mammalian genomes is transcribed than was previously anticipated. However, these non-genic transcripts often appear to be low in abundance, and their functional significance is not known.
Results
To systematically search for functional ncRNAs, we designed microarrays to detect 3,478 intergenic and intronic sequences that are conserved between the human, mouse, and rat genomes, and that score highly by other criteria that characterize ncRNAs. We probed these arrays with total RNA isolated from 16 wild-type mouse tissues. Among 55 candidates for highly-expressed novel ncRNAs tested by northern blotting, eight were confirmed as small, highly-and ubiquitously-expressed RNAs in mouse. Of the eight, five were also detected in rat tissues, but none were detected at appreciable levels in human tissues or cultured cells.
Conclusion
Since the sequence and expression of most known coding transcripts and functional ncRNAs is conserved between human and mouse, the lack of northern-detectable expression in human cells and tissues of the novel mouse and rat ncRNAs that we identified suggests that they are not functional or possibly have rodent-specific functions. Our results confirm that relatively little of the intergenic sequence conserved between human, mouse and rat is transcribed at high levels in mammalian tissues, possibly suggesting a limited role for transcribed intergenic and intronic sequences as independent functional elements.
==== Body
Background
Comparative genomics has revealed that approximately 5% of the mammalian genome is under purifying selection [1,2]. While exons make up roughly 1.5% of the genome [3], relatively little is known about the role of the remaining 3.5% of the highly conserved genomic regions, and even less about the functional potential of evolutionarily-diverged intergenic sequences. Large-scale microarray tiling analyses (i.e. using a set of probes designed to detect all or most of a targeted genome or genomic region), as well as high-throughput cDNA sequencing efforts, have indicated that the "transcriptome" is significantly larger than was previously appreciated, although the functional significance of the vast majority of the novel, apparently noncoding (nc) transcripts detected by these approaches has remained elusive [4-8]. To date, several studies have reported large-scale tiling efforts of the human genome [4,5,9,10]. In all cases a significantly higher proportion of transcribed sequence was reported than could be accounted for by existing exon annotation data, and much of the remainder did not appear to encode protein [4]. Comparison of datasets suggests that a high proportion of the novel transcripts are specific to tissues or cell lines [4,9]. This trend was particularly evident for cell lines, where novel cell-line specific transcripts were even more numerous than annotated cell-line specific exons [4], implying that many of these transcripts may not have endogenous functions in whole organisms. Further supporting this possibility was the observation that the majority of the novel transcripts were detected at very low levels [5].
A second source of evidence for a more extensive transcriptome arises from large-scale cDNA compilation efforts. The mouse cDNA sequencing effort led by the RIKEN Consortium identified 60,770 unique cDNA transcripts from a variety of mouse tissues and cell lines [7]. Approximately half (33,409 sequences) were derived from unique genomic locations (Transcriptional Units), of which 15,815 did not map to known or predicted coding genes in mouse [7]. Further refinement identified a set of 4,280 mRNA-like noncoding RNAs which had no homology to any known protein sequences and comprised of sequences mapped to regions located between predicted exon boundaries [6]. Many of these sequences were reported elsewhere in EST databases and displayed features of polymerase II transcripts [6]. However, unlike protein-coding mouse genes, of which 99% have homologs in the human genome [1], only 10.6% of the 4,280 apparent nc transcripts were represented by homologous sequences in the human genome [6,7]. In fact, Wang et al. [11] demonstrated that most of these transcripts are no more conserved than intergenic sequence in general, and less conserved than a comprehensive set of 321 known ncRNAs with established functional roles. In addition, expression profiling of a different but overlapping (FANTOM1) subset of cDNAs that do not map to known ESTs or protein sequences (3,388), revealed that most transcripts in this uncharacterized class were present at low abundance [12]. Collectively, these results demonstrate the transcription of uncharacterized sequence, but raise questions about the functional relevance of the novel "noncoding" set.
One possible explanation for the observed low-level expression of a much larger fraction of genomes than can be accounted for by known genes comes from the recent discovery of a nuclear posttranscriptional quality-control pathway that degrades "cryptic unstable transcripts" (CUTs) in yeast [13]. CUTs are transcribed by Pol II and are detectable by both microarrays and RT-PCR in wild-type yeast, and also appear to be frequently represented as single tags in SAGE libraries, but are undetectable by Northern blotting and do not contain significant open reading frames. However, in mutants in the quality-control pathway, they appear as a smear on Northern blots due to the fact that they have heterogeneous 3' ends [13]. The fact that a posttranscriptional quality control exists to prevent accumulation of CUTs suggests that they are aberrant and predicts that there should be little selection pressure on their expression. Moreover, these observations suggest that nonfunctional transcripts might be distinguished from bona fide functional transcripts on the basis of formation of a discrete species on Northern blotting, and by conservation of expression among different organisms.
In this study, we describe a systematic approach to predict and screen novel ncRNA transcripts in the mouse genome. We first identified non-exonic sequences that are most likely to encode functional ncRNAs (functional transcripts that do not encode proteins) by using the program QRNA, which searches for conserved regions with compensatory mutation patterns that are consistent with the evolutionary conservation of secondary structure in functional noncoding sequences [14]. These are hallmarks of most known functional ncRNAs, and QRNA has been used successfully to identify novel structural ncRNAs in E. coli [15] and S. cerevisiae [16]. However, even in these organisms, which have relatively compact genomes, a high false-positive prediction rate was observed [16], which presents a challenge for screening large genomes. We therefore used a custom oligonucleotide microarray [17] as an initial high-throughput screen for expression. We then tested the 55 highest-expressed candidates to ask whether they are detectable as discrete species on Northern blots. We report eight novel mouse transcripts identified using this approach. However, none of the eight appears to be expressed in humans, casting doubt on their role as independent functional elements. Taken together with the low proportion of intergenic sequences that we detected, our results suggest that much of the recently-discovered expanded transcriptome [4-7,9,10] may correspond to cryptic transcripts [13], suggesting a limited role for transcribed intergenic and intronic sequences as independent functional elements.
Results
Predicting novel ncRNA candidates
We identified novel structural ncRNA candidates on the basis of two features: high sequence conservation and a mutation pattern consistent with sequences being under selective pressure to maintain a conserved secondary structure [see Methods for details]. Figure 1 outlines the computational screen we employed for finding novel ncRNAs. We obtained human-mouse pairwise sequence alignments from UCSC [18] and subset them to alignments with a minimum of 85% sequence identity using a 200 nt scanning window. This eliminated sequences that are unlikely to be under evolutionary selection [1,11] and reduced the dataset to a computationally manageable size. We used QRNA v.1.1 [14-16] to screen the alignments for putative ncRNAs. This generated 106,320 predicted ncRNAs. We then removed redundant sequences and predicted ncRNAs with sequences similar or identical to coding genes annotated in the Mouse RefSeq mRNA database [19], RIKEN cDNA [7], Mouse Protein NR database [20], or coding genes in other organisms annotated in GenBank [21]. The remaining 36,756 predicted ncRNAs were sorted by logistic regression using four parameters that we identified to be useful for distinguishing known ncRNAs from QRNA predictions of putative new ncRNAs: 1) the QRNA logodds RNA score; 2) the thermodynamic stability of the predicted secondary structure of each prediction; 3) a genomic clustering score of closely mapped predictions on the genome (presumed to be multiple regions of a longer ncRNA transcript); and 4) an overlap between mouse-human and mouse-rat QRNA predictions (processed similarly to mouse-human alignments). The QRNA score was the strongest indicator of defined ncRNAs (data not shown), but combining the additional parameters increased the sorting power, especially for the top 10% of the predicted RNAs (Fig. 1B). Other parameters such as GC content and length of the sequences did not improve the sorting (data not shown). To further characterize the set of predicted ncRNAs, we screened them computationally for tRNAs and snoRNAs [22,23] and searched for similar sequences in the RIKEN FANTOM2 mRNA-like ncRNA collection. For a summary of these features, see Additional file 1.
Figure 1 Summary of computational prediction and sorting of novel mammalian noncoding RNA. (a) Whole genome alignments were downloaded from UCSC [18], subset to regions of greater than 85% sequence identity, and analyzed with QRNA [14]. Removal of redundant and coding sequences left 36,756 ncRNA candidates containing 62 known ncRNAs (of approximately 400 known ncRNAs). Candidates were then sorted and the top 3,478 predicted RNAs, which contained 38 known ncRNAs (representing a 1,700 fold enrichment of real ncRNAs over random selection from genomic sequence), were selected for further screening. (b) Sorting was based on the QRNA score, stability (measured as a predicted free energy change using Mfold [44]), overlap with rat-mouse QRNA predictions, and genomic clustering (many predictions close to one another are likely the same transcript). This combination of four criteria was more powerful in identifying real ncRNAs than using the QRNA score alone. *UCSC [18], **See (b).
Array design
Due to the generic nature of the algorithm, QRNA has a high false positive rate, much higher than coding gene-finding algorithms, thus experimental validation is essential. Using the prediction scheme and sorting criteria described above, we designed a microarray to detect 3,478 QRNA predictions with properties most indicative of ncRNAs. The microarray contained probes for the top 9.5% of the total ncRNA predictions, and probes for 38 known ncRNAs [see Additional file 2]. The design included 20,867 complementary DNA probes to the QRNA predictions (with six probes per prediction; three for each orientation), 200 random probe sequences and 305 intergenic probes that served as negative controls, and 705 positive control probes tiled across U4, U5 and mature rRNA transcripts. A list of all microarray probes is included in Additional file 3.
Analysis of RNA from diverse mouse tissues
Since ncRNAs can be expressed in a tissue-or developmental-stage-specific manner [24-29] we screened 16 mouse tissues/organs encompassing a variety of tissue sources, including two embryonic stages of development (Table 1). The intensity distribution over all measurements is shown in Figure 2. As expected, the majority of QRNA-prediction intensity measurements overlapped with the negative control probe measurements, presumably due to the high false-positive rate inherent in generating the predictions. However, the distribution was skewed to the right tail of the plot (i.e. higher intensity), and twice as many of measurements from the predicted ncRNAs were above the 99% negative control threshold (generated using random sequence probes) than would be expected based on a random distribution (Fig. 2). Of the 38 known ncRNAs that were among the QRNA predicted RNAs [see Additional file 2], we detected 15, including several snoRNAs, tRNAs, a Hox antisense transcript, and miRNAs, using the same intensity cutoff used for selecting novel candidates (see below). This illustrates that the sensitivity of this technique is sufficient to detect most known ncRNA types. We have also used this technique to survey miRNA expression [24].
Table 1 List of tissues used in microarray expression analysis
Tissues/Organs/Cells screened for novel noncoding RNAs
Bladder
Brain
Embryonic Stem Cells
Femur
Heart
Intestine
Liver
Lung
Mammary Gland
Muscle
Stomach
Teeth
Testis
12.5-day Embryo
15-day Embryo
9.5-day Placenta
Figure 2 Screening, selection, and confirmation of novel ncRNA predictions. (a) Total RNA from 16 tissues was hybridized to custom Agilent microarrays containing probes to the QRNA predictions. (b) Most intensity measurements overlapped negative-control intensities (both random probe sequences and probes corresponding to randomly sampled intergenic regions), although a right-tailed distribution overlapping rRNA levels indicated detection of potential abundant novel transcripts. (c) Expanded y-axis region from (b), axes denote absolute probe intensity counts. (d) Sample schematic of microarray spots corresponding to a transcript that was tested further by northern analysis. (e) 55 transcripts in total were screened by Northern analysis.
Validation by northern blotting
Although our microarray data indicated that many of the measurements arise from real transcripts, noise (e.g. spurious cross-hybridization) could also account for some proportion of the high-intensity measurements. Furthermore, microarray results cannot differentiate between a single RNA species and a heterogeneous population. We therefore used Northern blotting to validate our candidate novel ncRNAs. Northern blotting is more quantitative than RT-PCR, since there is no exponential amplification step. It is less sensitive for the same reason; however, using our methods, we have been able to detect all types of transcripts including structural ncRNAs, miRNAs, and mRNAs ([24,30] and data not shown). Importantly, since Northern blotting reveals the size of the RNA species detected, it can distinguish whether there is a single RNA product species and a heterogeneous transcript population. We tested all predicted ncRNAs detected by at least two of three probes (all in the same orientation) displaying signals above the 99%-negative control intensity threshold in at least one tissue. In total, this included 55 novel predicted ncRNAs, of which most appeared to be ubiquitously expressed. Northern analysis on this subset confirmed 8 novel transcripts (Fig. 3), all of which were detected ubiquitously in total RNA isolated from 16 wild-type mouse tissues. All eight transcripts were between 70 and 140 nt in length, none had tRNA or snoRNA structural or sequence characteristics, and five were located in intronic regions. It is possible that additional RNAs are expressed at low levels that are detectable by microarray and/or RT-PCR but not by Northern blotting, especially if they are heterogeneous in length [13]. We did not pursue this possibility, since it seemed that transcripts undetectable by Northern blotting are less likely to represent bona fide ncRNAs.
Figure 3 Abundantly and ubiquitously expressed novel mouse transcripts are not expressed in human tissues or cells. (a) Northern analysis of mouse transcripts using mouse-specific probes. U5 and U4 probes were used as loading controls as indicated and were co-hybridized with test probes. (b) No signal was detected in Northern analyses using human-specific probes. Human tissues were analyzed with a longer exposure (right panel) since short RNAs were slightly underrepresented in these commercially-obtained samples. Mouse-, and human-specific probe sequences complementary to the novel ncRNA predictions and images of all full-scale northern blots are available in Additional files 4 and 5.
Expression of novel mouse transcripts is not conserved in human cells/tissues
Nearly all sequence-conserved coding genes between human and mouse have a conserved expression pattern across multiple tissues [31]. Although they have not to our knowledge been comprehensively analyzed, ncRNAs are also generally expressed similarly across related species [32] and since most are required for cell proliferation, they tend to be expressed in all tissues, as were all eight of the novel transcripts we observed. However, Northern blotting revealed that none of the eight novel mouse transcripts were expressed at detectable levels in HeLa cells or in human tissues (Fig. 3). Moreover, only five of these were detected in rat (Fig. 3). Images of all full-scale Northern blots shown in Figure 3 and other supporting Northern data is available in the supplementary data [see Additional file 4].
We also compared our 3,478 QRNA predictions and the RNAs we verified with recently published high-density human tiling data from Cheng et al. [9]. We subset our QRNA predictions to regions surveyed by Cheng et al. [9] (i.e. the non-repetitive regions of human chromosomes 6, 7, 13, 14, 19, 20, 21, 22, X, and Y). We considered the "transfrags" (i.e. transcribed fragments: any transcribed genomic region) described by Cheng et al. [9] in poly-A-minus RNA from HepG2 cells, which should be most comparable to our data (Fig. 4A). We confirmed that the Cheng et al. [9] "transfrags" encompassed a larger number of known noncoding RNAs and mRNA exons than was obtained from random positioning of sequences of the same length (Fig. 4B), although the vast majority of "transfrags" do not overlap any annotated features. We did not see a marked difference in the overlap between the Cheng et al. "transfrags" and our QRNA predictions (Fig. 4B). This indicates that the "transfrags" are not enriched for conserved sequence with conserved secondary structure, consistent with our data showing a lack of conserved expression of our northern-confirmed QRNA transcripts in human tissues and cells. Only one of the eight northern-confirmed novel mouse transcripts we verified mapped to the regions surveyed by Cheng et al. [9] and it did not overlap a "transfrag", also consistent with our results (Fig. 3).
Figure 4 Overlap between known and predicted ncRNA types. (a) 3,478 QRNA predictions, 390 mouse annotated ncRNAs, and 716 human ribosomal exons were mapped to the human genome (hg17) regions surveyed by Cheng et al. [9] using poly-A minus RNA derived from HepG2 cells. 833 QRNA predictions, 151 mouse ncRNAs, and 260 exons are located in regions surveyed by Cheng et al. (b) Shown are distributions of the percentage of overlap of each sequence with a transfrag for actual genomic positions versus randomly positioned sequences in regions surveyed by Cheng et al. [9]. (c) Schematic of how an overlap was calculated for one QRNA prediction that overlaps a transfrag. This was repeated for all overlaps and the distributions are shown in (b).
Discussion
Using comparative genomics and an established ncRNA search method modified for high-throughput screening, we report eight novel mouse ncRNA transcripts that are all relatively short, ubiquitously expressed, and abundant. Despite their sequence and secondary structure conservation, none of the transcripts were expressed at detectable levels in human cells and tissues.
Given the large search space incorporated in this analysis, our results indicate that little intergenic or intronic sequence is expressed as distinct, stable transcripts at levels comparable to the expression of most known functional RNAs. This deduction contrasts somewhat with the conclusions of recent studies employing tiling arrays or large-scale cDNA sequencing [5,7,9,10]. Because results from other studies were obtained and validated in different ways, we cannot confirm or refute the basic observations of any other study. Numerous explanations exist for the breadth of the emerging transcriptome [10]. Nonetheless, we propose that claims regarding a dramatically larger transcriptome than is accounted for by current annotations should be addressed with scrutiny, particularly with regard to functional potential. Several factors indicate that a significant proportion of the newly measured transcripts may either be spurious or non-functional: 1) transcriptionally active regions identified in tiling analyses and potentially noncoding cDNAs are generally detected in low abundance [4,6]; 2) in yeast, many intergenic regions are also transcribed at low levels, apparently as heterogeneous species, and there is a specific mechanism for degrading these transcripts [13]; 3) much of the mammalian data available is from cell lines, including a high proportion of tumor-derived cell lines [5,9], which may lack the same degree of quality-control as found in normal cells and tissues; 4) most are relatively short (i.e. sequenced transcripts are shorter than the average coding gene [7], as are transcripts identified from tiling, which are on average less than 200 nt [4]); 5) potentially noncoding cDNAs correspond to regions not conserved at the sequence level [6,7]) and have evolved at a non-selective rate [11]; 6) there is little evidence for cross-species expression (only 2.6% of noncoding mouse cDNAs can be mapped to human ESTs [6,7]); 7) 70% of intergenic "transfrags" corresponding to novel transcribed regions could not be detected by northern analysis [4]); 8) the "transfrags" do not appear to be enriched in sequences with conserved secondary structures (Fig. 4), which is a hallmark feature of known structural ncRNAs [33].
How can we distinguish bona fide functional transcripts, in the absence of directed genetic experimentation? Sequence conservation alone is apparently not sufficient to distinguish sequences with critical functions, as large-scale non-genic deletions encompassing highly conserved regions can be tolerated in mice without detectable fitness disadvantages [34]. The presence of conserved expression improves the likelihood that sequence-conserved regions are functional since most characterized RNA classes, including coding mRNA, and noncoding rRNA, tRNA, snRNA, snoRNA, and miRNA, generally exhibit conserved expression patterns across evolutionarily-related species [31,32,35,36]. Of the few characterized mRNA-like ncRNAs, some also have conserved expression patterns [29,32,37]. The lack of conserved expression of the eight transcripts identified in our study, despite a high sequence and structural similarity, suggests that they are not functional, although it is possible that a subset of functional transcripts have species-or lineage-specific functions despite their high degree of sequence conservation. For example, a subset of the ncRNAs detected in mouse in our study were also detected in rat tissues, suggesting the possibility of conserved functions restricted to the rodent lineage. However, in reported cases of mouse-specific ncRNAs, such as BC1, Tsix, CIOR, and t-ncb [32], the RNAs were not conserved at the sequence level which is likely the reason for mouse-specific ncRNA differential regulation.
More cross-species and non-biased expression data is required to definitively address the likelihood of functionality of emerging transcriptomes. The most comprehensive approach will likely be an extension of whole genome tiling microarray analyses [38] using RNA derived from endogenous tissues from a variety of organisms. The approach of hybridizing covalently labeled total RNA (applied in this paper), as opposed to cRNA or cDNA derived from poly-adenylated RNA, presents a potential improvement to the unbiased nature of tiling analyses, since there is no amplification bias and strand information is retained. An added dimension of conserved expression will enable focused functional experimentation on transcripts that are likely to be important, although our data indicates that these cases will be the exception rather than the rule.
Conclusion
With the application of high-throughput transcriptional analyses it has been reported that more sequence is transcribed than was previously appreciated, with some estimates exceeding twice that of currently annotated transcripts. In a systematic search for sequence-conserved transcripts with hallmarks of structural ncRNAs, we identified only eight novel mouse transcripts with ubiquitous and abundant expression. This indicates that very little intergenic sequence is transcribed at high levels. Furthermore, despite meeting the stringent requirements of characterized ncRNAs, none of these eight transcripts were expressed at detectable levels in human cells or tissues. This suggests that these transcripts are unlikely to have conserved functional roles. We propose that newly-identified transcriptomes should be viewed with scrutiny, particularly with regard to function, until it is determined that they are functional or at least display properties of known functional elements.
Methods
Predicting novel ncRNAs
Whole genome pairwise human-mouse alignments were downloaded from the UCSC Genome Bioinformatics website ([18]; build 32, Nov. 2003). Repeat-masked [39] alignments were subset to segments with a minimum of 85% sequence identity. QRNA v.1.1 [14] was used to score the alignments for noncoding RNA potential using settings determined to work optimally on a test set of ncRNAs embedded in random alignments of equivalent sequence identity (default settings with -w 100 -x 50). The processing time was approximately 14 days on a 20-processor (1 GHz) linux cluster. Overlapping sequences with a positive logodds RNA score were concatenated into one sequence which was assigned the highest score of the original component sequences.
Selection of candidate ncRNAs
QRNA predictions were filtered by alignment to a variety of coding sequence databases using BLAT [40] with a default score cutoff of 30 ([alignment length] – [number of mismatches]) and a minimum sequence identity of 60% (-minIdentity = 60). The databases included: Mouse RefSeq mRNA [19] (build 29), ENSEMBL coding transcripts [41], RIKEN cDNA [7], ESTs [42] (download date: Nov. 2003), Unigene [43] (Nov. 2003), Protein NR [20] (build 29), and Genbank NT Database [21] (Nov. 2003). Redundant QRNA predictions and predictions that aligned to annotated coding sequences were removed. The remaining set was screened for tRNAs and Box C/D snoRNAs using tRNAScan SE [22] and snoscan [23] respectively. Sequences were extended by 100 bases in both directions from the genome to ensure complete coverage of potential tRNAs or snoRNAs.
QRNA predictions that did not map to annotated coding genes were sorted on a combination of criteria to maximize selection of known ncRNAs (as compiled in [24]). Sorting parameters included the minimum free energy as predicted by Mfold [44], overlap between mouse-human and mouse-rat QRNA predictions (blast, e-threshold 10-4), and proximity to adjacent predictions in the genome. Genomic proximity was scored by adding the number of QRNA predictions within 1000 bp of each other in the genome. Mouse-rat alignments [18] were processed identically to mouse-human alignments. Multiple linear regression was used to assign weights to these parameters in addition to the QRNA logodds RNA score and were subsequently used to calculate an overall score for each QRNA prediction. The top 3,478 predicted RNAs (limited by space on the array) were analyzed further by microarray. These contained 38 known ncRNAs of the approximately 400 known ncRNAs, representing a 1,700-fold enrichment. The level of enrichment was calculated as the ratio of the proportion of nucleotides that are real ncRNAs in the QRNA predicted set to the proportion of nucleotides of all known ncRNAs of the mouse genome (i.e. how much more likely one could select a nucleotide belonging to a known ncRNA in the QRNA set over the whole genome).
Microarray design
Six probe sequences were allotted for each ncRNA prediction; three for each orientation. Complementary DNA probes were designed to maximize spatial coverage of each predicted sequence and were normalized by length (i.e. probe lengths were adjusted) to a uniform melting temperature of 60°C. Probe sequences were on average 26.9 nt and were concatenated to 60 nucleotides. Probe sequences were submitted to Agilent Technologies for microarray production (Palo Alto, California). The designs included 200 60-mer probes containing random sequences, which were used as negative controls, and 696 positive control probes tiled across U4 and U5 snRNAs and 18S and 28S rRNAs. Additional file 3 contains a list of all of the probe sequences.
RNA extraction, labeling, and hybridizations
HeLa nuclear extract (NE) was prepared as described previously [45]. Total RNA from HeLa cells, HeLa NE, and mouse tissues was extracted using Trizol (Invitrogen) according to the manufacturer's instructions and was treated with DNase I (Fermentas). Total RNA derived from human tissues was purchased from Clontech (BD Biosciences, Mississauga, ON) and Ambion (Austin, TX). Integrity of rRNA was confirmed on 1% agarose-formaldehyde gels. 7 μg of total RNA was chemically labeled with Ulysis Alexa Fluor 546 or Ulysis Alexa Fluor 647 (Ulysis) according to manufacturer's instructions. This protocol labels G residues [46], and there were no predicted RNAs that lacked G residues. Samples were resuspended in 0.5 mL of hybridization buffer (1 M NaCl, 0.5% sodium sarcosine, 50 mM N-morpholino ethane sulfonate, pH 6.5, 33% formamide and 40 μg salmon sperm DNA), denatured by heating at 65°C for 5 minutes, and snap-cooled on ice prior to hybridization. Hybridizations were carried out for 16–24 h at 42°C in a rotating hyb oven. Slides were then washed (rocking ~30 seconds in 6× SSPE, 0.005% sarcosine, then rocking ~30 seconds in 0.06× SSPE) and scanned with a 4000A microarray scanner (Axon Instruments, Union City, CA).
Microarray data processing and normalization
TIFF images were quantified with GenePix 3.0 (Axon Instruments, Union City, CA). Individual channels were spatially detrended (i.e. overall correlations between spot intensity and position on the slide removed) by high-pass filtering [47] using 5% outliers. The 16 individual channels were then normalized using Variance Stabilization [48,49] and transformed to arcsinh values, which are similar to natural log values but are tolerant of negative numbers emerging from high-pass filtering.
Northern blotting
7 μg of total RNA from each tissue was separated on 10% polyacrylamide/TBE/urea gels, and electroblotted to Hybond N+ or Hybond-XL membranes (Amersham) using asemi-dry transfer apparatus(Bio-Rad) in 0.5X TBE according to the manufacturer's instructions. The membranes were UV cross-linked using a Stratalinker (Stratagene), hybridized overnight at 42°C in Church buffer with 5'-32P-end-labeled oligonucleotide probes, and washed with 2X SSC, 0.1% SDS and 0.1X SSC, 0.1% SDS for 5 minutes each at 42°C. Results were analyzed using a Phosphorimager (Bio-Rad Personal FX). Oligonucleotide probe sequences are listed in Additional file 5.
Calculating overlap with human tiling analyses
The 3,478 QRNA predictions analyzed by microarray were mapped to the mouse-human UCSC genomic alignments (mm6-hg17) and were subset to the same regions analyzed by Cheng et al. [9] (i.e. not repetitive regions, for example), which were determined from the probe positions used in the tiling analysis (coordinates were converted to the hg17 genome release using the UCSC LiftOver tool [17]). The tiling dataset we focused on was generated using nuclear poly-A minus RNA derived from HepG2 cells. For QRNA predictions that overlapped with a transfrag, the degree of overlap was calculated as a percentage of the length of the QRNA prediction that overlaps the transfrag. The distribution of QRNA overlaps was compared to overlaps from randomly positioned QRNA predictions in the human-surveyed regions. The random set consisted of a set of sequences identical in length to the actual QRNA predictions, but with randomized positions in the human surveyed regions. The same analysis was repeated using 390 (151 mapped to human surveyed regions) mouse annotated miRNAs, snoRNAs, snRNAs, and tRNAs downloaded from NONCODE [47] and Rfam [48] databases, and 716 human ribosomal protein exons (260 mapped to human surveyed regions) annotated in the Refseq database [18].
Data availability
All supplementary data is available at . The microarray design has been submitted to NCBI GEO in MIAME format under accession GSE2366. The 8 novel transcripts have been submitted to GenBank [Genbank:AY954743 – Genbank:AY954751].
Authors' contributions
TB carried out the data compilation, microarray analysis, and data analysis. BB and TH coordinated the study. All authors contributed to preparation of the manuscript.
Supplementary Material
Additional File 1
Summary of array-tested ncRNA predictions.
Click here for file
Additional File 2
List of known mouse ncRNAs represented on array.
Click here for file
Additional File 3
Microarray probe sequences.
Click here for file
Additional File 4
Whole-blot northern data.
Click here for file
Additional File 5
List of northern probe sequences.
Click here for file
Acknowledgements
We would like to thank Janet Rossant, Wen Zhang, and Eric Sat for sharing mouse tissues and dissection expertise. We would also like to thank Susan McCracken for sharing HeLa nuclear extract. This work was supported by grants to T.R.H. and B.J.B. from CIHR and CFI. T.B. was supported by an NSERC graduate scholarship.
==== Refs
Waterston RH Lindblad-Toh K Birney E Rogers J Abril JF Agarwal P Agarwala R Ainscough R Alexandersson M An P Antonarakis SE Attwood J Baertsch R Bailey J Barlow K Beck S Berry E Birren B Bloom T Bork P Botcherby M Bray N Brent MR Brown DG Brown SD Bult C Burton J Butler J Campbell RD Carninci P Cawley S Chiaromonte F Chinwalla AT Church DM Clamp M Clee C Collins FS Cook LL Copley RR Coulson A Couronne O Cuff J Curwen V Cutts T Daly M David R Davies J Delehaunty KD Deri J Dermitzakis ET Dewey C Dickens NJ Diekhans M Dodge S Dubchak I Dunn DM Eddy SR Elnitski L Emes RD Eswara P Eyras E Felsenfeld A Fewell GA Flicek P Foley K Frankel WN Fulton LA Fulton RS Furey TS Gage D Gibbs RA Glusman G Gnerre S Goldman N Goodstadt L Grafham D Graves TA Green ED Gregory S Guigo R Guyer M Hardison RC Haussler D Hayashizaki Y Hillier LW Hinrichs A Hlavina W Holzer T Hsu F Hua A Hubbard T Hunt A Jackson I Jaffe DB Johnson LS Jones M Jones TA Joy A Kamal M Karlsson EK Karolchik D Kasprzyk A Kawai J Keibler E Kells C Kent WJ Kirby A Kolbe DL Korf I Kucherlapati RS Kulbokas EJ Kulp D Landers T Leger JP Leonard S Letunic I Levine R Li J Li M Lloyd C Lucas S Ma B Maglott DR Mardis ER Matthews L Mauceli E Mayer JH McCarthy M McCombie WR McLaren S McLay K McPherson JD Meldrim J Meredith B Mesirov JP Miller W Miner TL Mongin E Montgomery KT Morgan M Mott R Mullikin JC Muzny DM Nash WE Nelson JO Nhan MN Nicol R Ning Z Nusbaum C O'Connor MJ Okazaki Y Oliver K Overton-Larty E Pachter L Parra G Pepin KH Peterson J Pevzner P Plumb R Pohl CS Poliakov A Ponce TC Ponting CP Potter S Quail M Reymond A Roe BA Roskin KM Rubin EM Rust AG Santos R Sapojnikov V Schultz B Schultz J Schwartz MS Schwartz S Scott C Seaman S Searle S Sharpe T Sheridan A Shownkeen R Sims S Singer JB Slater G Smit A Smith DR Spencer B Stabenau A Stange-Thomann N Sugnet C Suyama M Tesler G Thompson J Torrents D Trevaskis E Tromp J Ucla C Ureta-Vidal A Vinson JP Von Niederhausern AC Wade CM Wall M Weber RJ Weiss RB Wendl MC West AP Wetterstrand K Wheeler R Whelan S Wierzbowski J Willey D Williams S Wilson RK Winter E Worley KC Wyman D Yang S Yang SP Zdobnov EM Zody MC Lander ES Initial sequencing and comparative analysis of the mouse genome Nature 2002 420 520 562 12466850 10.1038/nature01262
Chiaromonte F Weber RJ Roskin KM Diekhans M Kent WJ Haussler D The share of human genomic DNA under selection estimated from human-mouse genomic alignments Cold Spring Harb Symp Quant Biol 2003 68 245 254 15338624 10.1101/sqb.2003.68.245
Lander ES Linton LM Birren B Nusbaum C Zody MC Baldwin J Devon K Dewar K Doyle M FitzHugh W Funke R Gage D Harris K Heaford A Howland J Kann L Lehoczky J LeVine R McEwan P McKernan K Meldrim J Mesirov JP Miranda C Morris W Naylor J Raymond C Rosetti M Santos R Sheridan A Sougnez C Stange-Thomann N Stojanovic N Subramanian A Wyman D Rogers J Sulston J Ainscough R Beck S Bentley D Burton J Clee C Carter N Coulson A Deadman R Deloukas P Dunham A Dunham I Durbin R French L Grafham D Gregory S Hubbard T Humphray S Hunt A Jones M Lloyd C McMurray A Matthews L Mercer S Milne S Mullikin JC Mungall A Plumb R Ross M Shownkeen R Sims S Waterston RH Wilson RK Hillier LW McPherson JD Marra MA Mardis ER Fulton LA Chinwalla AT Pepin KH Gish WR Chissoe SL Wendl MC Delehaunty KD Miner TL Delehaunty A Kramer JB Cook LL Fulton RS Johnson DL Minx PJ Clifton SW Hawkins T Branscomb E Predki P Richardson P Wenning S Slezak T Doggett N Cheng JF Olsen A Lucas S Elkin C Uberbacher E Frazier M Gibbs RA Muzny DM Scherer SE Bouck JB Sodergren EJ Worley KC Rives CM Gorrell JH Metzker ML Naylor SL Kucherlapati RS Nelson DL Weinstock GM Sakaki Y Fujiyama A Hattori M Yada T Toyoda A Itoh T Kawagoe C Watanabe H Totoki Y Taylor T Weissenbach J Heilig R Saurin W Artiguenave F Brottier P Bruls T Pelletier E Robert C Wincker P Smith DR Doucette-Stamm L Rubenfield M Weinstock K Lee HM Dubois J Rosenthal A Platzer M Nyakatura G Taudien S Rump A Yang H Yu J Wang J Huang G Gu J Hood L Rowen L Madan A Qin S Davis RW Federspiel NA Abola AP Proctor MJ Myers RM Schmutz J Dickson M Grimwood J Cox DR Olson MV Kaul R Raymond C Shimizu N Kawasaki K Minoshima S Evans GA Athanasiou M Schultz R Roe BA Chen F Pan H Ramser J Lehrach H Reinhardt R McCombie WR de la Bastide M Dedhia N Blocker H Hornischer K Nordsiek G Agarwala R Aravind L Bailey JA Bateman A Batzoglou S Birney E Bork P Brown DG Burge CB Cerutti L Chen HC Church D Clamp M Copley RR Doerks T Eddy SR Eichler EE Furey TS Galagan J Gilbert JG Harmon C Hayashizaki Y Haussler D Hermjakob H Hokamp K Jang W Johnson LS Jones TA Kasif S Kaspryzk A Kennedy S Kent WJ Kitts P Koonin EV Korf I Kulp D Lancet D Lowe TM McLysaght A Mikkelsen T Moran JV Mulder N Pollara VJ Ponting CP Schuler G Schultz J Slater G Smit AF Stupka E Szustakowski J Thierry-Mieg D Thierry-Mieg J Wagner L Wallis J Wheeler R Williams A Wolf YI Wolfe KH Yang SP Yeh RF Collins F Guyer MS Peterson J Felsenfeld A Wetterstrand KA Patrinos A Morgan MJ de Jong P Catanese JJ Osoegawa K Shizuya H Choi S Chen YJ Initial sequencing and analysis of the human genome Nature 2001 409 860 921 11237011 10.1038/35057062
Kampa D Cheng J Kapranov P Yamanaka M Brubaker S Cawley S Drenkow J Piccolboni A Bekiranov S Helt G Tammana H Gingeras TR Novel RNAs identified from an in-depth analysis of the transcriptome of human chromosomes 21 and 22 Genome Res 2004 14 331 342 14993201 10.1101/gr.2094104
Kapranov P Cawley SE Drenkow J Bekiranov S Strausberg RL Fodor SP Gingeras TR Large-scale transcriptional activity in chromosomes 21 and 22 Science 2002 296 916 919 11988577 10.1126/science.1068597
Numata K Kanai A Saito R Kondo S Adachi J Wilming LG Hume DA Hayashizaki Y Tomita M Identification of putative noncoding RNAs among the RIKEN mouse full-length cDNA collection Genome Res 2003 13 1301 1306 12819127 10.1101/gr.1011603
Okazaki Y Furuno M Kasukawa T Adachi J Bono H Kondo S Nikaido I Osato N Saito R Suzuki H Yamanaka I Kiyosawa H Yagi K Tomaru Y Hasegawa Y Nogami A Schonbach C Gojobori T Baldarelli R Hill DP Bult C Hume DA Quackenbush J Schriml LM Kanapin A Matsuda H Batalov S Beisel KW Blake JA Bradt D Brusic V Chothia C Corbani LE Cousins S Dalla E Dragani TA Fletcher CF Forrest A Frazer KS Gaasterland T Gariboldi M Gissi C Godzik A Gough J Grimmond S Gustincich S Hirokawa N Jackson IJ Jarvis ED Kanai A Kawaji H Kawasawa Y Kedzierski RM King BL Konagaya A Kurochkin IV Lee Y Lenhard B Lyons PA Maglott DR Maltais L Marchionni L McKenzie L Miki H Nagashima T Numata K Okido T Pavan WJ Pertea G Pesole G Petrovsky N Pillai R Pontius JU Qi D Ramachandran S Ravasi T Reed JC Reed DJ Reid J Ring BZ Ringwald M Sandelin A Schneider C Semple CA Setou M Shimada K Sultana R Takenaka Y Taylor MS Teasdale RD Tomita M Verardo R Wagner L Wahlestedt C Wang Y Watanabe Y Wells C Wilming LG Wynshaw-Boris A Yanagisawa M Yang I Yang L Yuan Z Zavolan M Zhu Y Zimmer A Carninci P Hayatsu N Hirozane-Kishikawa T Konno H Nakamura M Sakazume N Sato K Shiraki T Waki K Kawai J Aizawa K Arakawa T Fukuda S Hara A Hashizume W Imotani K Ishii Y Itoh M Kagawa I Miyazaki A Sakai K Sasaki D Shibata K Shinagawa A Yasunishi A Yoshino M Waterston R Lander ES Rogers J Birney E Hayashizaki Y Analysis of the mouse transcriptome based on functional annotation of 60,770 full-length cDNAs Nature 2002 420 563 573 12466851 10.1038/nature01266
Shabalina SA Spiridonov NA The mammalian transcriptome and the function of non-coding DNA sequences Genome Biol 2004 5 105 15059247 10.1186/gb-2004-5-4-105
Cheng J Kapranov P Drenkow J Dike S Brubaker S Patel S Long J Stern D Tammana H Helt G Sementchenko V Piccolboni A Bekiranov S Bailey DK Ganesh M Ghosh S Bell I Gerhard DS Gingeras TR Transcriptional Maps of 10 Human Chromosomes at 5-Nucleotide Resolution Science 2005 308 1149 1154 15790807 10.1126/science.1108625
Johnson JM Edwards S Shoemaker D Schadt EE Dark matter in the genome: evidence of widespread transcription detected by microarray tiling experiments Trends Genet 2005 21 93 102 15661355 10.1016/j.tig.2004.12.009
Wang J Zhang J Zheng H Li J Liu D Li H Samudrala R Yu J Wong GK Mouse transcriptome: neutral evolution of 'non-coding' complementary DNAs Nature 2004 431 1 p following 757; discussion following 757 15495343
Bono H Yagi K Kasukawa T Nikaido I Tominaga N Miki R Mizuno Y Tomaru Y Goto H Nitanda H Shimizu D Makino H Morita T Fujiyama J Sakai T Shimoji T Hume DA Hayashizaki Y Okazaki Y Systematic expression profiling of the mouse transcriptome using RIKEN cDNA microarrays Genome Res 2003 13 1318 1323 12819129 10.1101/gr.1075103
Wyers F Rougemaille M Badis G Rousselle JC Dufour ME Boulay J Regnault B Devaux F Namane A Seraphin B Libri D Jacquier A Cryptic pol II transcripts are degraded by a nuclear quality control pathway involving a new poly(A) polymerase Cell 2005 121 725 737 15935759 10.1016/j.cell.2005.04.030
Rivas E Eddy SR Noncoding RNA gene detection using comparative sequence analysis BMC Bioinformatics 2001 2 8 11801179 10.1186/1471-2105-2-8
Rivas E Klein RJ Jones TA Eddy SR Computational identification of noncoding RNAs in E. coli by comparative genomics Curr Biol 2001 11 1369 1373 11553332 10.1016/S0960-9822(01)00401-8
McCutcheon JP Eddy SR Computational identification of non-coding RNAs in Saccharomyces cerevisiae by comparative genomics Nucleic Acids Res 2003 31 4119 4128 12853629 10.1093/nar/gkg438
Hughes TR Mao M Jones AR Burchard J Marton MJ Shannon KW Lefkowitz SM Ziman M Schelter JM Meyer MR Kobayashi S Davis C Dai H He YD Stephaniants SB Cavet G Walker WL West A Coffey E Shoemaker DD Stoughton R Blanchard AP Friend SH Linsley PS Expression profiling using microarrays fabricated by an ink-jet oligonucleotide synthesizer Nat Biotechnol 2001 19 342 347 11283592 10.1038/86730
Karolchik D Baertsch R Diekhans M Furey TS Hinrichs A Lu YT Roskin KM Schwartz M Sugnet CW Thomas DJ Weber RJ Haussler D Kent WJ The UCSC Genome Browser Database Nucleic Acids Res 2003 31 51 54 12519945 10.1093/nar/gkg129
NCBI ftp://ftp.ncbi.nih.gov/genomes/M_musculus/RNA/
NCBI ftp://ftp.ncbi.nih.gov/genomes/M_musculus/protein/
NCBI ftp://ftp.ncbi.nih.gov/genbank/
Lowe TM Eddy SR tRNAscan-SE: a program for improved detection of transfer RNA genes in genomic sequence Nucleic Acids Res 1997 25 955 964 9023104 10.1093/nar/25.5.955
Lowe TM Eddy SR A computational screen for methylation guide snoRNAs in yeast Science 1999 283 1168 1171 10024243 10.1126/science.283.5405.1168
Babak T Zhang W Morris Q Blencowe BJ Hughes TR Probing microRNAs with microarrays: tissue specificity and functional inference Rna 2004 10 1813 1819 15496526 10.1261/rna.7119904
Cavaille J Vitali P Basyuk E Huttenhofer A Bachellerie JP A novel brain-specific box C/D small nucleolar RNA processed from tandemly repeated introns of a noncoding RNA gene in rats J Biol Chem 2001 276 26374 26383 11346658 10.1074/jbc.M103544200
Inoue A Kobayashi Y Ishizuka M Hirose S Hagiwara H Identification of a novel osteoblastic gene, inducible by C-type natriuretic peptide, whose transcript might function in mineralization as a noncoding RNA Calcif Tissue Int 2002 70 111 116 11870417 10.1007/s00223-001-1012-3
Lagos-Quintana M Rauhut R Yalcin A Meyer J Lendeckel W Tuschl T Identification of tissue-specific microRNAs from mouse Curr Biol 2002 12 735 739 12007417 10.1016/S0960-9822(02)00809-6
Takeda K Ichijo H Fujii M Mochida Y Saitoh M Nishitoh H Sampath TK Miyazono K Identification of a novel bone morphogenetic protein-responsive gene that may function as a noncoding RNA J Biol Chem 1998 273 17079 17085 9642273 10.1074/jbc.273.27.17079
Brannan CI Dees EC Ingram RS Tilghman SM The product of the H19 gene may function as an RNA Mol Cell Biol 1990 10 28 36 1688465
Hiley SL Babak T Hughes TR Global analysis of yeast RNA processing identifies new targets of RNase III and uncovers a link between tRNA 5' end processing and tRNA splicing Nucleic Acids Res 2005 33 3048 3056 15920104 10.1093/nar/gki608
Su AI Wiltshire T Batalov S Lapp H Ching KA Block D Zhang J Soden R Hayakawa M Kreiman G Cooke MP Walker JR Hogenesch JB A gene atlas of the mouse and human protein-encoding transcriptomes Proc Natl Acad Sci U S A 2004 101 6062 6067 15075390 10.1073/pnas.0400782101
Szymanski M Erdmann VA Barciszewski J Noncoding regulatory RNAs database Nucleic Acids Res 2003 31 429 431 12520042 10.1093/nar/gkg124
Eddy SR Computational genomics of noncoding RNA genes Cell 2002 109 137 140 12007398 10.1016/S0092-8674(02)00727-4
Nobrega MA Zhu Y Plajzer-Frick I Afzal V Rubin EM Megabase deletions of gene deserts result in viable mice Nature 2004 431 988 993 15496924 10.1038/nature03022
Eddy SR Noncoding RNA genes Curr Opin Genet Dev 1999 9 695 699 10607607 10.1016/S0959-437X(99)00022-2
Sempere LF Freemantle S Pitha-Rowe I Moss E Dmitrovsky E Ambros V Expression profiling of mammalian microRNAs uncovers a subset of brain-expressed microRNAs with possible roles in murine and human neuronal differentiation Genome Biol 2004 5 R13 15003116 10.1186/gb-2004-5-3-r13
Brockdorff N Ashworth A Kay GF McCabe VM Norris DP Cooper PJ Swift S Rastan S The product of the mouse Xist gene is a 15 kb inactive X-specific transcript containing no conserved ORF and located in the nucleus Cell 1992 71 515 526 1423610 10.1016/0092-8674(92)90519-I
Bertone P Stolc V Royce TE Rozowsky JS Urban AE Zhu X Rinn JL Tongprasit W Samanta M Weissman S Gerstein M Snyder M Global identification of human transcribed sequences with genome tiling arrays Science 2004 306 2242 2246 15539566 10.1126/science.1103388
Smit AFA Green P http://ftp.genome.washington.edu/RM/RepeatMasker.html
Kent WJ BLAT--the BLAST-like alignment tool Genome Res 2002 12 656 664 11932250 10.1101/gr.229202. Article published online before March 2002
Kasprzyk A Keefe D Smedley D London D Spooner W Melsopp C Hammond M Rocca-Serra P Cox T Birney E EnsMart: a generic system for fast and flexible access to biological data Genome Res 2004 14 160 169 14707178 10.1101/gr.1645104
NCBI ftp://ftp.ncbi.nih.gov/blast/db/
NCBI ftp://ftp.ncbi.nih.gov/repository/UniGene/
Mathews DH Sabina J Zuker M Turner DH Expanded sequence dependence of thermodynamic parameters improves prediction of RNA secondary structure J Mol Biol 1999 288 911 940 10329189 10.1006/jmbi.1999.2700
Dignam JD Lebovitz RM Roeder RG Accurate transcription initiation by RNA polymerase II in a soluble extract from isolated mammalian nuclei Nucleic Acids Res 1983 11 1475 1489 6828386
Wiegant JC van Gijlswijk RP Heetebrij RJ Bezrookove V Raap AK Tanke HJ ULS: a versatile method of labeling nucleic acids for FISH based on a monofunctional reaction of cisplatin derivatives with guanine moieties Cytogenet Cell Genet 1999 87 47 52 10640810 10.1159/000015390
Shai O Morris Q Frey BJ Spatial Bias Removal in Microarray Images University of Toronto Technical Report, http://hugheslabmedutorontoca/Zhang/tech%reportpdf 2003
Huber W von Heydebreck A Sultmann H Poustka A Vingron M Variance stabilization applied to microarray data calibration and to the quantification of differential expression Bioinformatics 2002 18 Suppl 1 S96 104 12169536
Zhang W Morris QD Chang R Shai O Bakowski MA Mitsakakis N Mohammad N Robinson MD Zirngibl R Somogyi E Laurin N Eftekharpour E Sat E Grigull J Pan Q Peng WT Krogan N Greenblatt J Fehlings M van der Kooy D Aubin J Bruneau BG Rossant J Blencowe BJ Frey BJ Hughes TR The functional landscape of mouse gene expression J Biol 2004 3 21 15588312 10.1186/jbiol16
|
16083503
|
PMC1199595
|
CC BY
|
2021-01-04 16:32:49
|
no
|
BMC Genomics. 2005 Aug 5; 6:104
|
utf-8
|
BMC Genomics
| 2,005 |
10.1186/1471-2164-6-104
|
oa_comm
|
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-1051609113210.1186/1471-2164-6-105Research ArticleThe complete mitochondrial genome of the stomatopod crustacean Squilla mantis Cook Charles E [email protected] Department and Museum of Zoology, University of Cambridge, Downing Street, Cambridge, CB2 3EJ, UK2 Natural Environment Research Council, British Antarctic Survey, Biological Sciences Division, High Cross, Madingley Road, Cambridge CB3 0ET, UK2005 9 8 2005 6 105 105 10 10 2004 9 8 2005 Copyright © 2005 Cook; licensee BioMed Central Ltd.2005Cook; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Animal mitochondrial genomes are physically separate from the much larger nuclear genomes and have proven useful both for phylogenetic studies and for understanding genome evolution. Within the phylum Arthropoda the subphylum Crustacea includes over 50,000 named species with immense variation in body plans and habitats, yet only 23 complete mitochondrial genomes are available from this subphylum.
Results
I describe here the complete mitochondrial genome of the crustacean Squilla mantis (Crustacea: Malacostraca: Stomatopoda). This 15994-nucleotide genome, the first described from a hoplocarid, contains the standard complement of 13 protein-coding genes, 22 transfer RNA genes, two ribosomal RNA genes, and a non-coding AT-rich region that is found in most other metazoans. The gene order is identical to that considered ancestral for hexapods and crustaceans. The 70% AT base composition is within the range described for other arthropods. A single unusual feature of the genome is a 230 nucleotide non-coding region between a serine transfer RNA and the nad1 gene, which has no apparent function.
I also compare gene order, nucleotide composition, and codon usage of the S. mantis genome and eight other malacostracan crustaceans. A translocation of the histidine transfer RNA gene is shared by three taxa in the order Decapoda, infraorder Brachyura; Callinectes sapidus, Portunus trituberculatus and Pseudocarcinus gigas. This translocation may be diagnostic for the Brachyura. For all nine taxa nucleotide composition is biased towards AT-richness, as expected for arthropods, and is within the range reported for other arthropods. Codon usage is biased, and much of this bias is probably due to the skew in nucleotide composition towards AT-richness.
Conclusion
The mitochondrial genome of Squilla mantis contains one unusual feature, a 230 base pair non-coding region has so far not been described in any other malacostracan. Comparisons with other Malacostraca show that all nine genomes, like most other mitochondrial genomes, share a bias toward AT-richness and a related bias in codon usage. The nine malacostracans included in this analysis are not representative of the diversity of the class Malacostraca, and additional malacostracan sequences would surely reveal other unusual genomic features that could be useful in understanding mitochondrial evolution in this taxon.
==== Body
Background
The mitochondria are extranuclear organelles present in all metazoans. They contain a circular genome, usually around 16 kilobases in length, with 37 genes (13 protein-coding, two ribosomal RNA genes, and 22 transfer RNA genes). This gene content is widely conserved, but gene order and the DNA sequences of the genes themselves are variable. Because of their small size many more mitochondrial genomes than nuclear genomes have been sequenced, and comparisons among them may serve as models for the evolution of the much larger nuclear genomes [1]. In addition, gene order rearrangements and mitochondrial gene sequences have been widely used for phylogenetic inference [2-7].
At present there are about 650 complete mitochondrial genomes available in public databases. Of these, about 75 percent are of vertebrates. By contrast only 129 complete mitochondrial genomes are available from arthropods, which are the most diverse and speciose phylum of animals. In addition, there is considerable taxonomic bias among the available arthropod sequences; 86, (67 percent) are hexapods. The subphylum Crustacea includes over 50,000 named species and is ecologically and morphologically the most diverse of the arthropod groups, and therefore of all the animals. Crustaceans occupy marine, terrestrial, and fresh water habitats from the deep sea to high mountains; range in adult size from less than one millimeter to more than four meters (leg span); and exhibit extensive variability in body plans when compared to other arthropod groups [8]. At present there are 23 complete crustacean mitochondrial genomes available. Within the Crustacea members of the class Malacostraca, which include crabs, lobsters, and shrimp, are perhaps the most well known to non-scientists. Due to their economic importance this group is often the focus of scientific enquiry. At present there are nine complete malacostracan mitochondrial genomes available, including that of the stomatopod shrimp Squilla mantis. In this paper I describe this genome and compare it to eight other mitochondrial genomes that are available from other Malacostraca.
Mantid shrimps, or stomatopods, are benthic predators distributed in the shallow waters of tropical and subtropical seas. They are best known for their raptorial appendages – pointed or clubbed – which they use to make lightning-fast attacks that disable prey animals by spearing or blunt trauma. Large individuals with the club-type appendages have been known to shatter the sides of aquaria [9]. Squilla mantis (Linnaeus, 1758) (Crustacea: Malacostraca: Stomatopoda), with a maximum length of around 20 cm, is distributed in shallow waters throughout the Mediterranean Sea and Eastern Atlantic [10]. S. mantis is widely consumed by humans throughout its range; UN Food and Agriculture Administration statistics indicate that total catches in the Mediterranean are currently in excess of 6500 tonnes per annum so this species is of some commercial importance [11].
Results and discussion
Mitochondrial genome composition
The mitochondrial genome of Squilla mantis (GenBank accession number AY639936) is a circular molecule of 15994 nucleotides that contains the same 13 protein-coding genes, 22 transfer RNA genes (tRNA), and two ribosomal RNA genes (rRNA) found in other metazoans [12,13]; the majority strand (i.e., the strand encoding the majority of genes) encodes nine protein-coding genes and 14 tRNAs while the minority strand encodes four protein-coding genes, eight tRNAs and both rRNA genes (Table 1). The S. mantis genome, like that reported for other arthropod genomes, is AT-rich and has an overall AT content of 70%. This frequency, as expected, varies for different regions of the genome. First and second codon positions average 62% and 63% AT, respectively, tRNA and rRNA genes average 72%, third codon positions average 79%, and putative non-coding regions reach up to 87% AT content (Table 2). There are no significant differences in AT frequency for genes encoded on the majority or minority strands. These values are within the range of 60–87% reported for other arthropods and are not unusual [14,15].
Table 1 Complete mitochondrial genomes available for the Malacostraca. Taxononomic classification from the NCBI taxonomy browser [37] and from reference [8].
Species GenBank accession Order Infraorder/Suborder Family
Callinectes sapidus NC_006281 Decapoda Brachyura Portunidae
Cherax destructor NC_011243 Decapoda Astacidea Parastacidae
Harpiosquilla harpax AY699271 Stomatopoda Unipeltata Squillidae
Macrobrachium rosenbergii AY659990 Decapoda Caridea Palaemonidae
Pagurus longicarpus NC_003058 Decapoda Anomura Paguridae
Panulirus japonicus NC_004251 Decapoda Palinura Palinuridae
Penaeus monodon NC_002184 Decapoda Dendrobranchiata Penaeidae
Portunus trituberculatus NC_005037 Decapoda Brachyura Portunidae
Pseudocarcinus gigas AY562127 Decapoda Brachyura Eriphiidae
Squilla mantis NC_006081 Stomatopoda Unipeltata Squillidae
Table 2 Nucleotide composition of Squilla mantis mitochondrial genome features. Major and minor strand genes show no significant differences in nucleotide composition.
Genome feature Number of nucleotides Proportion of nucleotides %AT
A C G T
All nucleotides (major strand) 15994 0.351 0.168 0.130 0.351 70
rRNA genes (minor strand) 1760 0.464 0.193 0.135 0.208 67
AT-rich region 1 (major strand) 230 0.448 0.061 0.065 0.426 87
AT-rich region 12(major strand) 861 0.423 0.137 0.095 0.345 77
tRNA genes (major strand) 945 0.361 0.126 0.157 0.357 72
tRNA genes (minor strand) 469 0.360 0.104 0.177 0.358 72
tRNA genes (both strands) 1414 0.361 0.119 0.163 0.357 72
Unassigned nucleotides (major strand) 76 0.316 0.132 0.013 0.539 86
Protein coding genes
All positions 11070 0.274 0.156 0.174 0.396 67
first codon positions 3690 0.292 0.136 0.241 0.331 62
second codon positions 3690 0.187 0.208 0.162 0.444 63
third codon positions 3690 0.359 0.106 0.102 0.433 79
Majority strand protein coding genes
All positions 6966 0.278 0.182 0.163 0.377 66
first codon positions 2262 0.287 0.154 0.242 0.317 60
second codon positions 2262 0.190 0.237 0.140 0.434 62
third codon positions 2262 0.380 0.130 0.079 0.411 79
Minority strand protein coding genes
All positions 4284 0.269 0.113 0.191 0.427 70
first codon positions 1428 0.300 0.107 0.239 0.353 65
second codon positions 1428 0.182 0.162 0.195 0.460 64
third codon positions 1428 0.325 0.068 0.138 0.469 79
Gene structure
The predicted structures of the 22 S. mantis tRNAs are shown in Figure 1. Twenty one of these genes were identified by tRNAscan-SE [16] and have secondary structures similar to those of other published metazoan tRNA genes. Two genes, trnS1 and trnQ, have single T-T mismatches in the acceptor stem, and one gene, trnM, has a single C-A mismatch in the stem of the TψC loop. The trnS1 gene was not identified by the tRNAscan software; rather, it was located by its conserved position in the genome. The variable loop of this gene, with nine nucleotides, is longer than the average of four or five for mitochondrial tRNA genes. This feature is characteristic of type 2 transfer RNA genes, which are uncommon in animal mitochondria but are the norm for bacterial and eukaryotic trnS genes.
Figure 1 Squilla mantis mitochondrial tRNA genes folded into inferred cloverleaf structures.
The large and small subunit ribosomal RNA (rRNA) genes (rnl, rns) have an AT content of 67%, within the range reported for other arthropod ribosomal RNA genes. Alignments of these genes with other arthropod homologues (not shown) as expected show both conserved and unconserved regions that correspond with the putative stems and loops within these genes. There are thus no unusual features to report for the two rRNA genes.
All of the 13 protein-coding genes, except cox1, have putative ATR methionine or ATT isoleucine start codons. The putative first codon of the cox1 gene is ACG threonine. The lack of a standard initiation codon in cox1 genes is common in arthropod mitochondria, so S. mantis is not unusual [17,18]. Two of the protein-coding genes, cox1 and nad6, lack a full TAA or TAG stop codon. These genes appear to terminate with a single T from which a stop codon is created by polyadenylation of the mRNA during processing. Again, this phenomenon has been observed in other arthropod mitochondrial genomes and is not unusual [17,19,20].
AT-rich regions
Arthropod mitochondrial genomes typically have a long region that has an AT content higher than that of mitochondrial coding regions. This AT-rich region, ranging from 263 to 4601 base pairs in length and usually located between the rns and trnI genes, is often termed the control region because it contains a number of regulatory elements including the origin of replication for the heavy strand of the mitochondrial genome [21,22]. In some arthropods the AT-rich region is reported to have any or all of these four different motifs: tandemly repeated sequences, a long sequence of T's, a subregion of even higher AT richness, and stem-loop structures [23,24].
In S. mantis there are two AT-rich regions, numbered 1 and 2 on Tables 2 and 3. AT-rich region 2 corresponds to the conventional arthropod region between rns and trnI; it is 862 base pairs long, well within the reported range for other arthropods, with an AT content of 77% compared to 70% for the entire S. mantis mitochondrion. However, this region has none of the four motif types that have been reported for arthropods, and I was not able to identify any putatively functional motifs.
Table 3 Annotation of the mitochondrial genome of Squilla mantis. Position numbers refer to positions on the major strand, with the first nucleotide of trnI assigned as number 1. Parentheses indicate genes encoded on the minor strand. Intergenic region numbers refer to the number of non-coding bases between features (positive integers) or the number of overlapping bases (negative integers). Asterisks (*) refer to incomplete termination codons that may be extended by post-transcriptional adenylation.
Feature Position First codon Stop codon Intergenic region
trnI 1–69
trnQ (68–135) -2
trnM 149–217 13
nad2 275–1219 ATA TAA 57
trnW 1218–1285 -2
trnC (1291–1354) 5
trnY (1356–1420) 1
cox1 1421–2959 ACG TAA 0
trnL2 2955–3021 -5
cox2 3026–3713 ATG T* 0
trnK 3714–3781 0
trnD 3781–3849 -1
atp8 3850–4002 ATC TAA 0
atp6 3996–4673 ATG TAA -7
cox3 4673–5461 ATG TAG -1
trnG 5461–5526 -1
nad3 5527–5880 ATG TAA 0
trnA 5884–5948 3
trnR 5953–6017 4
trnN 6019–6087 1
trnS1 6088–6155 0
trnE 6156–6221 0
trnF (6222–6288) 0
nad5 (6289–8001) ATA TAA 0
trnH (8020–8086) 18
nad4 (8086–9426) ATG TAG -1
nad4L (9420–9719) ATG TAA -7
trnT 9722–9790 2
trnP (9791–9857) 0
nad6 9900–10386 ATA T* 42
cob 10387–11523 ATG TAA 0
TrnS2 11523–11592 -1
AT-rich region 1 11593–11822 0
nad1 (11823–12803) ATA TAA 0
trnL1 (12801–12867) -3
rnl (12868–14250) 0
trnV (14251–14300) 0
rns (14301–15132) 0
AT-rich region 2 15133–15994 0
I therefore examined the shorter AT-rich region of 230 nucleotides between the trnS2 and nad1 genes for possible functional motifs. Most arthropod mitochondrial genomes have a few short non-coding regions between some genes, usually from a few bases to 20 bases long, but longer non-coding regions, such as AT-rich region 1 in S. mantis, are rare. It therefore seemed possible that this region might have taken over some of the functions putatively assigned to the longer AT-rich region. However, AT-rich region 1, like region 2, contains none of the motifs listed above. Furthermore, the AT content of this region, at 87%, is similar to that calculated collectively for other unassigned nucleotides in the S. mantis genome (Table 2) and is consistent with the hypothesis that this region has no function.
Unusual genomic features, such as this non-coding region or gene order rearrangements, can be useful as characters for reconstructing evolutionary relationships [13,25]. A second AT-rich region is notably absent even in Harpiosquilla harpax, which is also a member of the family Squillidae. A survey of other members of the genus Squilla for the presence of a similar region would perhaps enhance our understanding of the history of this unusual genomic feature.
Comparison with other malacostracan crustaceans
A number of features of mitochondrial genomes can be used to infer relationships among taxa. These include phylogenetic analysis using DNA and protein sequences, relative rates of sequence evolution, gene order, and the effective number of codons (Nc). I present a phylogenetic analysis and a discussion of rates of sequence evolution in arthropods (including S. mantis) elsewhere [26], and discuss gene order and Nc below.
Rearrangements of the mitochondrial genome are relatively rare events in evolutionary history. Such rare events can be used to infer relationships among taxa, and mitochondrial gene order rearrangements have proven useful in understanding some aspects of arthropod evolution [27-29]. Figure 2 shows the mitochondrial gene order for the nine Malacostraca for which there are complete mitochondrial genomes. Five of these genomes share the gene order considered ancestral for the Pancrustacea (Crustacea + Hexapoda) [27]. Callinectes sapidus, Portunus trituberculatus and Pseudocarcinus gigas share a single translocation of trnH compared to the ancestral gene order. The mitochondrial genome of Cherax destructor is considerably rearranged and evidences at least seven translocation events compared to the ancestral pancrustacean arrangement [20]. C. sapidus, P. trituberculatus and P. gigas are all decapods within the infraorder Brachyura (Table 1). The trnH translocation shared by these three taxa is therefore not surprising. It is possible that this character is shared among all of the Brachyura, and could therefore serve as a marker for membership in this group and might aid in rapid identification of unidentified individuals, such as larvae or processed materials in markets.
Figure 2 Mitochondrial gene orders for nine malacostracan crustaceans. The style of the figure is adapted from Figure 1 of Lavrov et al. [25]. Protein and ribosomal RNA genes (large boxes) are abbreviated as in the text. Transfer RNA genes are abbreviated with single letter codes (see Figure 1). The striped box represents the AT-rich region. The ancestral pancrustacean gene order is found for five of the nine taxa, including Squilla mantis. The position of AT-rich region 1 in the S. mantis genome is noted with an arrow. Genes are transcribed from right to left except when underlined. Shaded boxes indicate genes whose positions differ from their positions in the ancestral pancrustacean sequence. The number in parentheses next to taxa names represents the minimum number of rearrangement events that separates that gene arrangement from the ancestral pancrustacean gene order (see Miller et al. [20] for a fuller discussion of the rearrangements in C. destructor).
The effective number of codons used in a gene, Nc, is a statistic developed by Wright [30] to quantify how far codon usage in a gene departs from the equal use of all synonymous codons. The value of Nc can range from 20, the theoretical extreme in which only one codon is used for each amino acid, to 61 when the use of all synonymous codons is equally likely. This statistic, initially developed to compare codon usage between different genes in the same genome, can also be used to compare codon usage between genomes. I calculated Nc for each of the nine malacostracan mitochondrial genomes using the program CodonW [31]. Table 4 shows Nc and GC content for majority strand, minority strand, and all protein genes in the malacostracan mitochondrial genomes. GC rather than AT content is presented to conform to the convention for these comparisons. The Nc values, which range from 38 to 53, are all below the value of 61 that indicates random codon usage, so codon usage in all nine genomes is non-random. There are no obvious similarities in the values for related taxa (i.e., the decapods), but extensive additional sampling among the Malacostraca would be necessary to confirm this observation.
Table 4 Codon usage in nine malacostracan mitochondrial genomes. Columns indicate GC content and effective number of codons (Nc) for majority strand genes, minority strand genes, and all genes. The least squares linear regression equation and the coefficient of determination are also shown for each data set.
Taxon Majority strand genes Minority strand genes All genes
Nc GC Nc GC Nc GC
Callinectes sapidus 43.36 0.348 39.05 0.302 44.79 0.33
Cherax destructor 52.79 0.423 49.65 0.366 55.25 0.401
Harpiosquilla harpax 40.17 0.325 40.23 0.309 41.08 0.319
Macrobrachium rosenbergii 43.92 0.418 39.74 0.369 52.49 0.399
Pagurus longicarpus 38.17 0.319 34.85 0.279 39.6 0.304
Panulirus japonicus 49.71 0.382 49.8 0.361 51.3 0.374
Penaeus monodon 39.79 0.322 34.99 0.284 39.44 0.307
Portunus trituberculatus 40.71 0.326 37.29 0.289 40.97 0.312
Pseudocarcinus gigas 48.57 0.332 49.89 0.308 45.96 0.315
Squilla mantis 42.48 0.328 38.45 0.303 41.57 0.319
Least sqaures equation y = 84.83x+14.08 y = 116.49x+4.47 y = 145.8x-4.044
Coefficient of determination (R2) R2 = 0.5015 R2 = 0.4493 R2 = 0.9139
In Figure 3 Nc and GC values are plotted. The distribution of the points suggests a linear relationship between Nc and GC content. A similar association of Nc and GC content was observed by Negrisolo et al. [32]. Equations representing a least squares linear regression analysis are shown for all three data sets in Table 2. Only the regression line for the all genes data set is shown on Figure 3 to prevent clutter. These equations are not statistically probative, but the distribution of points around the line shown in Figure 3 does add to the qualitative impression of a relationship between Nc and GC content. I also calculated the coefficient of determination (R2) for each data set. The values for the majority and minority strand columns are near 0.5, suggesting that around 50% of the variation in one variable is associated with the other. That is, if GC content is taken as independent then 50% of the codon bias in the majority and minority strands is due to the influence of the bias towards low GC values. When both data sets are combined R2 rises to 0.91, suggesting that codon bias and GC content are very closely associated. This discordance between R2 for each strand separately and R2 for all protein-coding genes is puzzling and merits additional study. However, if one assumes that GC content is driven by other biochemical factors then it is clear that much, if not most, of the codon bias observed in these mitochondrial genomes is a consequence of this nucleotide bias.
Figure 3 Relationship between GC content and effective number of codons for mitochondrial protein-coding genes. The line represents the least squares linear regression calculated for the all genes data set. This equation is shown in Table 4.
Conclusion
This is the first formal description of the mitochondrial genome of a stomatopod crustacean. This genome maintains the same genes and gene order that are inferred as ancestral in the Pancrustacea, but does contain one unusual feature: a 230 base pair AT-rich region between the trnS2 and nad1 genes. This feature has no discernable function, but it may prove useful as a character in understanding the evolutionary history of the genus Squilla. Three other arthropod mitochondrial genomes have two large non-coding regions; the ostracod crustacean Vargula hilgendorfii, a millipede Thyropygus sp., and the tick Boophilus microplus [15,33,34]. Of these the latter two are clearly duplications of the orginal control region. Only V. hilgendorfii has, like S. mantis, two apparently unrelated non-contiguous AT-rich regions.
A comparison of nine malacostracan genomes, including that of S. mantis, shows that all nine exhibit the nucleotide composition bias favoring A and T nucleotides that is commonly observed for arthropod genomes, and that this bias is responsible at least in part for the observed codon usage bias in these genomes. However, there are no observable patterns of nucleotide composition bias or codon usage bias that unite particular taxa into common groups. These nine taxa represent only a small fraction of the diversity of the Malacostraca, and additional sequencing from across the diversity of this taxon would provide additional data for understanding the evolution of mitochondrial genomes of the class.
Methods
Samples and DNA extraction
A single freshly caught specimen of S. mantis was purchased from the fish market at Heraklion, Crete, Greece. Six grams of abdominal muscle were dissected from the specimen and immediately frozen at -70°C. Approximately one half gram of the frozen tissue was shaved from the specimen using a sterile razor blade and genomic DNA extracted using a QIAGEN genomic-tip 20 and the associated buffer set according to the manufacturer's protocol.
PCR, sequencing, and annotation
Short fragments (300–1000 nucleotides) of the mitochondrial genome were amplified at low stringency (50–55 degree annealing temperatures) using primers designed to work on all arthropods (Table 4). Amplification products were cloned into the T-Easy vector (Promega) and at least three clones from each PCR product were sequenced with vector-specific primers using ABI Big-Dye chemistry. Squilla-specific primers were designed and used to amplify longer fragments of 1000–4000 nucleotides that spanned the gaps between the short fragments. Longer fragments were also ligated into the T-Easy vector and at least three clones from each ligation were isolated. Each clone was sequenced using a primer-walking strategy initiated with vector-specific primers. Sequences were assembled using Sequencher v. 3.1 (GeneCodes Corp.). Protein-coding genes were identified using BLAST searches [35] and by comparison with other arthropod mitochondrial genome sequences. Transfer RNA genes were identified using tRNAscan-SE [36]. Transfer RNA sequences were folded by eye, but made use of the tRNAscan-SE server output when that was available. The effective number of codons, Nc was calculated using the software package CodonW [31].
Abbreviations
Protein genes: cox1, cox 2, cox 3, cytochrome oxidase subunits I, II, and III; cob, cytochrome oxidase b; atp6, atp8, ATP synthase subunits 6 and 8; nad1, nad2, nad3, nad4, nad4L, nad5, nadD6, NADH dehydrogenase subunits 1–6 and 4L. Large and small subunit ribosomal RNA genes are abbreviated rnl and rns. Transfer RNA genes are listed as trnA, trnC, etc., where the final letter is the single letter abbreviation for that amino acid. This nomenclature follows that of Lavrov et al. [25].
Table 5 Primers used to amplify short fragments of the Squilla mantis genome. Majority/minority strand are with reference to the ancestral pancrustacean sequence. These primers will work for almost all arthropods and for many other metazoans as well. Reference "JB" is J. Boore pers. comm. Reference "CEC" refers to primers designed for this project.
Gene Majority strand Sequence Minority strand Sequence Reference
COX1 Lco1490 GGTCAACAAATCATAAAGATATTGG Hco2198 TAAACTTCAGGGTGACCAAAAAATCA 38
COX1-COX2 CO1DL CCWCGWCGWTAYTCWGAYTAYCCWGA CO2DL CWGAATARRCATAWSWTCARTATCATTG JB
ATP6-COX3 ATP6.AANMM GCCGTACGGCTTGCAGCNAAYATRAT CO3DL2 ACWACGTCKACGAAGTGTCARTATCA CEC, JB
COX3 CO3DL1 TGGTGGCGAGATGTKKTNCGNGA CO3DL2 ACWACGTCKACGAAGTGTCARTATCA JB
ND5 ND5-R-DL TARAAKCCWGMTARAAAWGGKAWWCC ND5-F-DL TWYTATTAGGKTGAGATGGKYTNGG JB
ND4 ND4-R_DL GARGAWCAKAWWCCRTGAGCAATYAT ND4-F-DL CCKAARGCYCAYGTKGARGCYCC JB
CYTB Cytb424-449 GGWTAYGTWYTWCCWTGRGGWCARAT Cytb876-847 GCRTAWGCRAAWARRAARTAYCAYTCWGG JB
ND1 ND1DL1 CCTTCWGCAAAATCGAAAGGGGYHCG ND1.YIQIR AAGATCCTTGGATAYATYCARATYCG JB, CEC
ssrRNA 12Sb-3' GAGGGTGACGGGCGGTGTGT 12sa-5' AAACTGGGATTAGATACCCTATTAT JB
lsrRNA 16SBRH CCGGTCTGAACTCAGATCACGT 16SARL CGCCTGTTTATCAAAAACAT 39
Acknowledgements
This work was funded by the UK Biotechnology and Biology Research Council. CEC is currently funded by the UK Natural Environment Research Council. Many thanks to M. Averof for purchasing an individual S. mantis in Crete and to J. Boore for primer sequences and helpful discussions regarding methodology, annotation, and data presentation.
==== Refs
Boore J Complete mitochondrial genome sequence of Urechis caupo, a representative of the phylum Echiura BMC Genomics 2004 5 67 15369601 10.1186/1471-2164-5-67
García-Machado E Pempera M Dennebouy N Oliva-Saurez M Mounolou JC Monnerot M Mitochondrial genes collectively suggest the paraphyly of Crustacea with respect to Insecta J Mol Evol 1999 49 142–149 10368442
Arnason U Adegoke JA Bodin K Born EW Esa YB Gullberg A Nilsson M Short RV Xu X Janke A Mammalian mitogenomic relationships and the root of the eutherian tree Proc Natl Acad Sci U S A 2002 99 8151 8156 12034869 10.1073/pnas.102164299
Nardi F Spinsanti G Boore JL Carapelli A Dallai R Frati F Hexapod origins: monophyletic or paraphyletic? Science 2003 299 1887 1889 12649480 10.1126/science.1078607
Phillips MJ Penny D The root of the mammalian tree inferred from whole mitochondrial genomes Mol Phylogenet Evol 2003 28 171 185 12878457 10.1016/S1055-7903(03)00057-5
Harrison GLA McLenachan PA Phillips MJ Slack KE Cooper A Penny D Four new avian mitochondrial genomes help get to basic evolutionary questions in the late Cretaceous Mol Biol Evol 2004 21 974 983 14739240 10.1093/molbev/msh065
Negrisolo E Minelli A Valle G The mitochondrial genome of the house centipede Scutigera and the monophyly versus paraphyly of myriapods Mol Biol Evol 2004 21 770 780 14963096 10.1093/molbev/msh078
Martin JW Davis GE An Updated Classification of the Recent Crustacea Science Series No 39 2001 Los Angeles, Natural History Museum of Los Angeles County
Caldwell RL Dingle H Stomatopods Scientific American 1976 234 80–89 1251183
Crustikon http://www.tmu.uit.no/crustikon
Fishbase http://fishbase.sinica.edu.tw/
Clary DO Wolstenholme DR The mitochondrial DNA molecule of Drosophila yakuba: nucleotide sequence, gene organization, and genetic code J Mol Evol 1985 22 252 271 3001325
Boore JL Animal mitochondrial genomes Nucleic Acids Res 1999 27 1767 1780 10101183 10.1093/nar/27.8.1767
Machida RJ Miya MU Nishida M Nishida S Complete mitochondrial DNA sequence of Tigriopius japonicus (Crustacea: Copepoda) Mar Biotechnol 2002 4 406—417 14961252 10.1007/s10126-002-0033-x
Ogoh K Ohmiya Y Complete mitochondrial DNA sequence of the sea-firefly, Vargula hilgendorfii (Crustacea, Ostracoda) with duplicate control regions Gene 2004 327 131 139 14960368 10.1016/j.gene.2003.11.011
Lowe TM Eddy SR tRNAscan-SE: a program for improved detection of transfer RNA genes in genomic sequence Nucleic Acids Res 1997 25 955 964 9023104 10.1093/nar/25.5.955
Stewart JB Beckenbach AT Phylogenetic and genomic analysis of the complete mitochondrial DNA sequence of the spotted asparagus beetle Crioceris duodecimpunctata Mol Phylogenet Evol 2003 26 513 526 12644408 10.1016/S1055-7903(02)00421-9
Wilson K Cahill V Ballment E Benzie J The complete sequence of the mitochondrial genome of the crustacean Penaeus monodon: are malacostracan crustaceans more closely related to insects than to branchiopods? Mol Biol Evol 2000 17 863 874 10833192
Yamauchi M Miya M Nishida M Complete mitochondrial DNA sequence of the Japanese spiny lobster, Panulirus japonicus (Crustacea: Decapoda) Gene 2002 295 89 12242015 10.1016/S0378-1119(02)00824-7
Miller AD Nguyen TT Burridge CP Austin CM Complete mitochondrial DNA sequence of the Australian freshwater crayfish, Cherax destructor (Crustacea: Decapoda: Parastacidae): a novel gene order revealed Gene 2004 331 65 72 15094192 10.1016/j.gene.2004.01.022
Brown WM RJ MI The mitochondrial genome of animals Molecular Evolutionary Genetics 1985 New York, Plenum 95–130
Goddard JM Wolstenholme DR Origin and direction of replication in mitochondrial DNA molecules from the genus Drosophila Nucleic Acids Res 1980 8 741–757 6253922
Shao R Barker SC The highly rearranged mitochondrial genome of the plague thrips, Thrips imaginis (Insecta: Thysanoptera): convergence of two novel gene boundaries and an extraordinary arrangement of rRNA Genes Mol Biol Evol 2003 20 362 370 12644556 10.1093/molbev/msg045
Zhang DX Hewitt GM Nuclear integrations: challenges for mitochondrial DNA markers Trends in Ecology and Evolution 1996 11 247 251 10.1016/0169-5347(96)10031-8
Lavrov DV Brown WM Boore JL Phylogenetic position of the Pentastomida and (pan)crustacean relationships Proc Biol Sci 2004 271 537–544 15129965
Cook CE Yue Q Akam M Mitochondrial genomes suggest that hexapods and crustaceans are mutually paraphyletic Proc Biol Sci 2005 272 1295–1304 16024395
Boore JL Lavrov DV Brown WM Gene translocation links insects and crustaceans Nature 1998 392 667–668 9565028 10.1038/33577
Boore JL Collins TM Stanton D Daehler LL Brown WM Deducing the pattern of arthropod phylogeny from mitochondrial DNA rearrangements Nature 1995 376 163–165 7603565 10.1038/376163a0
Rokas A Holland PWH Rare genomic changes as a tool for phylogenetics TREE 2001 15 454–459 11050348
Wright F The effective number of codons used in a gene Gene 1990 87 23–29 2110097 10.1016/0378-1119(90)90491-9
codonw: Correspondence Analysis of Codon Usage
Negrisolo E Minelli A Valle G Extensive gene order rearrangement in the mitochondrial genome of the centipede Scutigera coleoptrata J Mol Evol 2004 58 413 423 15114420 10.1007/s00239-003-2563-x
Lavrov DV Boore JL Brown WM Complete mtDNA sequences of two millipedes suggest a new model for mitochondrial gene rearrangements: duplication and nonrandom loss Mol Biol Evol 2002 19 163 169 11801744
Campbell NJH Barker SC The novel mitochondrial gene arrangement of the cattle tick, Boophilus microplus: fivefold tandem repetition of a coding region Mol Biol Evol 1999 16 732 740 10368952
BLAST NCBI
Server RNASE
Homepage NCBIT http://www.ncbi.nlm.nih.gov/Taxonomy/taxonomyhome.html/
Folmer O Black M Hoeh W Lutz R Vrijenhoek R DNA primers for amplification of mitochondrial cytochrome c oxidase subunit I from diverse metazoan invertebrates Mol Mar Biol Biotechnol 1994 3 294 299 7881515
Palumbi SR Martin A Romano S McMillan WO Stice L Grabowki G The Simple Fool's Guide to PCR 1996 Honolulu, Kewalo Marine Laboratory and University of Hawaii
|
16091132
|
PMC1199596
|
CC BY
|
2021-01-04 16:32:48
|
no
|
BMC Genomics. 2005 Aug 9; 6:105
|
utf-8
|
BMC Genomics
| 2,005 |
10.1186/1471-2164-6-105
|
oa_comm
|
==== Front
BMC Infect DisBMC Infectious Diseases1471-2334BioMed Central London 1471-2334-5-551600017710.1186/1471-2334-5-55Research ArticleSubclinical iron deficiency is a strong predictor of bacterial vaginosis in early pregnancy Verstraelen Hans [email protected] Joris [email protected] Kristien [email protected] Stijn [email protected] Geert [email protected] Marleen [email protected] Department of Obstetrics & Gynaecology, Faculty of Medicine and Health Sciences, Ghent University, Ghent University Hospital, De Pintelaan 185, B-9000 Ghent, Belgium2 Department of Clinical Chemistry, Microbiology and Immunology, Faculty of Medicine and Health Sciences, Ghent University, Ghent University Hospital, De Pintelaan 185, B-9000 Ghent, Belgium3 Department of Intensive Care, Faculty of Medicine and Health Sciences, Ghent University, Ghent University Hospital, De Pintelaan 185, B-9000 Ghent, Belgium2005 6 7 2005 5 55 55 27 1 2005 6 7 2005 Copyright © 2005 Verstraelen et al; licensee BioMed Central Ltd.2005Verstraelen et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Bacterial vaginosis (BV) is the single most common vaginal infection in women of childbearing age and associated with a sizeable infectious disease burden among both non-pregnant and pregnant women, including a significantly elevated risk of adverse pregnancy outcome. Overall, little progress has been made in identifying causal factors involved in BV acquisition and persistence. We sought to evaluate maternal iron status in early pregnancy as a putative risk factor for BV, considering that micronutrients, and iron deficiency in particular, affect the host response against bacterial colonization, even in the setting of mild micronutrient deficiencies.
Methods
In a nested case-control study, we compared maternal iron status at entry to prenatal care (mean gestational age 9.2 ± 2.6 weeks) between eighty women with healthy vaginal microflora and eighteen women with vaginosis-like microflora. Vaginal microflora status was assessed by assigning a modified Nugent score to a Gram-stained vaginal smear. Maternal iron status was assayed by an array of conventional erythrocyte and serum indicators for iron status assessment, but also by more sensitive and more specific indicators of iron deficiency, including soluble transferrin receptors (sTfR) as an accurate measure of cellular and tissue iron deficiency and the iron deficiency log10[sTfR/ferritin] index as the presently most accurate measure of body storage iron available.
Results
We found no statistically significant correlation between vaginal microflora status and routinely assessed iron parameters. In contrast, a highly significant difference between the healthy and vaginosis-like microflora groups of women was shown in mean values of sTfR concentrations (1.15 ± 0.30 mg/L versus 1.37 ± 0.38 mg/L, p = 0.008) and in mean iron deficiency log10[sTfR/ferritin] index values (1.57 ± 0.30 versus 1.08 ± 0.56, p = 0.003), indicating a strong association between iron deficiency and vaginosis-like microflora. An sTfR concentration >1.45 mg/L was associated with a 3-fold increased risk (95%CI: 1.4–6.7) of vaginosis-like microflora and after controlling for maternal age, gestational length, body mass, parity, and smoking habits with an adjusted odds ratio of 4.5 (95%CI: 1.4–14.2).
Conclusion
We conclude that subclinical iron deficiency, presumably resulting from inadequate preconceptional iron supplies, is strongly and independently associated with vaginosis-like microflora during early pregnancy.
==== Body
Background
Bacterial vaginosis (BV) is the single most common vaginal infection in women of childbearing age. Basically, BV involves a shift from the predominant hydrogen peroxide-producing Lactobacillus species, such as L. crispatus and L. jensenii [1-3] to a polymicrobial flora that includes gram-variable and gram-negative anaerobes such as Gardnerella vaginalis, Prevotella spp., and Mobiluncus spp. [4,5], and more recently associated gram-positive anaerobes such as Peptostreptococcus spp. [6] and Atopobium vaginae [3,7-9].
Besides a nuisance problem causing vaginal discomfort [10], BV is also associated with a sizeable disease burden [11]. In pregnancy, the presence of vaginosis-like microflora especially during early gestation has been consistently and strongly associated with spontaneous preterm labour and preterm prelabour rupture of membranes, and hence, BV is a major determinant to the prematurity-related disease burden [12]. In addition, though not considered a conventionally defined sexually transmitted infection (STI), BV strongly predisposes to the acquisition of pandemic STIs, such as Neisseria gonorrhoeae [13], Chlamydia trachomatis [13], and most notably HIV-1 among both pregnant [14,15] and non-pregnant women [16-19].
A few factors are known to increase the risk of BV, including demographic factors such as black ethnicity, and behavioural factors including smoking, vaginal douching, IUD contraceptive use, and sexual behaviour-related factors [20]. Overall, little progress has been made however in identifying causal factors involved in BV acquisition and recurrence [21,22]. In particular, there is a striking dearth of data on intrinsic or biological risk factors for BV. Nonetheless, since BV is not deemed a traditionally defined STI, it is all the more likely that intrinsic factors elicit a pivotal role in the acquisition of disturbed vaginal microflora and, in some women, are underlying the apparent instability of the vaginal ecosystem. In particular, it is increasingly assumed that subtle differences in the type and magnitude of the host-pathogen response at the level of the vaginal mucosa may explain the differential susceptibility to perturbation of the vaginal niche [23]. Some very recent findings in the field of the innate vaginal mucosal immune response, including the demonstration of genetic differences, such as Toll 4-like receptor gene mutations [24], and of phenotypic differences such as impaired expression of anti-inflammatory cytokines [25], being strongly correlated with the susceptibility to bacterial vaginosis are unique examples to the above paradigm.
From this perspective, we hypothesized that micronutrient status during early pregnancy may represent yet another putative biological risk factor for BV, considering that micronutrients, and iron deficiency in particular, affect the host response against bacterial colonization [26], even in the setting of mild micronutrient deficiencies [27]. Moreover, maternal iron deficiency has been consistently associated with adverse pregnancy outcome [28,29] and it has therefore been reiterated that micronutrient status during early pregnancy warrants further scrutiny even among well-nourished women from high-income countries [28].
Methods
Study population and design
As part of a prospective cohort study basically involving the study of vaginal microflora during early pregnancy in relation to pregnancy outcome, we conducted a nested case-control study comprising 115 unselected pregnant women, which were consecutively enrolled on the occasion of their first antenatal visit, between March 3 and November 6, 2003 at the outpatient obstetric clinic of the Ghent University Hospital. The Ghent University Hospital Ethical Board approved the study protocol and all study subjects agreed to participate through written informed consent.
Sample collection
Vaginal samples were collected for the purpose of vaginal microflora status assessment by inserting a sterile cotton-tipped wooden swab into the vagina. The swab was rolled round through 360 degrees against the vaginal wall at the midportion of the vault and carefully withdrawn to prevent contamination. Swabs were then smeared on a plain glass slide and air-dried at room temperature. The slides were Gram stained and assigned a modified Nugent score [4] according to Ison and Hay [30]. Accordingly, Gram-stained vaginal smears were initially categorized as normal (grade I), intermediate (grade II), and bacterial vaginosis (grade III). To the purpose of the present study, we subsequently pooled the latter two categories into a single category unless otherwise specified, and therefore further denote two vaginal microflora status categories: normal or healthy microflora (corresponding to grade I microflora or a Nugent score 0–3) and disturbed or bacterial vaginosis-like microflora (corresponding to grade II and III or a Nugent score 4–10).
Blood samples were drawn within one hour following vaginal swabbing and processed within four hours. A first venous blood sample was allowed to clot, and centrifuged (1000 g for 10 minutes at room temperature). The supernatant serum was collected for analysis. Serum ferritin, soluble transferrin receptors (sTfR), and C-reactive protein (CRP) were assayed by fixed-time immunonephelometry using commercial rabbit anti-human antisera on a BN II nephelometer (Dade Behring), calibrated against the CRM 470 certified reference material. Serum transferrin concentration was assessed by immunoturbidimetry using commercial reagents on a Modular P analyzer (Roche Diagnostics). Serum iron concentration was measured by spectrophotometry (ferrozine method). Serum transferrin saturation (TS) was calculated as TS (%) = [serum iron (μmol/L)/serum transferrin (g/L)] × 3.98.
Additional indices of iron deficiency were calculated following log10 transformation of serum ferritin and sTfR concentrations, including the log10 [sTfR]/log10[ferritin] and log10 [sTfR/ferritin] indices [31]. These combined measures have recently evolved as highly specific and sensitive measures of iron deficiency [31]. In particular, the logarithm of the ratio of the soluble transferrin receptor to ferritin concentration (log10 [sTfR/ferritin]) is currently the most precise measure of body storage iron available [32,33].
A second venous blood sample was simultaneously collected in EDTA tubes to assess plasma haemoglobin (Hgb), red blood cell count (RBC) and haematocrit (Hct), mean corpuscular volume (MCV) and mean corpuscular haemoglobin (MCH) (Sysmex SE-9500; Toa Medical Electronics).
All women had a vaginal ultrasound (US) scan and if embryonal biometry was inconsistent with the calculated gestational length, the latter was adjusted according to US biometry. Body mass index or BMI (kg/m2) was calculated as weight (kg)/[length (m)] 2 after standardized assessment of maternal weight and length at entry to prenatal care. All other clinical data were collected in a routine manner.
Exclusion criteria
Exclusion criteria were first attendance beyond fourteen completed weeks of gestation, multiple gestation, and pre-existing maternal systemic disease, e.g. diabetes mellitus. Women who were unacquainted with the Dutch language could not be included because we had no means to obtain informed consent from these in a proper way.
Definitions
For initial assessment of subjects, iron deficiency and anaemia were defined according to the WHO criteria of serum ferritin ≤ 12 g/L and haemoglobin < 11 g/dL, respectively. Iron deficiency was further investigated by the use of more sensitive indicators, in particular maternal sTfR concentrations and the log10 [sTfR/ferritin] index; however, no reference values for pregnant women are available at present for these parameters.
Statistical analyses
Following the one-sample Kolmogorov-Smirnov test procedure for each variable, we established that all continuous variables could be analysed under the parametric assumption (p-value to the Z-statistic >0.5), except maternal serum CRP. Means are presented as arithmetic means and standard deviation to the mean. Means between two groups were compared with the independent samples t test. Strength of bivariate correlations was expressed as Pearson correlation coefficients (R). Strength of association was calculated as prevalence or risk ratios (RR) in a univariate analysis and (adjusted) odds ratios (OR) in a multivariable analysis with 95% confidence intervals (CI) and p-values to the 95% CI. Multivariable analysis was performed using a stepwise binary logistic regression model and likelihood ratio tests were used to compare different models. For any reported measure, statistical significance was accepted, as the two-tailed probability level was <0.05.
All statistical analyses were performed using the statistical software package SPSS v12.0 (Chicago, Illinois).
Results
Study population
We excluded seventeen women (17/115 or 14.8%) who had given their consent for participation in the study, from the analysis, because they refused part of the sampling procedure, or because of incomplete or inadequate sampling, including delay between vaginal and blood sampling, or because of early (missed) abortion. As a group, the subjects excluded did not differ significantly from the remainder of women in terms of baseline characteristics (mean maternal age, mean body mass index, smoking habits, mean gestational age, and median parity and gravidity).
From each patient included for the final analysis (98/115 or 85.2%) a vaginal swab and venous blood samples were obtained at a single point in time at a mean gestational age of 9.2 ± 2.6 weeks. Eighty women had normal or grade I microflora on Gram stain and eighteen women presented with disturbed or vaginosis-like microflora, i.e. intermediate (grade II) microflora or overt bacterial vaginosis (grade III) microflora.
Basic clinical characteristics of study participants, who were all of white Caucasian origin, are displayed in table 1. Women with disturbed vaginal flora in the index pregnancy were significantly more likely to have delivered a child previously and tended to have a higher body mass index (Table 1).
Table 1 Basic clinical characteristics of the study population.
Healthy vaginal microflora (n = 80) Disturbed vaginal microflora (n = 18) p-value
Maternal age (years) – mean ± SD 30.5 ± 4.7 30.2 ± 5.7 0.8
Gestational age (weeks) – mean ± SD 9.3 ± 2.6 8.7 ± 2.8 0.4
Body mass index (kg/m2) – mean ± SD 22.7 ± 4.0 24.9 ± 5.8 0.05
Body mass index (kg/m2) – % (n)
<25 77.5% (62) 50.0% (9) 0.04
≥25 22.5% (18) 50.0% (9)
Parity – % (n)
0 43.8% (35) 16.7% (3) 0.04
≥1 56.3% (45) 83.3% (15)
Gravidity – % (n)
0 35.0% (28) 11.1% (2) 0.05
≥1 65.0% (52) 88.9% (16)
Smoking – % (no)
Yes 11.1% (2) 22.2% (4) 0.1
No 88.9% (16) 77.8% (14)
Healthy vaginal flora is defined as grade I or lactobacilli-dominated microflora on Gram stain (corresponding to a Nugent score 0 – 3) and disturbed vaginal flora is defined as grade II and grade III flora, this is mixed or gram-negative rods-dominated microflora on Gram stain (corresponding to a Nugent score ≥4). Body mass index (BMI) refers to the BMI at entry to prenatal care and is calculated as weight (kg)/[length (m)]2.
Traditional indicators of iron deficiency in relation to vaginal microflora status
In this study sample, 10.2% of subjects (10/98) had depleted iron stores during early pregnancy according to the conventional criterion of serum ferritin ≥ 12 g/L. 4.1% of women (4/98) presented with anaemia defined as Hgb < 11 g/dL and 10.2% with a Hgb concentration below 12 g/dL (10/98), though only one of these (1/98 or 1.0%) had true iron-deficient anaemia when accounting for both serum ferritin (≥ 12 g/L) and haemoglobin (< 11 g/dL). Any other traditional compound indices of iron-deficiency, e.g. by accounting for mean cell volume, mean corpuscular haemoglobin, and transferrin saturation did not identify any additional cases of clinical overt iron deficiency.
There was no statistically significant correlation between vaginal microflora status as assessed by Gram stain and any of the conventionally assessed iron and red blood cell indices, including red blood cell counts, serum haemoglobin, haematocrit, mean corpuscular volume, mean corpuscular haemoglobin, serum iron, serum ferritin, serum transferrin and transferritin saturation (Table 2). Similarly, we observed no significant association between vaginal microflora status and maternal CRP concentration during early pregnancy. In this cohort there were no cases of overt systemic inflammation according to maternal CRP, nor were ferritin and transferrin concentrations significantly correlated with CRP.
Table 2 Maternal serum iron and red blood cell indices according to vaginal microflora status. Healthy vaginal flora is defined as grade I or lactobacilli-dominated microflora on Gram stain (corresponding to a Nugent score 0 – 3) and disturbed vaginal flora is defined as grade II and grade III flora, this is mixed or gram-negative rods-dominated microflora on Gram stain (corresponding to a Nugent score ≥4). RBC denotes red blood cell counts, Hgb plasma haemoglobin concentration, Hct hematocrit, MCV mean corpuscular volume of red blood cells, MCH mean corpuscular haemoglobin mass of red blood cells, Fe serum iron concentration, and sTfR (serum) soluble transferrin receptor concentration, respectively.
Healthy vaginal microflora (n = 80) Disturbed vaginal microflora (n = 18) p-value
RBC (×106/μL) 4.3 ± 0.4 4.4 ± 0.4 0.5
HgB (g/dL) 13.1 ± 0.9 13.1 ± 1.1 0.8
Hct (%) 38.8 ± 2.6 39.4 ± 3.2 0.4
MCV (fL) 89.4 ± 4.6 89.1 ± 5.3 0.8
MCH (pg/cell) 30.2 ± 1.8 29.7 ± 1.5 0.3
Fe (μg/dL) 113.0 ± 39.1 105.2 ± 43.3 0.5
Ferritin (mg/L) 46.2 ± 35.9 52.5 ± 56.6 0.6
Transferrin (mg/dL) 2.8 ± 0.5 2.9 ± 0.5 0.7
Transferrin saturation (%) 29.5 ± 12.1 26.7 ± 12.3 0.4
sTfR (mg/L) 1.15 ± 0.30 1.37 ± 0.38 0.01
Log10[sTfR]/log10[ferritin] 0.04 ± 0.12 0.10 ± 0.15 0.04
Log10[sTfR/ferritin] 1.57 ± 0.30 1.08 ± 0.56 0.003
Soluble transferrin receptors in relation to vaginal microflora status
In contrast to the above, we observed a trend by which maternal serum transferrin receptor (sTfR) concentrations during early pregnancy were negatively correlated with lactobacillary grading and hence positively correlated with the degree of vaginal microflora alteration (R = 0.26, p = 0.01). When vaginal microflora status was handled as a dichotomous variable ('healthy' versus 'disturbed'), women with healthy vaginal microflora (n = 80) during early pregnancy had a mean sTfR concentration of 1.15 ± 0.30 mg/L as compared to a mean sTfR of 1.37 ± 0.38 mg/L among women (n = 18) with disturbed microflora (p = 0.008) (Figure 1).
Figure 1 Distribution of sTfR concentrations according to vaginal microflora status. Box-and-whisker plots of the sTfR distributions according to vaginal microflora status during early pregnancy. The thick line represents the median sTfR value, the horizontal box lines the 25th percentile and 75th percentile, and the outer short horizontal lines the boundaries of the sTfR range. Healthy vaginal flora is defined as grade I or lactobacilli-dominated microflora on Gram stain (corresponding to a Nugent score 0 – 3) and disturbed vaginal flora is defined as grade II and grade III flora, this is mixed or gram-negative rods-dominated microflora on Gram stain (corresponding to a Nugent score ≥4).
Given the significant overlap in sTfR distributions for the healthy and disturbed vaginal microflora groups of women, classification plots were constructed and the sTfR value with the highest discriminative value between both groups was chosen as the sTfR cut-off level. Serum transferrin receptor concentrations > 1.45 mg/L were associated with a prevalence or risk ratio of 3.0 (95% CI: 1.4 – 6.7, p = 0.014) for disturbed vaginal flora. The accuracy of the sTfR assay at this threshold was 79% (95% CI: 0.69–0.86). Sensitivity and the positive predictive value (PPV) were however low and estimated at 39% (95% CI: 0.18–0.64) and 41% (95% CI: 0.19–0.67), respectively. In contrast, specificity and consequently the negative predictive value (NPV) of the assay were as high as 88% (95% CI: 0.78–0.94) and 86% (95% CI: 0.77–0.93), respectively.
Of the parameters entered into in the multivariable analysis only smoking, body mass index and gravidity/parity were expected to act as true confounders of the association under study considering these variables impinge both on serum transferrin receptor concentrations and on vaginal microflora status. We also controlled for maternal age, gestational age at sampling, and CRP as covariates in the model. Yet, the only significant variable retained from the multivariable analysis was the maternal sTfR concentration, suggesting that raised sTfR concentrations during early pregnancy are independently associated with vaginal microflora alteration. The adjusted odds ratio of an sTfR-concentration > 1.45 mg/L for vaginosis-like vaginal microflora was 4.5 (95%CI: 1.4–14.2, p = 0.011).
The log10 [sTfR/ferritin] iron deficiency index in relation to vaginal microflora status
Since sTfR concentrations may reflect both rates of erythropoiesis as well as cellular iron needs, several combined measures following log transformation of sTfR and/or ferritin have been used as highly specific and sensitive measures of iron deficiency in particular, including sTfR/log10[ferritin], log10[sTfR]/log10[ferritin], and log10[sTfR/ferritin] [31].
The logarithm of the ratio of the soluble transferrin receptor to ferritin concentration (log10 [sTfR/ferritin]) is currently the most precise measure of body storage iron available [32,33]. We found a highly significant difference (difference in means = 0.49, 95% CI 0.30–0.68, p < 0.0001) between mean values of the log10 [sTfR/ferritin] index (1.57 ± 0.30 versus 1.08 ± 0.56) between both groups (Table 2) further indicating that the observed association between maternal sTfR concentrations and vaginal microflora status indeed relates to depleted available body iron stores and cellular iron avidity.
Though our sample size was rather small, post hoc analysis revealed that the above difference in the mean iron deficiency index values (log10[sTfR/ferritin]) between both groups was demonstrated at a two-sided significance level of α = 0.05 with a statistical power of more than 90% (1-β = 92.7) under the parametric assumption and by accounting for unequal variances.
Supplemental iron intake as a potential determinant of iron indices
Of note is that 65.3% of women in this cohort (64/98) were already taking oligo-elements or vitamin supplements at the time of sampling, most often as combined preparations (35/98 or 35.7%) or as folate supplements (27/98 or 27.6%). We found no significant association between iron supplementation (37/98 or 37.8%) and maternal sTfR concentrations (p = 0.95) or with the combined sTfR-ferritin indexes following log transformation (p = 0.20 to 0.97), while there was a marginally significant association between serum ferritin concentrations and supplemental iron intake (p = 0.053).
Discussion
The association of subclinical iron deficiency with vaginosis-like microflora
We found that maternal serum concentrations of soluble transferrin receptors during early pregnancy were positively correlated with decreased lactobacillary grading and hence with degree of vaginal microflora alteration. An sTfR concentration above 1.45 mg/L, this is approximately one SD (0.33 mg/L) above the mean sTfR concentration (1.19 mg/L), was associated with a 3-fold increased risk of vaginosis-like microflora (RR = 3.0, 95%CI: 1.4–6.7), and the risk did not change when accounting for potential differences in the distributions of maternal age, gestational age, body mass index, parity, CRP, and smoking habits.
Albeit sTfR is considered a marker of both iron status and erythropoiesis, sTfR acts as a marker of erythropoiesis only when iron stores are adequate and sTfR additionally becomes a marker of iron status in the setting of tissue iron deficiency with or without adequate iron stores [31]. In this cohort of pregnant women, the sTfR distribution among women presenting with disturbed vaginal flora was significantly skewed to the right partly in the absence of overt iron deficiency as measured by serum ferritin. We therefore accounted for the log10 [sTfR/ferritin] index, which is considered the most accurate measure of body storage iron available [32,33] and documented a highly significant difference in the log10 [sTfR/ferritin] distributions between the normal and disturbed vaginal microflora groups of women. We therefore believe that the observed risk of vaginosis-like microflora associated with increased sTfR concentrations can reliably be attributed to insufficiently available body storage iron and cellular or tissue iron deficiency during early pregnancy.
Documenting the association between iron and vaginal microflora status during pregnancy depends on the accuracy of the iron assays applied
Serum transferrin receptor synthesis is up-regulated in iron-deprived tissues and it can therefore be argued that assessment of iron status at the tissue level is of more functional importance when examining the effects of iron depletion on disease occurrence than conventional assessment of iron stores. Several recent studies have shown that sTfR is a very sensitive and specific index of iron deficiency during pregnancy [34,35], that sTfR assaying is superior to conventional methods for the assessment of iron status [32,33] and that the accuracy with which iron deficiency can be diagnosed is further increased by combining sTfR and ferritin [31-33]. In contrast, conventional indicators of iron status, such as red blood cell indices and markers of iron transport tend to be less sensitive or are altered by gestation independently of iron status [31,34], and therefore also have low specificity. Markers of iron storage and transport, ferritin and transferrin, may also act as acute phase reactants which may further intricate the interpretation of their values with regard to iron status, though there was no evidence of confounding by inflammation in our study.
Consequently, it may not be surprising that the array of conventional iron and red blood cell assays that were performed in our study did not pick up the association with vaginal microflora status. Ferritin is however, a very early marker in the setting of inadequate iron supplies and could therefore reasonably be expected to reflect the observed association as well, though not obvious from this study. A possible explanation is that our study lacked the power to demonstrate the effect of iron deficiency as assessed by ferritin alone, due to its biological variability, considering serum ferritin shows a wide range of values within the normal range.
If anything, it cannot be ignored that when accounting for both ferritin and soluble transferrin receptor concentrations in the log10 [sTfR/ferritin] index as an established highly accurate measure of body storage iron, the difference between the healthy and disturbed vaginal microflora groups of women became even more apparent than when accounting for sTfR alone.
Iron deficiency in early pregnancy may ensue from critical preconceptional iron supplies which are further compromised by iron restriction during early pregnancy
Soluble transferrin concentrations have been shown to be steadily low during early pregnancy [31,34,36,37] and relatively decreased as compared to prepregancy concentrations by some [31] or at least not significantly elevated from the non-pregnant state until the second trimester by others [37]. This has been attributed to blunted erythropoietin production resulting in decreased erythropoiesis [36] with low peripheral reticulocyte counts during the first pregnancy trimester [37].
Decreased erythropoiesis has in turn been considered to concur with decreased iron requirements during early pregnancy due to cessation of menstrual losses [38]. Yet, the biological significance of this relative erythropoietic quiescence has not been fully explained.
It is likely that the disproportionate increases in maternal plasma volume and red cell mass leading to the physiologic anaemia of pregnancy from early gestation until term represent an important hemodynamic and hemostatic protective feature of normal pregnancy [39]. Several studies have demonstrated unfavourable pregnancy outcomes associated with high Hgb concentrations early in pregnancy, as well as in situations where Hgb concentrations fail to decline in the mid-trimester [40]. Consequently, the relative erythropoietic slowdown during early pregnancy might be targeted at rapid enhancement of such hemodilution. For instance, by the end of the first trimester, maternal plasma volume has expanded by some 15% on average [41], though not paralleled by a concomitant boost in erythropoietin release [36], as would be expected [42].
Of note is that intestinal iron absorption during the first trimester is also reduced [38], while rapidly increasing after that time with the amount of dietary iron transferred to the foetus regulated in response to maternal iron status at the level of the gut [43]. It is therefore plausible that restrictive iron absorption during early pregnancy at the time of critical processes such as placental development and organogenesis also concurs with other protective mechanisms, notably anti-oxidant and anti-infectious defence [28,40].
Against this background, it is conceivable that women entering pregnancy with impending iron deficiency may share a conflict of interest between these potentially protective iron- and erythropoiesis-restricting mechanisms on one hand, and insufficient iron availability to comply with increased basic metabolic rates and increased iron and oxygen demands on the other hand. It has been recognized that the high gestational iron needs during the second and third trimesters are met by increasing intestinal iron absorption rates beyond the first trimester up to term, but also by mobilizing available prepregnancy reserves [34]. High STfR concentrations during early pregnancy among a subset of women in this cohort may therefore reflect cellular iron deficiency as a result of subclinical or latent prepregnancy iron shortage.
Iron deficiency and the host response against vaginal bacterial colonization
As it comes down to the biological plausibility of the observed association, it is conceivable that iron deficiency may affect both innate and cellular immune responsiveness [26,27] at the level of the vaginal mucosa, though we did not identify any previous study on micronutrient status and bacterial vaginosis. The significant association of iron deficiency anaemia with infection has however extensively been demonstrated and most importantly, these effects have been attributed to the adverse effects of iron deficiency on the immune system, even in the setting of mild micronutrient deficiencies [26,27,29]. Interestingly, our observations on the association between iron deficiency and bacterial vaginosis concur with the consistent association between maternal iron deficiency in early pregnancy and a greater risk of preterm delivery [28,29] on one hand, and with the strong and consistent association between bacterial vaginosis in early pregnancy and preterm delivery [11,12,20] on the other hand.
Limitations of the study
Our results should be taken with caution considering our sample size was limited. Therefore our findings undeniably need to be confirmed in much larger, prospective cohort studies, preferably including non-pregnant women as well. Though the present case-control study was actually nested within a larger cohort study, failure to consistently follow-up patients enrolled at entry to prenatal care hampered any further conclusions being drawn that might be of interest to our results presented above.
As to confounding, we were able to control for most established confounders that may impinge on vaginal microflora status and on sTfR concentrations, including age, gestational length, BMI, smoking habits, and parity. Among these body mass index and parity warrant particular scrutiny, considering these variables were significantly associated with vaginal microflora status in our series, while also being known determinants of maternal sTfR concentrations. Maternal sTfR concentrations were however not significantly correlated with BMI (p = 0.4). The correlation between sTfR and parity was also not apparent from our data, this is, when parity was handled as the raw, categorical variable (p = 0.2), yet the correlation became highly significant when parity was handled as a binary variable (0.005). Therefore, it cannot be ignored that the observed association between maternal sTfR concentrations and vaginal microflora status concurs with the association between sTfR and parity. In particular, we found that multiparous women had on average a significantly higher sTfR (p = 0.003) and were also significantly more likely to have disturbed vaginal microflora (p = 0.04) as compared to nulliparous women. Though these interactions were cancelled in the multivariable analysis by the correlation between sTfR and vaginal microflora status, it still needs to be considered that parity may act as a confounder to the former association. If anything, as it is only plausible that parity affects mean sTfR rather than sTfR concentrations determining parity, the most conceivable explanation would be that sTfR concentrations are the explanatory variable to the association between parity and vaginal microflora status. Ethnicity was not a confounder to our study as all women were of white Caucasian origin. We did not collect any data on sexual behaviour-related characteristics nor on vaginal douching, which have consistently been associated with BV, and therefore it cannot be precluded that differential sexual behaviour and differences in use of vaginal hygiene products between both groups may have confounded our results at least to some extent.
It should also be acknowledged that, owing to the design of the study, we are also unaware of the timing of the exposure relative to the timing of the outcome. There is good evidence from the literature however, that the sTfR concentrations at these early gestational ages most likely reflect preconceptional sTfR concentrations, as discussed above.
Finally, at the sTfR cut-off level chosen, the strength of the association was convincingly strong and this was further reflected by the high specificity and the high NPV indicating that above this threshold, women with disturbed vaginal microflora were overrepresented among subjects with a high sTfR and hence with impaired iron availability or tissue iron needs. On the contrary, though subjects with sTfR concentrations below the threshold represented the preponderance of women with normal vaginal microflora, a substantial proportion of the women with disturbed vaginal microflora also had no evidence of iron deficiency, and therefore the sensitivity and the PPV were actually low. This observation is not unexpected to the extent BV – much alike most conditions, is thought of as a multifactorial condition that obviously can occur in the absence of iron deficiency.
Iron supplementing and bacterial vaginosis: an opportunity for preconceptional prevention of adverse pregnancy outcome?
We could not establish a significant relationship between iron supplementing and various indicators of iron deficiency in this population with the exception of a marginally significant association with ferritin values, but this finding should be taken with caution considering we had no information on duration of iron supplementing preceding iron status assessment. However, since study subjects were enrolled at their first antenatal visit, the time span between commencement of iron supplementing and iron status assessment must have been rather short. If anything, our study lacked the power to substantiate such an association considering the prevalence of iron deficiency was fairly low and rates of supplemental iron intake rather high. Previous studies did demonstrate a significant decrease of sTfR concentrations following iron administration among both pregnant [34] and non-pregnant subjects [44].
From the above notions on iron metabolism in early pregnancy it may be inferred however, that adequate preconceptional iron stores and therefore preconceptional iron supplementing rather than first trimester supplements may better serve the goal of preventing BV-associated adverse pregnancy outcome and preterm birth in particular. Of note is that in animal models, preconceptional nutrional status has recently also been associated with non-infectious preterm birth [45].
Conclusion
We conclude that subclinical iron deficiency, presumably resulting from inadequate preconceptional iron supplies, is strongly and independently associated with vaginosis-like microflora during early pregnancy, after accounting for maternal age, gestational length, body mass index, parity, CRP, and smoking habits as potential confounders. The strong association between tissue iron deficiency with vaginosis-like flora during early pregnancy was documented by assessment of highly sensitive and specific markers, in particular maternal soluble transferrin receptors and the log10 [sTfR/ferritin] iron deficiency index, while the association was not apparent from conventional iron and red blood cell indicators.
These findings need to be confirmed in larger, prospective cohort studies, preferably including non-pregnant women as well. If so, the association of latent, subclinical or functional iron deficiency with bacterial vaginosis may be of paramount interest to the primary and secondary prevention of bacterial vaginosis and bacterial vaginosis-associated disease, including preterm birth.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
HV and MT participated in the development of the study design, the collection of the study samples, the collection, analysis and interpretation of the data, and in the writing of the report. JD participated in the development of the study design, the analysis of the blood samples, the interpretation of the data, and provided important intellectual content to the manuscript. KR and SB participated in the collection of the study samples, the collection of the data, and provided important intellectual content to the manuscript. GC participated in the development of the study design, conducted the analysis of all vaginal samples, and provided important intellectual content to the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This work was supported through a research grant by the Marguerite-Marie Delacroix Foundation. As the main funding source, the Marguerite-Marie Delacroix Foundation was not involved in the development of the study design; the collection, analysis, and interpretation of the data; in the writing of the report; nor in the decision to submit the paper for publication.
==== Refs
Hillier SL Krohn MA Rabe LK Klebanoff SJ Eschenbach DA The normal vaginal flora, H2O2-producing lactobacilli, and bacterial vaginosis in pregnant women Clin Infect Dis 1993 273 81 8399882
Antonio MA Hawes SE Hillier SL The identification of vaginal Lactobacillus species and the demographic and microbiologic characteristics of women colonized by these species J Infect Dis 1999 180 1950 6 10558952 10.1086/315109
Verhelst R Verstraelen H Claeys G Verschraegen G Delanghe J Van Simaey L De Ganck C Temmerman M Vaneechoutte M Cloning of 16S rRNA genes amplified from normal and disturbed vaginal microflora suggests a strong association between Atopobium vaginae, Gardnerella vaginalis and bacterial vaginosis BMC Microbiol 2004 4 16 15102329 10.1186/1471-2180-4-16
Nugent RP Krohn MA Hillier SL Reliability of diagnosing bacterial vaginosis is improved by a standardized method of gram stain interpretation J Clin Microbiol 1991 29 297 301 1706728
Hillier SL Krohn MA Klebanoff SJ Eschenbach DA The relationship of hydrogen peroxide-producing lactobacilli to bacterial vaginosis and genital microflora in pregnant women Obstet Gynecol 1992 79 369 73 1738516
Delaney ML Onderdonk AB Nugent score related to vaginal culture in pregnant women Obstet Gynecol 2001 98 79 84 11430961 10.1016/S0029-7844(01)01402-8
Ferris MJ Masztal A Aldridge KE Fortenberry JD Fidel PL JrMartin DH Association of Atopobium vaginae, a recently described metronidazole resistant anaerobe, with bacterial vaginosis BMC Infect Dis 2004 4 5 15018635 10.1186/1471-2334-4-5
Burton JP Devillard E Cadieux PA Hammond JA Reid G Detection of Atopobium vaginae in postmenopausal women by cultivation-independent methods warrants further investigation J Clin Microbiol 2004 42 829 31 14766863 10.1128/JCM.42.4.1829-1831.2004
Verstraelen H Verhelst R Claeys G Temmerman M Vaneechoutte M Culture-independent analysis of vaginal microflora: The unrecognized association of Atopobium vaginae with bacterial vaginosis Am J Obstet Gynecol 2004 191 1130 2 15507931 10.1016/j.ajog.2004.04.013
Anderson MR Klink K Cohrssen A Evaluation of vaginal complaints JAMA 2004 291 1368 79 15026404 10.1001/jama.291.11.1368
Sobel JD Bacterial vaginosis Annu Rev Med 2000 51 349 56 10774469 10.1146/annurev.med.51.1.349
Leitich H Bodner-Adler B Brunbauer M Kaider A Egarter C Husslein P Bacterial vaginosis as a risk factor for preterm delivery: a meta-analysis Am J Obstet Gynecol 2003 189 139 47 12861153 10.1067/mob.2003.339
Wiesenfeld HC Hillier SL Krohn MA Landers DV Sweet RL Bacterial vaginosis is a strong predictor of Neisseria gonorrhoeae and Chlamydia trachomatis infection Clin Infect Dis 2003 36 663 8 12594649 10.1086/367658
Taha TE Gray RH Kumwenda NI Hoover DR Mtimavalye LA Liomba GN Chiphangwi JD Dallabetta GA Miotti PG HIV infection and disturbances of vaginal flora during pregnancy J Acquir Immune Defic Syndr Hum Retrovirol 1999 20 52 59 9928730
Royce RA Thorp J Granados JL Savitz DA Bacterial vaginosis associated with HIV infection in pregnant women from North Carolina J Acquir Immune Defic Syndr Hum Retrovirol 1999 20 382 6 10096583
Sewankambo N Gray RH Wawer MJ Paxton L McNaim D Wabwire-Mangen F Serwadda D Li C Kiwanuka N Hillier SL Rabe L Gaydos CA Quinn TC Konde-Lule J HIV-1 infection associated with abnormal vaginal flora morphology and bacterial vaginosis Lancet 1997 350 546 550 9284776 10.1016/S0140-6736(97)01063-5
Taha TE Hoover DR Dallabetta GA Kumwenda NI Mtimavalye LA Yang LP Liomba GN Broadhead RL Chiphangwi JD Miotti PG Bacterial vaginosis and disturbances of vaginal flora: association with increased acquisition of HIV AIDS 1998 12 1699 1706 9764791 10.1097/00002030-199813000-00019
Martin H Nyange PM Richardson BA Martin HL Richardson BA Nyange PM Lavreys L Hillier SL Chohan B Mandaliya K Ndinya-Achola JO Bwayo J Kreiss J Vaginal lactobacilli, microbial flora, and risk of human immunodeficiency virus type 1 and sexually transmitted disease acquisition J Infect Dis 1999 180 1863 8 10558942 10.1086/315127
Hashemi FB Ghassemi M Roebuck KA Spear GT Activation of human immunodeficiency type I expression by Garnerella vaginalis J Infect Dis 1999 179 924 10068588 10.1086/314674
Morris M Nicoll A Simms I Wilson J Catchpole M Bacterial vaginosis: a public health review BJOG 2001 108 439 50 11368127 10.1016/S0306-5456(00)00124-8
Hay PE Recurrent bacterial vaginosis Dermatol Clin 1998 16 769 73 9891678
Wilson J Managing recurrent bacterial vaginosis Sex Transm Infect 2004 80 8 11 14755028 10.1136/sti.2002.002733
Genc MR Onderdonk A Witkin SS Innate Immune System Gene Polymorphisms in Women with Vulvovaginal Infections Curr Infect Dis Rep 2004 6 462 468 15538984
Genc MR Vardhana S Delaney ML Onderdonk A Tuomala R Norwitz E Witkin SS MAP Study Group Relationship between a toll-like receptor-4 gene polymorphism, bacterial vaginosis-related flora and vaginal cytokine responses in pregnant women Eur J Obstet Gynecol Reprod Biol 2004 116 152 6 15358455 10.1016/j.ejogrb.2004.02.010
Simhan H Caritis S Hillier S Krohn M Cervical anti-inflammatory cytokine concentrations are decreased among pregnant women with bacterial vaginosis [abstract] Am J Obstet Gynecol 2004 11 15295338 10.1016/j.ajog.2004.09.060
Bendich A Micronutrients in women's health and immune function Nutrition 2001 17 858 67 11684393 10.1016/S0899-9007(01)00649-9
Bhaskaram P Immunobiology of mild micronutrient deficiencies Br J Nutr 2001 75 80 11227036
Scholl TO Reilly T Anemia, iron and pregnancy outcome J Nutr 2000 443 7
Allen LH Anemia and iron deficiency: effects on pregnancy outcome Am J Clin Nutr 2000 1280 4
Ison CA Hay PE Validation of a simplified grading of Gram stained vaginal smears for use in genitourinary medicine clinics Sex Transm Infect 2002 78 413 5 12473800 10.1136/sti.78.6.413
Beguin Y Soluble transferrin receptor for the evaluation of erythropoiesis and iron status Clin Chim Acta 2003 329 9 22 12589962 10.1016/S0009-8981(03)00005-6
Skikne BS Flowers CH Cook JD Serum transferrin receptor: a quantitative measure of tissue iron deficiency Blood 1990 75 1870 6 2331526
Cook JD Flowers CH Skikne BS The quantitative assessment of body iron Blood 2003 101 3359 64 12521995 10.1182/blood-2002-10-3071
Akesson A Bjellerup P Berglund M Bremme K Vahter M Soluble transferrin receptor: longitudinal assessment from pregnancy to postlactation Obstet Gynecol 2002 99 260 6 11814507 10.1016/S0029-7844(01)01684-2
Akesson A Bjellerup P Berglund M Bremme K Vahter M Serum transferrin receptor: a specific marker of iron deficiency in pregnancy Am J Clin Nutr 1998 68 1241 6 9846853
Beguin Y Lipscei G Thoumsin H Fillet G Blunted erythropoietin production and decreased erythropoiesis in early pregnancy Blood 1991 78 89 93 2070061
Choi JW Im MW Pai SH Serum transferrin receptor concentrations during normal pregnancy Clin Chem 2000 46 725 7 10794761
Bothwell TH Iron requirements in pregnancy and strategies to meet them Am J Clin Nutr 2000 72 257 264
Yip R Significance of an abnormally low or high hemoglobin concentration during pregnancy: special consideration of iron nutrition Am J Clin Nutr 2000 272 9
Casanueva E Viteri FE Iron and oxidative stress in pregnancy J Nutr 2003 1700 8
Bernstein IM Ziegler W Badger GJ Plasma volume expansion in early pregnancy Obstet Gynecol 2001 97 669 72 11339913 10.1016/S0029-7844(00)01222-9
Vora M Gruslin A Erythropoietin in obstetrics Obstet Gynecol Surv 1998 53 500 8 9702790 10.1097/00006254-199808000-00023
O'Brien KO Zavaleta N Abrams SA Caulfield LE Maternal iron status influences iron transfer to the fetus during the third trimester of pregnancy Am J Clin Nutr 2003 77 924 30 12663293
Radtke H Tegtmeier J Rocker L Salama A Kiesewetter H Daily doses of 20 mg of elemental iron compensate for iron loss in regular blood donors: a randomized, double-blind, placebo-controlled study Transfusion 2004 44 1427 32 15383014 10.1111/j.1537-2995.2004.04074.x
Bloomfield FH Oliver MH Hawkins P Campbell M Phillips DJ Gluckman PD Challis JR Harding JE A periconceptional nutritional origin for noninfectious preterm birth Science 2003 300 606 12714735 10.1126/science.1080803
|
16000177
|
PMC1199597
|
CC BY
|
2021-01-04 16:28:15
|
no
|
BMC Infect Dis. 2005 Jul 6; 5:55
|
utf-8
|
BMC Infect Dis
| 2,005 |
10.1186/1471-2334-5-55
|
oa_comm
|
==== Front
BMC Infect DisBMC Infectious Diseases1471-2334BioMed Central London 1471-2334-5-671612020910.1186/1471-2334-5-67Research ArticleSafety and tolerability of nevirapine-based antiretroviral therapy in HIV-infected patients receiving fluconazole for cryptococcal prophylaxis: a retrospective cohort study Manosuthi Weerawat [email protected] Nopphanath [email protected] Achara [email protected] Somnuek [email protected] Bamrasnaradura Institute, Ministry of Public Health, Nonthaburi, 11000, Thailand2 Faculty of Medicine Ramathibodi Hospital, Mahidol University, Bangkok, 10400, Thailand2005 24 8 2005 5 67 67 29 5 2005 24 8 2005 Copyright © 2005 Manosuthi et al; licensee BioMed Central Ltd.2005Manosuthi et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
To compare the adverse events after initiation of nevirapine-based ART among HIV-infected patients who did not receive fluconazole (group A), received fluconazole 400 mg/week (group B), and received fluconazole 200 mg/day (group C).
Methods
A retrospective cohort study was conducted among HIV-infected patients who began NVP-based ART between December 2003 and September 2004. Patients were followed up for 6 months. Clinical hepatitis, elevated aspartate aminotransferase (AST) and alanine aminotransferase (ALT) (> 3 times from baseline), and skin rashes were studied.
Results
There were 686 patients; 225, 392, and 69 patients in group A, B, and C, respectively. Baseline characteristics including age, previous opportunistic infections, use of antituberculous drugs, and baseline aminotransferase levels among the three groups were similar. Group C had a higher proportion of men (p = 0.016). Baseline median (IQR) CD4 cell counts were 85 (21–159), 18 (7–48), and 16 (5–35) cell/mm3 in group A, B, and C, respectively (p < 0.001). Of 2/225 (0.9%), 4/392 (1.0%), and 0/69 (0%) patients in group A, B, and C developed clinical hepatitis (p = 0.705). There were no significant difference of elevated AST or ALT among the three groups (p > 0.05). By logistic regression, receiving fluconazole was not predictive of clinical hepatitis, elevated aminotransferase, or skin rashes. At 6 months after initiating NVP, 174 (77.3%) patients in group A, 309 (78.8%) patients in group B, and 58 (84.1%) patients in group C remained on NVP.
Conclusion
Initiation of NVP-based ART among Thais with advance HIV disease receiving fluconazole is safe and well-tolerated. nevirapine should not be contraindicated for patients receiving fluconazole for treatment or prophylaxis of cryptococcosis.
==== Body
Background
Highly active antiretroviral therapy (HAART) has been widely used for the treatment of human immunodeficiency virus (HIV) infected patients with successful immune restoration and reductions in morbidity and mortality [1,2]. However, access to antiretroviral therapy for HIV-infected patients in resource-limited countries is still a major obstacle [3,4]. HIV-infected patients in these areas often presented with advanced HIV disease. Nevirapine (NVP) is a non-nucleoside reverse transcriptase inhibitor (NNRTI) that has shown efficacy even in advanced HIV disease [5-8]. NVP-based HAART has been widely used in the resource-limited countries because of its efficacy, relative low cost, and availability. Additionally, NVP is part of two of the four World Health Organization-recommended generic combinations for the 3 × 5 program in resource-limited countries [9]. NVP-related adverse events particularly skin rashes and hepatotoxicity have been well recognized as the limitation of NVP use [10-12]. NVP-associated skin rashes usually appear after one to four weeks of treatment [10]. The risk of hepatotoxicity is greatest in the first six weeks of treatment and continues through 18 weeks of treatment [13].
Fluconazole is a commonly used for the prophylaxis and treatment of cryptococcosis in HIV-infected patients. Fluconazole 400 mg/day has been used as standard therapy for treatment of cryptocoocal meningitis after completion of two weeks of amphotericin B [12]. Fluconazole 200 mg/day is used for secondary prophylaxis until immune reconstitution occurs with HAART [14]. Fluconazole 400 mg/week is recommended for primary prophylaxis among HIV-infected patients with CD4 cell count <100 cells/mm3 in Thailand and some developing countries because of a high prevalence of cryptococcosis and the evidence of a survival benefit [15]. Nonetheless, the potential drug-drug interaction between NVP and fluconazole is a major concern. Both drugs can induce cytochrome P450 isoenzymes in the liver [16-18]. Geel et al recently reported that fluconazole significantly raised plasma NVP levels and may cause serious hepatotoxicity [19]. So far, there have not been enough data or any recommendations to adjust NVP dosage for the concomitant use of both drugs in order to avoiding adverse events. A previous study demonstrated that genetic disposition may play a role in NVP hypersensitivity reactions [20]. There is little data of safety and tolerability for concurrent use of NVP and fluconazole in Asian populations. We therefore conducted this retrospective cohort study to compare the frequencies of hepatotoxicity and skin rashes among HIV-infected patients who received fluconazole and subsequently received NVP-based HAART regimens.
Methods
A retrospective cohort study was conducted among antiretroviral-naïve HIV-infected patients who began a NVP-based HAART regimens between December 2003 and September 2004 at Bamrasnaradura Institute, Ministry of Public Health, Nonthaburi, Thailand. Inclusion criteria were as follows: (1) HIV-infected patients > 15 years of age, (2) naïve to antiretroviral therapy, (3) began antiretroviral therapy with a NVP-based HAART regimen, (4) used NVP 200-mg once-daily lead-in dose, prior to escalation to 200 mg twice daily. Each eligible patient was classified into one of three groups according to whether he or she received fluconazole or not and the different dosages of fluconazole as follows: not received fluconazole (group A), received fluconazole 400 mg/week (group B), and received fluconazole 200 mg/day (group C). All patients were followed for 6 months after initiating NVP-based HAART. The primary objective was to compare the frequency of clinical hepatitis among the three groups. The secondary objectives of interest were to compare the frequencies of elevated aspartate aminotransferase (AST) and alanine aminotransferase (ALT) > three times from baseline at six months after initiation of NVP and to compare the frequencies skin rashes among the three groups.
The patients' clinical statuses including clinical hepatitis and skin rashes were assessed monthly by the attending physician. Liver functions test was performed at 6 months after initiating NVP in case of no evidence of hepatitis. Clinical hepatitis was defined as symptomatic hepatitis and either AST or ALT elevated > five times the upper limit of normal. Signs and symptoms of hepatitis may include anorexia, malaise, jaundice, nausea/vomiting, hepatomegaly, and hepatic tenderness. Skin rash was determined by medical record of skin reactions including diffuse erythematous or maculopapular rash, urticarial rash, serum sickness-like reaction, Stevens-Johnson syndrome, and toxic epidermal necrolysis.
Mean (± standard deviation, SD), median (interquartile range, IQR) and frequencies (%) were used to describe patients' characteristics in each treatment groups. Chi-square test and Kruskal-Walis test were used to compare categorical and continuous variables among the treatment groups, respectively. Logistic regression was used to determine predicting factors for each adverse event of interest. The possible risk factors were entered into the model of binary logistic regression. Spearman correlation was use to study the relationship between skin rashes and clinical hepatitis, skin rashes and aminotransferase levels. All analyses were performed using SPSS program version 11.5. A P value less than 0.05 was considered statistically significant. The study was approved from the institute review board.
Results
A total of 686 patients were eligible and included in the study. Of 686 patients; 225, 392, and 69 patients were classified into group A, B, and C, respectively. The baseline characteristics of patients prior to initiation of HAART are described and compared among the three groups as shown in Table 1. Baseline characteristics including age, previous major opportunistic infections (OIs), use for antituberculous drugs therapy, baseline AST and ALT, and baseline HIV RNA prior to HAART were not different among the three groups. Group C patients had a greater proportion of men than the other groups (p = 0.016). Group B and group C patients had more advanced HIV disease [i.e., lower body weight (p = 0.008), lower baseline CD4 cell counts prior to initiation of HAART (p < 0.001), and higher proportion receiving co-trimoxazole(p < 0.001)] than patients in group A.
Table 1 Baseline characteristic of 686 patients receiving NVP-based HAART.
Baseline characteristics Not receiving fluconazole Receiving fluconazole P value
(n = 225) 400 mg/week (n = 392) 200 mg/day (n = 69)
Gender: Male 116 (51.6%) 213 (54.3%) 49 (71.0%) 0.016
Female 109 (48.4%) 179 (45.7%) 20 (29.0%)
Age, years, mean ± SD 36.0 ± 7.5 35.0 ± 7.4 34.7 ± 8.03 0.200
Weight, kg., mean ± SD 54.9 ± 10.6 52.3 ± 10.0 51.4 ± 8.7 0.008
Previous opportunistic infections, patients (%)
Any major opportunistic infections 53 (23.6%) 120 (30.6%) 24 (34.8%) 0.368
Tuberculosis 32 (14.2%) 64 (16.3%) 17 (24.6%) 0.124
PCP 31 (13.8%) 47 (12.0%) 8 (11.6%) 0.787
Histoplasmosis 3 (1.3%) 1 (0.3%) - 0.190
Penicilliosis 3 (1.3%) 0 (0.0%) - 0.046
Toxoplasmosis - 5 (1.3%) - 0.151
MAC 2 (0.9%) 2 (0.5%) - 0.669
Concurrent medication
Cotrimoxazole 176 (78.2%) 363 (93.3%) 57 (82.6%) < 0.001
Antituberculous drugs 47 (20.9%) 82 (20.9%) 22 (31.9%) 0.095
Baseline CD4 cell count, cell/mm3, median (IQR) 85 (21–159) 18 (7–48) 16 (5–35) < 0.001
Baseline % CD4, median (IQR) 7 (2–11) 2 (1–5) 2 (1–5) < 0.001
Baseline plasma HIV RNA, 207,500 270,000 169,000 0.095
copies/ml, median (IQR) (68,375–549,500) (118,000–545,000) (76,750–415,000)
Baseline AST, U/L, median (IQR) 31 (23–46.5) 32 (25–53.2) 34 (24–53) 0.358
Baseline ALT, U/L, median (IQR) 23 (17–39) 28 (17–45) 32 (17–49.5) 0.177
Baseline TB, mg/dl, median (IQR) 0.5 (0.4–0.65) 0.5 (0.4–0.7) 0.5 (0.4–0.8) 0.843
One hundred and four of 255 (40.8%) in group A, 188 of 392 (48.0%) in group B, and 36 of 69 (52.2%) in group C had AST testing at 6 months after initiating NVP. Seventy one of 255 (29.4%), 95 of 392 (24.2%), and 22 of 69 (31.9%) in corresponding groups had ALT testing at 6 months after initiating NVP. The prevalence of patients who developed adverse events including clinical hepatitis, AST and ALT increase > 3 times from baseline, and skin rashes are shown in Table 2. The prevalence of clinical hepatitis, elevated AST, and elevated ALT were not different among the three groups (p > 0.05). There was a significant difference among the groups in the prevalence of skin rashes (p = 0.016).
Table 2 Frequency of adverse events among the three groups.
Adverse events Not receiving fluconazole Receiving fluconazole P value
(n = 225) 400 mg/week (n = 392) 200 mg/day (n = 69)
Clinical hepatitis 2/225 (0.9%) 4/392 (1.0%) 0/69 (0%) 0.705
Elevated AST > 3 times from baseline 3/104 (2.9%) 5/188 (2.7%) 3/36 (8.3%) 0.192
Elevated ALT > 3 times from baseline 6/71 (8.5%) 7/95 (7.4%) 2/22 (9.1%) 0.938
Skin Rash 23/225 (10.2%) 27/392 (6.9%) 0/69 (0%) 0.016
Results of logistic regression analysis indicated that receiving anti-tuberculous drugs was a significant predictor of clinical hepatitis (p = 0.025) (Table 3). Gender, body weight, baseline CD4 cell counts, or receiving fluconazole were not predictive of clinical hepatitis. Logistic regression analyses of models including gender, body weight, baseline CD4 cell count, receiving anti-tuberculous drugs, and receiving fluconazole, were not significant predictors of elevated AST > 3 times from baseline (χ2 (5, N = 328) = 3.72, p = 0.592) and elevated ALT > 3 times from baseline (χ2 (5, N = 188) = 7.26, p = 0.298). Logistic regression analysis of skin rashes indicated that receiving anti-tuberculous drugs was a significant predictor of skin rashes (Table 4). Gender, body weight, baseline CD4 cell counts, or receiving fluconazole were not predictive of skin rashes.
Table 3 Logistic regression of possible risk factors and clinical hepatitis
Risk factors Odd Ratios 95% confidence interval P value
Female gender 1.25 0.21–7.49 0.809
Body weight 1.08 0.98–1.17 0.076
Baseline CD4 cell count 0.99 0.97–1.01 0.352
Receiving anti-tuberculous drugs 7.42 1.27–43.31 0.026
Receiving fluconazole 1.12 0.18–6.90 0.901
Receiving Cotrimoxazole 0.84 0.21–5.44 0.675
Table 4 Logistic regression of possible risk factors and skin rashes.
Factors Odd Ratios 95% confidence interval P value
Female gender 1.26 0.67–2.39 0.470
Body weight 1.02 0.98–1.05 0.347
Baseline CD4 cell count 1.00 0.99–1.01 0.873
Receiving anti-tuberculous drugs 2.96 1.04–8.47 0.043
Receiving fluconazole 1.56 0.79–3.08 0.201
Receiving Cotrimoxazole 1.21 0.68–2.91 0.594
The results of correlation analyses show that there was no correlation between skin rashes and clinical hepatitis (rs = 0.026, p = 0.491), skin rashes and AST (rs = 0.003, p = 0.948), or skin rashes and ALT (rs = 0.034, p = 0.637). The tolerability of NVP-based HAART is shown in Table 5. Regarding NVP-associated skin rashes, the majority (> 90%) of patients developed diffuse maculopapular rashes. Only 1 of 25 patient in group A and 2 of 26 patients in groups B developed Steven-Johnson syndromes. Nine patients died, 1 in group A and 8 in group B. The majority of patients died from opportunistic infections (i.e., 2 cryptococcosis, 2 bacterial pneumonitis, 1 Mycobacterium avium complex infection, and 3 wasting syndromes). No patient died from hepatitis.
Table 5 Tolerability of NVP-based HAART among the three groups.
Tolerability outcomes Not receiving fluconazole Receiving fluconazole P value
(n = 225) 400 mg/week (n = 392) 200 mg/day (n = 69)
On NVP-based HAART at six months 174 (77.3%) 309 (78.8%) 58 (84.1%) 0.532
Reasons of discontinuation 0.448
Adverse events 26 (11.5%) 29 (7.4%) 3 (4.3%)
Skin rashes 25/26 (96.2%) 26/29 (89.7%) 3 of 3 (100.0%)
Hepatotoxicity 1/26 (2.8%) 3/29 (10.3%) -
Death 1 (0.4%) 8 (2.0%) -
Loss to follow-up 20 (8.9%) 39 (9.9%) 7 (10.1%)
Refer 4 (1.8%) 7 (1.8%) 1 (1.4%)
Discussion
In the present study, we have shown that the prevalence of clinical hepatitis and AST and ALT elevation 6 months after initiating HAART among HIV-infected patients who concurrently received NVP and fluconazole were not different from those who received NVP- based HAART without fluconazole. These findings suggest that administration of NVP-based HAART regimens in HIV-infected patients who receive fluconazole dose not increase the risk of hepatotoxicity. The prevalence of clinical hepatitis and aminotransferase elevations in our study was lower than that reported in studies conducted in African HIV-infected patients [19,20]. There may be several explanations for this discrepancy. First, the patients in this study had very low CD4 cell counts (i.e., 97% had CD4 cell counts < 200 cells/mm3 and 80% had CD4 cell counts < 100 cells/mm3). High CD4 cell count is a predictor of hepatotoxicity [11,19]. Second, there was a higher proportion of men in our study than the previous studies [17,18]. CD4 cell counts are more predictive of hepatotoxicity among men than women (i.e., men who have a CD4 cell counts > 400 cells/mm3 are at a greater risk of hepatitis than women with CD4 cell counts > 250 cells/mm3) [13,21]. Third, our study included Asians, which was different from the previous study [11,17,18]. Sanne et al reported that African women may have an increased likelihood of hepatotoxicity [20]. The previous studies demonstrate that there are differences in drug and alcohol metabolism that may be related to genetic or environmental factors (e.g., differences in diet and protein intake) [22-24]. There is little data on differences in NVP metabolism based on genetic disposition.
While logistic regression analyses in the present study show that receiving fluconazole was not predictive for any adverse events, receiving anti-tuberculous drugs was significantly predictive for both clinical hepatitis and skin rashes. The patients in our large cohort had overall 16% of concurrent tuberculosis and the results from the regression analyses can demonstrate this important issue. These findings point out that concurrent concomitant use of NVP and anti-tuberculous drugs may a cause of more concern. We suggest to closely monitoring liver function test in patients with concomitant use of NVP and anti-tuberculous drugs. Further well-designed prospective study is needed.
Regarding NVP-associated skin rashes, the overall prevalence in the present study was similar to previous study in Thai patients with advanced HIV disease [25]. The incidence of NVP-associated skin rashes in the patients who received Fluconazole was lower than the other groups. This may be explained by the lower CD4 cells count in this group. As the previous studies have demonstrated that persons with a higher baseline CD4 cell counts are likely to develop rash from NVP [26]. Given there was no correlation between skin rashes and clinical hepatitis or elevated aminotransferase, these two entities of adverse events may need different approaches for monitoring and management.
Overall about 80% of the patients in our cohort continued on NVP-based HAART at six months. Although all patients were initiated with NVP 200-mg once-daily lead-in dose prior to escalation to 200 mg twice daily, skin rash was still the major adverse event led to discontinuation of NVP-based HAART. The previous studies using short-course prednisolone were failed to prevent NVP-associated skin rashes [26,27]. The further study to minimize this major limitation is needed.
NVP-based HAART is a common regimen which widely used for treatment of HIV-infected patients in resource-limited countries due to its affordability. Until the other options are more accessible, NVP-based HAART is still a key regimen to scale up treatment of HIV in resource-limited countries. Our study may provide the safety data of concomitant use of NVP and fluconazole and support the use of NVP in patients receiving fluconazole for treatment or prophylaxis of cryptococcal diseases.
The limitation of the present study is the nature of retrospective study that patients' clinical status may be underreported. First, not all patients had been checked for aminotransferase level during the first 6 months. Only 328 and 188 patients had been checked for AST and ALT levels, respectively. However, we had analyzed the demographics and adverse events between patients who had and did not have AST or ALT levels and found that there were no statistically significant differences. We did not have data of plasma NVP level in this cohort due to retrospective nature. However, Dailly et al reported that there was no association between plasma NVP level and hepatotoxicity [28]. Ultimately, the present study did not assess patients who initiated NVP while received fluconazole 400 mg/day.
Conclusion
In summary, receiving of fluconazole for treatment or prophylaxis of cryptococcosis does not increase the risk of hepatotoxicity, elevated aminotransferase, or skin rashes from NVP. Initiation of NVP-based HAART among HIV-infected patients receiving fluconazole is safe and well-tolerated. Use of anti-tuberculous drugs is predictive of clinical hepatitis and skin rashes. Initiation of NVP-based HAART in HIV-infected patients receiving anti-tuberculous needs closely monitoring and prompt management.
List of abbreviations
HAART: Highly active anti-retroviral therapy, HIV: Human immunodeficiency virus, NVP: Nevirapine, NNRTI: Non-nucleoside reverse transcriptase inhibitor, AST: Aspartate aminotransferase, ALT: Alanine aminotransferase
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
WM participated in the design of the study. NC performed statistical analysis. AC participated in the design of the study. SS participated in the design of the study and statistical analysis.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We would like to thank all the study patients in this study and all the attending staffs in Department of Medicine, Bamrasnaradura Institute. The authors would like to thank Michael Martin for his comments.
==== Refs
Hogg RS Heath KV Yip B Craib KJ O'Shaughnessy MV Schechter MT Montaner JS Improved survival among HIV-infected individuals following initiation of antiretroviral therapy Jama 1998 279 450 454 9466638 10.1001/jama.279.6.450
Palella FJJ Delaney KM Moorman AC Loveless MO Fuhrer J Satten GA Aschman DJ Holmberg SD Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. HIV Outpatient Study Investigators N Engl J Med 1998 338 853 860 9516219 10.1056/NEJM199803263381301
Dionisio D Esperti F Messeri D Vivarelli A Priority strategies for sustainable fight against HIV/AIDS in low-income countries Curr HIV Res 2004 2 377 393 15544458 10.2174/1570162043351057
Vermund SH Powderly WG Developing a human immunodeficiency virus/acquired immunodeficiency syndrome therapeutic research agenda for resource-limited countries: a consensus statement Clin Infect Dis 2003 37 S4 12 12822127 10.1086/375367
Floridia M Bucciardini R Ricciardulli D Fragola V Pirillo MF Weimer LE Tomino C Giannini G Galluzzo CM Andreotti M Cargnel A Alberici F De Rienzo B Leoncini F Fiaccadori F Francisci D Grillone W Ortona L Piazza M Scalzini A Nigra E Tumietto F Vella S A randomized, double-blind trial on the use of a triple combination including nevirapine, a nonnucleoside reverse transcriptase HIV inhibitor, in antiretroviral-naive patients with advanced disease J Acquir Immune Defic Syndr Hum Retrovirol 1999 20 11 19 9928724
Manosuthi W Sungkanuparph S Vibhagool A Rattanasiri S Thakkinstian A Nevirapine- versus efavirenz-based highly active antiretroviral therapy regimens in antiretroviral-naive patients with advanced HIV infection HIV Med 2004 5 105 109 15012650 10.1111/j.1468-1293.2004.00195.x
Podzamczer D Ferrer E Consiglio E Gatell JM Perez P Perez JL Luna E Gonzalez A Pedrol E Lozano L Ocana I Llibre JM Casiro A Aranda M Barrufet P Martinez-Lacasa J Miro JM Badia X Casado A Lupo S Cahn P Manos M Estela J A randomized clinical trial comparing nelfinavir or nevirapine associated to zidovudine/lamivudine in HIV-infected naive patients (the Combine Study) Antivir Ther 2002 7 81 90 12212928
van Leth F Phanuphak P Ruxrungtham K Baraldi E Miller S Gazzard B Cahn P Lalloo UG van der Westhuizen IP Malan DR Johnson MA Santos BR Mulcahy F Wood R Levi GC Reboredo G Squires K Cassetti I Petit D Raffi F Katlama C Murphy RL Horban A Dam JP Hassink E van Leeuwen R Robinson P Wit FW Lange JM Comparison of first-line antiretroviral therapy with regimens including nevirapine, efavirenz, or both drugs, plus stavudine and lamivudine: a randomised open-label trial, the 2NN Study Lancet 2004 363 1253 1263 15094269 10.1016/S0140-6736(04)15997-7
Scaling up Antiretroviral Therapy in Resource-limited Settings: Treatment Guidelines for a public Health Approach 2003 Geneva, Joint United Nations Programme on HIV/AIDS (UNAIDS) and World Health Organization (WHO)
Milinkovic A Martinez E Nevirapine in the treatment of HIV Expert Rev Anti Infect Ther 2004 2 367 373 15482202 10.1586/14787210.2.3.367
Pollard RB Robinson P Dransfield K Safety profile of nevirapine, a nonnucleoside reverse transcriptase inhibitor for the treatment of human immunodeficiency virus infection Clin Ther 1998 20 1071 1092 9916603 10.1016/S0149-2918(98)80105-7
Knudtson E Para M Boswell H Fan-Havard P Drug rash with eosinophilia and systemic symptoms syndrome and renal toxicity with a nevirapine-containing regimen in a pregnant patient with human immunodeficiency virus Obstet Gynecol 2003 101 1094 1097 12738113 10.1016/S0029-7844(02)02620-0
Stern JO Robinson PA Love J Lanes S Imperiale MS Mayers DL A comprehensive hepatic safety analysis of nevirapine in different populations of HIV infected patients J Acquir Immune Defic Syndr 2003 34 Suppl 1 S21 33 14562855
Saag MS Graybill RJ Larsen RA Pappas PG Perfect JR Powderly WG Sobel JD Dismukes WE Practice guidelines for the management of cryptococcal disease. Infectious Diseases Society of America Clin Infect Dis 2000 30 710 718 10770733 10.1086/313757
Chetchotisakd P Sungkanuparph S Thinkhamrop B Mootsikapun P Boonyaprawit P A multicentre, randomized, double-blind, placebo-controlled trial of primary cryptococcal meningitis prophylaxis in HIV-infected patients with severe immune deficiency HIV Med 2004 5 140 143 15139978 10.1111/j.1468-1293.2004.00201.x
Albengres E Le Louet H Tillement JP Systemic antifungal agents. Drug interactions of clinical significance Drug Saf 1998 18 83 97 9512916
Kuper JI D'Aprile M Drug-Drug interactions of clinical significance in the treatment of patients with Mycobacterium avium complex disease Clin Pharmacokinet 2000 39 203 214 11020135
Joly V Yeni P [Non-nucleoside reverse transcriptase inhibitors] Ann Med Interne (Paris) 2000 151 260 267 10922953
Geel J Pitt J Orrell CJ Van Dyk M Wood R The effect of fluconazole on nevirapine pharmacokinetics: July 11-16; Bangkok. 2004 1 369
Sanne I Mommeja-Marin H Hinkle J Bartlett JA Lederman MM Maartens G Wakeford C Shaw A Quinn J Gish RG Rousseau F Severe hepatotoxicity associated with nevirapine use in HIV-infected subjects J Infect Dis 2005 191 825 829 15717255 10.1086/428093
Reisler K High hepatotoxicity rate seen among HAART patients AIDS Alert 2001 16 118 119 11547496
Stranges S Freudenheim JL Muti P Farinaro E Russell M Nochajski TH Trevisan M Greater hepatic vulnerability after alcohol intake in African Americans compared with Caucasians: a population-based study J Natl Med Assoc 2004 96 1185 1192 15481746
Ma B Yeo W Hui P Ho WM Johnson PJ Acute toxicity of adjuvant doxorubicin and cyclophosphamide for early breast cancer -- a retrospective review of Chinese patients and comparison with an historic Western series Radiother Oncol 2002 62 185 189 11937245 10.1016/S0167-8140(02)00003-8
Critchley JA Nimmo GR Gregson CA Woolhouse NM Prescott LF Inter-subject and ethnic differences in paracetamol metabolism Br J Clin Pharmacol 1986 22 649 657 3567011
Anekthananon T Ratanasuwan W Techasathit W Sonjai A Suwanagool S Safety and efficacy of a simplified fixed-dose combination of stavudine, lamivudine and nevirapine (GPO-VIR) for the treatment of advanced HIV-infected patients: a 24-week study J Med Assoc Thai 2004 87 760 767 15521230
Montaner JS Cahn P Zala C Casssetti LI Losso M Hall DB Wruck J McDonough M Gigliotti M Robinson PA Randomized, controlled study of the effects of a short course of prednisone on the incidence of rash associated with nevirapine in patients infected with HIV-1 J Acquir Immune Defic Syndr 2003 33 41 46 12792354
Knobel H Miro JM Domingo P Rivero A Marquez M Force L Gonzalez A De Miguel V Sanz J Boix V Blanco JL Locutura J Failure of a short-term prednisone regimen to prevent nevirapine-associated rash: a double-blind placebo-controlled trial: the GESIDA 09/99 study J Acquir Immune Defic Syndr 2001 28 14 18 11579272
Dailly E Billaud E Reliquet V Breurec S Perre P Leautez S Jolliet P Bourin M Raffi F No relationship between high nevirapine plasma concentration and hepatotoxicity in HIV-1-infected patients naive of antiretroviral treatment or switched from protease inhibitors Eur J Clin Pharmacol 2004 60 343 348 15156302 10.1007/s00228-004-0769-5
|
16120209
|
PMC1199598
|
CC BY
|
2021-01-04 16:28:15
|
no
|
BMC Infect Dis. 2005 Aug 24; 5:67
|
utf-8
|
BMC Infect Dis
| 2,005 |
10.1186/1471-2334-5-67
|
oa_comm
|
==== Front
BMC MicrobiolBMC Microbiology1471-2180BioMed Central London 1471-2180-5-491611149310.1186/1471-2180-5-49Research ArticleIdentification of a new family of enzymes with potential O-acetylpeptidoglycan esterase activity in both Gram-positive and Gram-negative bacteria Weadge Joel T [email protected] John M [email protected] Anthony J [email protected] Department of Molecular and Cellular Biology, University of Guelph, Guelph, Ontario N1G 2W1 Canada2005 19 8 2005 5 49 49 27 5 2005 19 8 2005 Copyright © 2005 Weadge et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The metabolism of the rigid bacterial cell wall heteropolymer peptidoglycan is a dynamic process requiring continuous biosynthesis and maintenance involving the coordination of both lytic and synthetic enzymes. The O-acetylation of peptidoglycan has been proposed to provide one level of control on these activities as this modification inhibits the action of the major endogenous lytic enzymes, the lytic transglycosylases. The O-acetylation of peptidoglycan also inhibits the activity of the lysozymes which serve as the first line of defense of host cells against the invasion of bacterial pathogens. Despite this central importance, there is a dearth of information regarding peptidoglycan O-acetylation and nothing has previously been reported on its de-acetylation.
Results
Homology searches of the genome databases have permitted this first report on the identification of a potential family of O-Acetylpeptidoglycan esterases (Ape). These proteins encoded in the genomes of a variety of both Gram-negative and Gram-positive bacteria, including a number of important human pathogens such as species of Neisseria, Helicobacter, Campylobacter, and Bacillus anthracis, have been organized into three families based on amino acid sequence similarities with family 1 being further divided into three sub-families. The genes encoding these proteins are shown to be clustered with Peptidoglycan O-acetyltransferases (Pat) and in some cases, together with other genes involved in cell wall metabolism. Representative bacteria that encode the Ape proteins were experimentally shown to produce O-acetylated peptidoglycan.
Conclusion
The hypothetical proteins encoded by the pat and ape genes have been organized into families based on sequence similarities. The Pat proteins have sequence similarity to Pseudomonas aeruginosa AlgI, an integral membrane protein known to participate in the O-acetylation of the exopolysaccaride, alginate. As none of the bacteria that harbor the pat genes produce alginate, we propose that the Pat proteins serve to O-acetylate peptidoglycan which is known to be a maturation event occurring in the periplasm. The Ape sequences have amino acid sequence similarity to the CAZy CE 3 carbohydrate esterases, a family previously known to be composed of only O-acetylxylan esterases. They are predicted to contain the α/β hydrolase fold associated with the GDSL and TesA hydrolases and they possess the signature motifs associated with the catalytic residues of the CE3 esterases. Specific signature sequence motifs were identified for the Ape proteins which led to their organization into distinct families. We propose that by expressing both Pat and Ape enzymes, bacteria would be able to obtain a high level of localized control over the degradation of peptidoglycan through the attachment and removal of O-linked acetate. This would facilitate the efficient insertion of pores and flagella, localize spore formation, and control the level of general peptidoglycan turnover.
==== Body
Background
Peptidoglycan is an essential component of most eubacterial cell walls (the exceptions are the mycobacteria). It provides osmotic stability by serving as an exo-skeleton through which cellular shape is imparted. The tensile strength of this multilayered polymer is provided by two types of covalent linkages. The first consists of β-1,4 glycosidic bonds between N-acetylmuramic acid (MurNAc) and N-acetylglucosamine (GlcNaC), while the second occurs through cross-linking of the tetrapeptides associated with MurNAC residues on adjacent glycan strands (Figure 1).
Peptidoglycan biosynthesis occurs in three stages (reviewed in [1]). The first involves the synthesis of soluble precursors in the cytoplasm leading to the production of UDP-linked MurNAc-pentapeptide. This muropeptide is then transferred to the membrane carrier bactoprenol via a pyrophosphate group and a GlcNAc residue is added to produce the membrane-associated, lipid II precursor. By an unknown mechanism, this lipid-linked precursor is translocated to the outer face of the cytoplasmic membrane. Now exposed to the periplasm in this third stage of biosynthesis, the available muropeptide is polymerized into the existing peptidoglycan sacculus, a reaction catalyzed by the transglycosylase domain of the class A high-molecular weight penicillin-binding proteins (PBPs). Once polymerized, the transpeptidase domain of both the class A and class B high-molecular weight PBPs catalyze the formation of cross-links between the peptides of neighboring glycan strands.
The peptidoglycan sacculus which surrounds the entire bacterium is not a static structure as it must permit the growth and division of the bacterial cell [1]. Furthermore, insertion of cell wall structures that transverse the peptidoglycan layer, such as secretion/transport complexes, flagella, and pili, require the sacculus to be constantly remodeled and reinforced [2]. These processes are performed by the coordinated action of peptidoglycan synthesizing PBPs and the peptidoglycan lytic enzymes. The peptidoglycan cleaving enzymes are numerous and there exists a specific lytic activity for each linkage in peptidoglycan. The amidases cleave the stem peptide from the MurNAc residues, glucosaminidases hydrolyze the β-1,4 linkage between GlcNAc and MurNAc residues, and both the lytic transglycosylases and muramidases cleave the other β-1,4 glycosidic bond that exists between MurNAc and GlcNAc (reviewed in [3]). Collectively, these enzymes are called autolysins because their uncontrolled activity will lead to the lysis of bacterial cells. Given the potential lethal activity of the autolysins, the cell has to control their activity. This is accomplished, at least in part, by associating these enzymes in a holoenzyme complex that includes both the lytic enzymes, which provide sites for the insertion of new peptidoglycan precursors, and the synthesizing PBPs. The major glycolytic enzyme associated with these holoenzyme complexes appears to be the lytic transglycosylases [1]. These enzymes are not hydrolases but cleave the β-1,4 linkage between GlcNAc and MurNAc with the concomitant formation of 1,6-anhydromuramoyl residues [4] (Fig. 1).
Given the inherent importance of peptidoglycan, it is not surprising that lysozyme, one of the most important components of the first line innate immune response of eukaryotic hosts to invading bacteria, targets this polymer. Hydrolysis of the β-1,4 glycosidic linkage between GlcNAc and MurNAc destabilizes the integrity of the sacculus and ultimately leads to cell lysis. A number of pathogenic bacteria are able to avoid the potentially lethal effects of lysozyme through the addition of an acetate group to the C-6 position of MurNAc (reviewed in [5,6]) (Figure 1). These pathogens include both Gram-positive and Gram-negative bacteria, such as Staphylococcus aureus and Neisseria gonorrhoeae, respectively. Several studies have shown that a direct correlation exists between the extent of O-acetylation and susceptibility to lysozyme-catalyzed hydrolysis of peptidoglycan [7-11]. However, the O-acetylation of peptidoglycan has also been shown to preclude the action of the endogenous lytic transglycosylases [12]. Intuitively, this would be expected given that the mechanism of action of these latter enzymes requires a free C-6 hydroxyl group on MurNAc to produce their reaction product, the corresponding 1,6-anhydro derivative. This has prompted the proposal of two roles for peptidoglycan O-acetylation; one to protect the bacterium from a host immune response, while the second would be to regulate the action of endogenous autolysins [13].
There is a dearth of information concerning the pathway for the O-acetylation of peptidoglycan, and nothing is known about its subsequent removal. All available evidence indicates that O-acetylation is a maturation event, occurring in the periplasm after the cross-linking of newly incorporated muropeptides (reviewed in [5,6]). Two models have been proposed to account for the transfer of acetate from cytoplasmic pools of acetyl-CoA to muramoyl residues in the peptidoglycan sacculus [6]. The first is a two protein system involving an integral membrane protein responsible for the translocation of acetate from the cytoplasm to the periplasm where it is accepted by a second more soluble protein which transfers it specifically to the C-6 hydroxyl of muramoyl residues in peptidoglycan. This system is modeled on the proposed pathway for the O-acetylation of the exopolysaccharide, alginate, produced by the opportunistic human pathogen Pseudomonas aeruginosa [14,15]. The other model invokes a single-enzyme system, exemplified by the action of Nod X for the O-acetylation of the carbohydrate-based nodulation factor by Rhizobium leguminosarum [16], which functions to concomitantly translocate and transfer the acetate from the cytoplasm to peptidoglycan. Recently, an enzyme has been identified within S. aureus that appears to function according to the second pathway [17], but more experimental evidence will be required to bear this out.
Herein, we describe the identification of a putative cluster of ORFs that encode both hypothetical O-acetyltransferases and O-acetyl esterases that we propose act on peptidoglycan to provide a dynamic process for attachment and removal of O-acetate, respectively. This proposal is supported by experimental data which demonstrate that representative bacteria harboring these ORFs do O-acetylate their respective peptidoglycan sacculi. This study thus represents the first to identify the presence of O-acetylpeptidoglycan in a variety of other human pathogens and the proteins involved in both the attachment and removal of this O-linked acetate.
Results and discussion
Initial identification of a peptidoglycan O-acetylation cluster in N. gonorrhoeae
N. gonorrhoeae is known to O-acetylate its peptidoglycan [5,6], however, the genes responsible for this process remain unidentified. Based on the available data, we have postulated that peptidoglycan O-acetylation may follow a similar pathway to that involved with the O-acetylation of the exopolysaccharide alginate produced by Pseudomonas aeruginosa [6]. Thus, to help identify potential genes associated with O-acetylation, the N. gonorrhoeae genome was probed with algI (U50202) [14,15], a P. aeruginosa gene encoding a putative membrane-bound, family 1 O-acetyltransferase [6]. A homologue of AlgI with 33% amino acid identity (51% similarity) was discovered in the raw sequence data of the unfinished genome project of N. gonorrhoeae FA1090 [18] even though this bacterium does not produce alginate. The ORF encodes 478 amino acids and sequence alignments revealed that the hypothetical protein possesses the signature motifs associated with the family 1 O-acetyltransferases [6], including the common F-W-R/N-R-W-H motif in the central region of the proteins (Fig. 2). Given that N. gonorrhoeae does not produce alginate, we propose that the hypothetical membrane-protein functions to O-acetylate peptidoglycan which is known to occur as a periplasmic, maturation event [5]. Thus, we have named the gene patA for peptidoglycan O-acetyltransferase A.
Analysis of sequences downstream from the patA locus led to the discovery of two more ORFs which, when subjected to BLASTP sequence similarity searches, permitted their tentative assignment of function to be associated also with the metabolism of acetate (Fig. 3). Thus, the third ORF in this cluster is predicted to be a hypothetical 44,087 Da protein of 397 amino acids having 49% sequence similarity to the catalytic domain of an acetyl xylan esterase (EC 3.1.1.72) from Ruminococcus flavefaciens (Fig. 4). This R. flavefaciens enzyme belongs to the family 3 carbohydrate esterases (CE3) of the CAZY classification [19]. Each of the other previously known members of family CE3 are also acetyl xylan esterases which aid in the degradation of xylan by removing the blocking O-linked acetate (reviewed in [20]). Although not well characterized, sequence similarity comparisons have permitted the identification of potential catalytic Asp, His, and Ser residues which could form a catalytic triad characteristic of the GDSL and TesA hydrolases [21]. As seen in Figure 4, these invariant residues are also present in the hypothetical N. gonorrhoeae protein together with associated signature motifs. Given the significant sequence similarity, it follows that the N. gonorrhoeae protein also has similar secondary structural elements compared to the family CE3 proteins, as predicted by the Garnier and PSI PRED programs (Fig. 4).
As N. gonorrhoeae is not a xylanolytic bacterium and its genome does not appear to encode xylanases, we postulate that the hypothetical family CE3 esterase functions on peptidoglycan to specifically catalyze its de-O-acetylation. This hypothetical protein was thus named Ape1 for O-acetylpeptidoglycan esterase 1. It has a high predicted pI value of 8.782 which is typical of periplasmic proteins associated with the metabolism of bacterial cell walls and thus provides further evidence in support of its proposed function. It should be noted that Ape1 (and all of the Ape proteins described below) do not have any significant sequence homology to either the Streptococcus pneumoniae peptidoglycan N-acetylglucosamine deacetylase (EC 3.1.1.-) [22] or the recently discovered N-acetylmuramic acid deacetylase (EC 3.5.1.-) from Bacillus subtilis [23], each of which belong to family CE4.
BLASTP analysis of the second ORF in this potential cluster of genes suggested that its deduced sequence of 335 amino acids is weakly similar to lysophospholipase and related esterases (E value 4e-04). Searches for known domains within its sequence using the CD-Search program indicated that it too resembled the GDSL and TesA hydrolases despite having only 18.6% identity with Ape1. Notwithstanding this low identity, the alignment revealed 53.4% similarity between the two hypothetical proteins and further analysis indicated that the second hypothetical protein does possess some of the Ape1 sequence motifs, including the GDS and Dx(V/T)H motifs containing the potential catalytic Ser, Asp, and His residues, respectively. Hence, this ORF was tentatively labeled ape2.
The patA, ape1, and ape2 gene sequences discovered on the N. gonorrhoeae chromosome were used in BLASTP searches of both the finished and unfinished genome sequences of the NCBI database. These searches led to the identification of 18 other bacteria that possess this cluster of genes (Fig. 3). This list includes some important human pathogens, both Gram-negative and Gram-positive, such as Campylobacter jejuni, Helicobacter pylori, and Bacillus anthracis, respectively.
Homology of Pat sequences
The recognition of homologs to the acetyltransferase AlgI from P. aeruginosa and their organization into a family has been described previously [6]. In the current study, we extend the list with new family members that include sequences from a variety of human pathogens such as species of Helicobacter, Campylobacter, Neisseria, and Bacillus, in particular, Bacillus anthracis (Fig. 2). All of the additional sequences from these diverse bacteria have at least 49% amino acid similarity to the P. aeruginosa AlgI despite the fact that none produce the exopolysaccharide alginate. In addition, each of these sequences contain the membrane-bound O-acetyltransferase (MBOAT) domain (pfam03062) [24]. However, none of the hypothetical proteins have significant homology to the acyltransferase 3 domain (pfam 01757) in the recently identified peptidoglycan O-acetyltransferase, OatA, from Staphylococcus aureus [17].
Architecture of the hypothetical O-acetylpeptidoglycan esterases (Ape)
The in silico identification of signature sequence motifs within the hypothetical Ape proteins was performed using a variety of programs. Predictions made using PSI PRED indicated that the sequence of secondary structural elements was very similar amongst each of the deduced proteins and that each contained the α/β hydrolase fold associated with the GDSL and TesA hydrolases (data not shown). This analogy was extended using Clustal W alignments and MEME/MAST motif discovery searches which demonstrated the strict conservation of two signature motifs that are also associated with the GDSL and TesA hydrolases (Fig. 5). The first of these, motif I, is characterized by the sequence GDS(H/F) which contains the potential catalytic serine residue, while the Dx(V/T)H of the other (motif VII) includes both the catalytic aspartyl and histidyl residues. In addition to these common sequence motifs, unique signature motifs were discerned within the intervening regions of the hypothetical proteins which permitted the subdivision of the individual sequences into a total of three separate Ape families, with family 1 being further divided into subfamilies (labeled a, b, and c) (Fig. 6). While the intervening sequences motifs distinguish the three Ape families, they nonetheless maintain the overall α/β fold of the proteins.
The Ape1 family of proteins are characterized by three additional common segments of sequence similarity which exist toward the C-terminal end of the proteins and thus closer to motif VII (Figs. 5 and 6). Three respective consensus motifs, labeled IV, V, and VI, were identified within these regions and hence, the Ape1 sequences are characterized by a total of five common sequence motifs.
The subfamily classification of the family 1 proteins is based on the presence or absence of a large sequence of approximately 100 amino acids immediately following motif I. Ape1a and Ape1b subfamilies contain this segment while it is absent in the family 1c sequences. Within this segment, only one conserved motif could be identified (motif II) and differences within it delineate the two subfamilies 1a and 1b. Although the rest of this extra segment is not highly conserved amongst the Ape1a and 1b homologs, a high amount of β-sheet structure is predicted for each. In an attempt to identify its possible function, the entire segment from each of the 12 hypothetical proteins was subjected to BLASTP searches. The only known protein that could be retrieved with any significance (E ≥ 1.0) from the database by this search was penicillin-binding protein 4 (PBP 4) from Bacillus subtilis which was obtained by probing with the segment from B. thetaiotamicron Ape1b. Interestingly, amino acid sequences similar to this N-terminal region of B. subtilis PBP4 are found in the N-terminal regions of a variety of enzymes related to peptidoglycan metabolism, including S. aureus PBP4, Streptocococcus pneumoniae PBP3, Streptomyces K15 D-transpeptidase, Escherichia coli PBP5, and TEM-1 β-lactamase. In S. aureus PBP4 [25] and Streptococcus pneumoniae PBP3 [26] (1TVF and 1XP4, respectively), the homologous sequences comprise a lobe of four surface exposed α-helices as part of the N-terminal, D,D carboxypeptidase-like domain. These helices are positioned on the back side of the active site cleft of the respective enzymes and perhaps participate in protein-protein interactions that contribute to the formation of peptidoglycan biosynthetic complexes.
The family 1a sequences are characterized by the presence of an additional region of sequence homology which is unique to these proteins. Amongst the residues that comprise the consensus motif (motif III) identified within this segment are two of the relatively rare tryptophan residues present in these proteins. This segment may thus contribute to the proteins binding specificity given the prominent role that tryptophanyl residues play in the binding subsites of carbohydrate-active enzymes through stacking interactions.
The sequences of the family 2 Ape proteins are highly homologous (Fig. 7) and, like the family 1a proteins, are characterized by having five distinct segments of high similarity separating the proposed catalytic motifs I and VII (Fig. 7). However, the consensus motifs identified are unique to this family (Fig. 7). Whereas the GDS(H/F) sequence is conserved in motif I, the conserved residues separating the catalytic D and H of consensus motif VII are distinctively G and I/V. As with the family 1 proteins, the family 2 sequences are encoded within the genomes of a number of important human pathogens, such as species of Neisseria, Campylobacter and Helicobacter.
The family 3 proteins have very little in common with either the family 1 or family 2 proteins (Fig. 6). In fact, the GDS(H/F) motif is replaced with G(T/S)S and there does not appear to be any consensus in residues separating the catalytic D and H residues of motif VII (Fig. 8). Whereas two other segments of some similarity are present between the catalytic motifs, there is a large region with minimal sequence similarity. This variation seen within the grouping suggests that further division of this family may occur once more genome sequences become available. Nonetheless, again this family does include sequences from some important pathogens such as Helicobacter pylori and Bacillus anthracis.
Predicted cellular localization of Ape
The SignalP, PSORT and DAS prediction programs indicated that almost all of the hypothetical Pat and Ape protein sequences have an N-terminal localization signal. The only exception appears to be with the Ape sequence from H. hepaticus. Family 1 proteins from H. hepaticus, P. luminescnes, Z. mobilis and P. gingivalis may have a cytoplasmic membrane association due to the retention of an α-helical transmembrane anchor. However, the other Ape1 sequences contain a predicted signal peptidase I cleavage site that should release the expressed proteins from the cytoplasmic membrane. In the case of the Ape2 homologs, the Campylobacter sp. and P. luminescens sequences are predicted to associate with the cytoplasmic membrane, while the Neisseria sp. and C. violaceum proteins would be cleaved by signal peptidase I releasing them to the periplasm. The family 3 Ape from H. pylori is also predicted to remain associated with the cytoplasmic membrane. These predictions are thus consistent with the proposed function of the expressed proteins as participating in the O-acetylation of peptidoglycan as a maturation event occurring within the periplasm [5].
Genomic organization of the OAP cluster
Despite their distant phylogenic relationship, the patA and ape genes are closely associated on respective chromosomes and comprise a small cluster we have called O-acetylpeptidoglycan (OAP) (Fig. 3). However, different gene organizations were observed which appear to be specified by the type of ape genes present. Thus, species containing the Ape1a proteins possess the strict gene order of pat immediately followed by ape2 and then ape1a. In contrast, the ape1b gene always appears to be located upstream of pat, sometimes with the addition of an intervening ape1c gene. The only ape2 not located downstream of pat is found in Zymomonas mobilis where instead it is positioned upstream together with an ape1b gene. All of the other ape1b genes are situated at the start of the OAP cluster and are followed in order by an ape1c and then pat.
Interestingly, the family 3 Ape appears to be capable of replacing the function(s) of both the family 1 and 2 proteins because its gene is situated alone immediately downstream of pat. Thus, this gene arrangement more closely resembles that of distantly related organisms like Bacillus anthracis, Bradyrhizobium japonicum and Gloeobacter violaceus than that of H. hepaticus.
Extent of peptidoglycan O-acetylation in Pat- and Ape-producing bacteria
With the exception of Agrobacterium tumefaciens, none of the bacteria listed in Fig. 3 are xylanolytic suggesting that the discovered genes may be used by the respective bacteria for the modification of their peptidoglycans as proposed for N. gonorrhoeae. However, of these listed bacteria, only N. gonorrhoeae was previously known to O-acetylate its peptidoglycan [5,6,27]. Hence, we investigated the presence of this modification in a selection of the bacteria possessing the OAP cluster of genes.
Following the isolation and purification of the peptidoglycans from cultures of the respective cells grown to late exponential phase, the quantification of base-labile, ester-linked acetate was determined by the HPLC-based method previously described [27]. Of the nine species tested, eight were found to possess O-acetylated peptidoglycan (Table 1). The levels of this O-acetylation were not stoichiometric, consistent with previous observations made with each of the other known bacteria that modify their peptidoglycan in this manner [5,6,27]. Thus, the extent of O-acetylation ranged from 14.7 % to 65.6 % relative to muramic acid content. These data thus provide preliminary support for the ascribed function of the OAP cluster of genes for the O-acetylation of peptidoglycan. With A. tumefaciens, only minimal levels of peptidoglycan O-acetylation were detected suggesting that the "Ape" proteins produced by this xylanolytic bacterium may function as authentic xylan esterases. Alternatively, environmental signals may be required by this plant pathogen to induce the O-acetylation of its peptidoglycan.
Conclusion
A new cluster of genes putatively involved in regulating O-linked acetate levels has been identified in a number of bacteria. These O-acetylpeptidoglycan clusters all consist of an O-acetyltransferase and at least one O-acetyl esterase. The O-acetyltransferases demonstrate high amino acid similarity to members of the MBOAT family of acetyltransferases [24] while the O-acetyl esterases share similarity to the CAZy family 3 esterases (CE3). The previously recognized members of family CE3 are all acetyl xylan esterases; no de-N-acetyltransferases belong to this family. Acetyl xylan esterases aid in the degradation of xylan by removing the blocking O-linked acetate and thereby permitting its subsequent hydrolysis by xylanases (reviewed in [9]). As many of the bacteria identified with these O-acetyl esterases are not xylanolytic and their genomes do not appear to encode xylanases, we postulated that these hypothetical esterases function on peptidoglycan to specifically catalyze de-O-acetylation. Consequently, they have been termed Ape proteins for their hypothetical O-acetylpeptidoglycan esterase activity.
Recognition of different domain structures for each of the Ape proteins prompted us to divide them into three distinct families. Of the families, only Ape3 appears to lack a clear conserved domain structure and this may represent a group that needs more members to permit its definitive subdivision. However, each of the bacteria possessing an Ape3 protein contain the unique cluster order of an O-acetyltransferase followed by a single O-acetyl esterase. The other Ape proteins are associated with an O-acetyltransferase, but always seem to be found in pairs consisting of Ape1b and Ape1c or Ape1a/1b and an Ape2. This ORF organization may translate into the evolutionary development of the Ape families. In fact, evolutionary divergence seems plausible for the Ape1 and Ape2 proteins, where duplication of an Ape1a/1b protein and divergence into an entirely new protein could have easily occurred. However, the ORF organization in Helicobacter species makes this theory far more confounded. Alternatively, differences between the Ape families may be linked to separate cellular functions. Some of the Ape proteins, especially those associated with either the cytoplasmic membrane and cytoplasm, may be involved in peptidoglycan recycling, while those in the periplasm and/or outer membrane of the gram-negative bacteria may be involved in the general maintenance of the peptidoglycan sacculus.
Regardless of their specific role in peptidoglycan metabolism, the activity of Ape proteins would be essential for the removal of acetate which would otherwise preclude the lysis of peptidoglycan by endogenous enzymes. Indeed, the major lytic enzymes associated with both the biosynthesis and maintenance of peptidoglycan are the lytic transglycosylases, enzymes that require a free C-6 hydroxyl group on muramoyl residues to produce their 1,6-anhydro product. Given that the bacteria harboring genes apparently involved with this activity were demonstrated to produce O-acetylated peptidoglycan, Ape activity would be required to function in advance of the lytic transglycosylase, many of which act processively along peptidoglycan strands [1], to provide appropriate sites for cleavage. Although strictly conjecture at this point, it is likely that these Ape proteins would be associated with the peptidoglycan 'replicase' holoenzyme complexes [1] which have been shown to include both the lytic transglycosylases and penicillin-binding proteins [28-30]. In this regard, it should be noted that many of the ape genes are in close proximity to other genes encoding enzymes involved in cell wall associated functions. For example, the three Campylobacter sp. contain a ferritin gene (cft and pfr) immediately upstream, while both C. coli and C. upsaliens also possess genes encoding a potential ABC transporter and an outer membrane protein directly downstream of the OAP cluster. Similarly, both Bacteroides sp. encode a putative cation-efflux transporter upstream of their OAP clusters, while the Z. mobilis chromosome encodes a nearby cell wall hydrolase. Hence, through a dynamic process of removal and attachment of O-acetate, the bacteria would be able to obtain a high level of localized control over the degradation of peptidoglycan. This would facilitate, for example, the efficient insertion of pores and flagella, localize spore formation, and control the level of general peptidoglycan turnover.
It is difficult to determine whether or not a correlation exists between O-acetylation levels and the type or number of O-acetyl esterases present in a particular organism. In the case of the Ape1b/1c and Ape3 this may be because so few organisms have been tested for peptidoglycan O-acetylation. However, even Neisseria sp., C. violaceum, and the Campylobacter sp., which all possess Ape1a and Ape2, contain O-acetylation levels that vary significantly. This suggests that the level of O-acetylpeptidoglycan esterase activity may be regulated by compartmentalization or at the level of transcription/translation. The Ape family 1a/1b proteins are different from the Ape1c proteins due to the presence of an extra amino acid segment. The variation in each of these segments raises questions of substrate specificity and function that must be proven, but it is interesting to note that one of the segments has partial similarity to PBP 4, a D,D carboxypeptidase. These proteins are peptidoglycan maintenance proteins, and therefore suggest that the extra protein segment found in the Ape1a/b proteins may be a peptidoglycan specific fragment or one directing protein-protein interactions. This would certainly explain its absence from the O-acetylxylan esterase CE3 family, but not necessarily from the other Ape families.
Two models of peptidoglycan O-acetylation have been proposed in a recent review [6]. The first model involves the sequential action of two proteins. The first protein would be located within the cytoplasmic membrane and facilitates the transfer of acetate from acetyl-CoA to the periplasm, where a second protein accepts the acetate and attaches it to peptidoglycan. The second model consists of a single membrane bound O-acetyltransferase that performs both the transport of acetate across the cytoplasmic membrane and the peptidoglycan O-acetylation steps. A peptidoglycan O-acetyltransferase recently identified in S. aureus [17] is expected to act by the second peptidoglycan O-acetylation pathway. This O-acetyltransferase bears no resemblance to the O-acetyltransferases identified in the present study. This strongly suggests that, as with the O-acetyl esterases together with the lytic transglycosylases [31] and the PBPs [32], different classes of peptidoglycan O-acetyltransferases exist.
In summary, experiments are still required to conclusively demonstrate that the O-acetyltransferases and Ape proteins identified in this study are actually involved in controlling peptidoglycan O-acetylation. However, it is important to note that all the organisms that demonstrated some level of peptidoglycan O-acetylation also contain the putative O-acetylation cluster. Verification of peptidoglycan O-acetylation in the other bacteria containing the O-acetylation cluster are in progress, and should further strengthen the link between these two events. In addition, experiments characterizing the gene product of N. gonorrhoeae Ape1a have been initiated and preliminary evidence supports its function as an O-acetylpeptidoglycan esterase.
Methods
Chemicals and reagents
DNase I, RNase A, and pronase were purchased from Roche Molecular Biochemicals (Laval, PQ), while Fisher Scientific (Nepean, ON), provided Luria Bertani (LB) growth medium. All other growth media were purchased from Difco Laboratories (Detroit, MI), and all other chemicals and reagents were from Sigma Chemical Co. (St. Louis, MO).
Bacterial strains and growth media
Agrobacterium tumefaciens and Photorhabdus luminescens were both cultured in Tryptic Soy Broth (TSB) for 48 h at ambient temperature with shaking (200 rpm) while Bacillus cereus was grown in Penassay Broth as previously described [33]. Chromobacterium violaceum was cultured in LB broth with shaking (200 rpm) for 48 h at ambient temperature. Bacteroides fragilis and B. thetaiotaomicron were grown at 37°C under strict anaerobic conditions using Bacteroides Bile Esculin Broth (BBE) supplemented with 100 mg/mL gentamicin while Ruminococcus flavefaciens was cultured anaerobically using Dehority's Complete Artificial Medium as previously described [34]. Helicobacter pylori and Campylobacter jejuni were cultured at 37°C under microaerophilic conditions (10% (vol/vol) O2, 5% (vol/vol) CO2, and 85% (vol/vol) N2) in Brain Heart Infusion Broth supplemented with 10% irradiated horse serum and TSB supplemented with 0.6% (w/vol) yeast extract, respectively. Bradyrhizobium japonicum was grown in Yeast Extract Mannitol broth at ambient temperature as previously described [35].
Peptidoglycan isolation and determination of extent of O-acetylation
O-Acetylated peptidoglycan was isolated and purified from the various bacteria, taking the necessary precautions to prevent the facile hydrolysis of O-acetyl groups as previously described [36]. Quantification of O-acetyl content was performed by both the HPLC-based organic acid analysis using an Aminex HPX-87H Bio-Rad column as previously described [27] and using the Megazyme Acetic Acid Assay kit (Megazyme International Ireland Ltd., Wicklow, Ireland). The extent of O-acetylation is presented as a percentage of muramic acid content as determined by quantitative aminosugar analysis [37].
Identification and genomic analysis of a putative cluster of peptidoglycan O-acetylation ORFs
The raw nucleotide sequence data of the N. gonorrhoeae FA1090 genome project [18] was searched for hypothetical proteins involved in acetate transport using the Pseudomonas aeruginosa protein sequence for AlgI as the probe (algI encodes a putative acetate translocator involved with the biosynthesis of alginate [14,15]). Identification of open reading frames, protein translations, and isoelectric points (pI) were performed using Clone Manager 5. Signal sequence and cellular localization predictions were performed using signalP version 3 [38], PSORT [39], and the DAS Transmembrane Prediction Server [40]. Searches for homologs to the characterized N. gonorrhoeae ORFs in the NCBI database were carried out using the BLASTP search engine to tentatively identify functions for each of the ORFs in the cluster[41]. Sequences with an (E) value threshold of 1e-06 were used to generate multiple alignments using ClustalW Version 1.8 software [42]. Manual correction of these alignments was performed based on the original BLASTP results in conjunction with the MEME/ MAST Version 3.0 motif discovery tool [43]. Predictions of secondary structure were performed using the GOR IV Secondary Structure Prediction method [44] and PSI PRED [45], while searches for known domains were performed using CD-Search [46] at NCBI.
Authors' contributions
JTW and AJC conceived the study, participated in the sequence alignments and analyses, and wrote the manuscript. JMP performed the quantification of peptidoglycan O-acetylation, and all authors read and approved the final manuscript.
Acknowledgements
This work was supported by an operating grant to AJC from the Canadian Institutes of Health Research (MOP 62772) and a post-graduate fellowship to JTW from the Natural Sciences and Engineering Research Council of Canada.
Figures and Tables
Figure 1 (A) Structure of O-acetylpeptidoglycan and (B) reaction pathway of lytic transglycosylases (LT). A. The O-acetylation of peptidoglycan occurs at the C-6 hydroxyl group of MurNAc residues (shown in red) to generate the corresponding 2-N-6-O-diacetylmuramyl residue (OAcMurNAc). B. The lytic transglycosylases (LT) require a free C-6 hydroxyl group to catalyze the formation of 1,6-anhdyroMurNAc products. The R denotes the stem peptide associated with the C-3 lactyl moiety of MurNAc residues.
Figure 2 Sequence alignment of N. gonorrhoeae PatA and related hypothetical proteins found in the genome databases with family 1 O-acetyltransferases. The family of proteins presented in the upper block of sequences includes AlgI from both P. aeruginosa PAO1 and Azotobacter vinelandii, a transporter of acetate involved in the O-acetylation of the exopolysaccharide alginate [6]. The residues in bold face and yellow highlight denote greater than 50% and 80% identity, respectively, within the separate blocks of sequences. The asterisks identify conservation between the two blocks of sequences while the residues in red are invariant amongst all sequences. Abbreviations (accession numbers): Pa, P. aeruginosa (U50202); Av, Azotobacter vinelandii, (AF027499); Hp, Helicobacter pylori (AE000596); Tp, Treponema pallidum (AE001232); Lc, Lactobacillus casei, (P35855); Sa, Staphylococcus aureus (D86240); Sm, Streptococcus mutans (AF049357); Bs, Bacillus subtilis (P39580), Ng, N. gonorrhoeae (AAW89272); Nm, N. meningitidis (NP284200); Cv, Chromobacterium violaceum (NP903736); Pl, Photorhabdus luminescens (NP927855); Hh, Helicobacter hepaticus (NP860615); Cj, Campylobacter jejuni, (NP281794); Cc, Campylobacter coli (ZP00366780); Zm, Zymomonas mobilis (YP162181); Bj, Bradyrhizobium japonicum (NP767015); Gv, Gloeobacter violaceus (NP923893); Ba, Bacillus anthracis (NP843401); Bc, B. cereus (NP977299); Pg, Porphyromonas gingivalis (NP905383); Bt, Bacteroides thetaiotaomicron (NP811593); Bf, Bacteroides. fragilis (YP101412).
Figure 3 Gene organizations of O-acetylpeptidoglycan (OAP) cluster composed of pat and ape genes. Abbreviations for other gene products: cft, ferritin; pfr, ferritin; flgG, flagellar distal rod protein; GuaC, GMP reductase.
Figure 4 Sequence alignment and secondary structure predictions of N. gonorrhoeae Ape1 and CAZy CE3 carbohydrate esterases (O-acetylxylan esterases). The residues in bold face denote at least 50% identity amongst the CE-3 esterases, while those highlighted in yellow denote identity with N. gonorrhoeae Ape1. Invariant residues are identified by the asterisks and the putative catalytic residues are in red. The numbers in brackets denote the length of intervening sequences. Abbreviations (accession numbers): Pa, P. aeruginosa (YP207682); Rf, Ruminococcus flavefaciens (XynB, Q52753; CesA, Q9RLB8); Nf, Neocallimastix frontalis (U66253); Np, N. patriciarum (O13497); An, Aspergillus nidulans, (EAA66937); Nsp, Nomomuraea species (Q7WZ50); Gv, Gloeobacter violaceus (Q7NGX3); Nc, Neurospora crassa (XM323878).
Figure 5 Sequence alignment of family 1 O-acetylpeptidoglycan esterases (Ape) and identification of consensus motifs. The residues in bold face and yellow highlight denote greater than 50% identity amongst the subfamilies (1A, 1B, 1C) and the entire family, respectively, while those in red are invariant amongst all sequences. The numbers in brackets denote the length of intervening sequences and the potential catalytic residues are identified with the red asterisks. Abbreviations (accession numbers): Ng, N. gonorrhoeae (AAW89270); Nm, N. meningitidis (NP284202); Cv, Chromobacterium violaceum (NP903734); Pl, Photorhabdus luminescens (NP927853); Cj, Campylobacter jejuni, (NP281792); Cc, Campylobacter coli (ZP00366778); Cu, Campylobacter upsaliensis (ZP00369960); Hh, Helicobacter hepaticus (NP860613); Bt, Bacteroides thetaiotaomicron (1b, NP811595; 1c, NP811594); Bf, Bacteroides. fragilis (1b, YP101414; 1c, NP811594); Pg, Porphyromonas gingivalis (1b, NP905385; 1c, NP905384); Zm, Zymomonas mobilis (1b, YP162180).
Figure 6 Distribution of consensus sequence motifs within the three families of Ape proteins. The motifs of the family 1 Apes are colour coded, and the asterisks denote the locations of the catalytic serine and aspartyl/histidyl residues of motifs I and VII, respectively. The numbers at the beginning of each sequence denote the length of the respective N-terminal sequences.
Figure 7 Sequence alignment of family 2 O-acetylpeptidoglycan esterases (Ape2) and identification of consensus motifs. The residues in bold face and yellow highlight denote greater than 50% and 80% identity, respectively, while those in red are invariant amongst all sequences. The numbers in brackets denote the length of intervening sequences and the potential catalytic residues are identified with the red asterisks. Abbreviations (accession numbers): Ng, N. gonorrhoeae (AAW89271); Nm, N. meningitidis (NP284201); Cv, Chromobacterium violaceum (NP903735); Pl, Photorhabdus luminescens (NP927854); Cj, Campylobacter jejuni, (NP281793); Cc, Campylobacter coli (ZP00366779); Cu, Campylobacter upsaliensis (ZP00369961); Hh, Helicobacter hepaticus (NP860614); Zm, Zymomonas mobilis (1b, YP162179).
Figure 8 Sequence alignment of family 3 O-acetylpeptidoglycan esterases (Ape3) and identification of consensus motifs. The residues in bold face and yellow highlight denote greater than 50% and 80% identity, respectively, while those in red are invariant amongst all sequences. The numbers in brackets denote the length of intervening sequences and the potential catalytic residues are identified with the red asterisks. Abbreviations (accession numbers): Hp, Helicobacter pylori (NP207650); Bj, Bradyrhizobium japonicum (NP767014); Gv, Gloeobacter violaceus (NP923892); Ba, Bacillus anthracis (NP843402); Bc, B. cereus (NP977300).
Table 1 Extent of peptidoglycan O-acetylation in bacteria possessing OAP gene clusters
Species Strain % O-Acetylation1
Agrobacterium tumefaciens A6 3.44 ± 0.79
371 1.52 ± 0.36
Bacillus cereus ATCC 10702 24.7 ± 0.09
Bacteroides fragilis ATCC 25285 69.4 ± 1.2
Bacteroides thetaiotamicron ATCC 29741 27.0 ± 1.2
Bradyrhizobium japonicum 532 C 18.5 ± 2.2
Campylobacter jejuni ATCC 700819 62.7 ± 5.6
NCTC 11168 55.7 ± 1.4
Chromobacterium violaceum ATCC 12472 14.7 ± 2.5
Helicobacter pylori ATCC 700392 45.6 ± 3.2
Photorhabdus luminescens ATCC 29999 65.6 ± 1.6
Ruminococcus flavefaciens 007 33.4 ± 3.4
1. Average ± s.d.(n = 3)of base-labile acetate content relative to muramic acid concentrations in isolated and purified peptidoglycan.
==== Refs
Höltje J-V Growth of the stress-bearing and shape-maintaining murein sacculus of Escherichia coli Microbiol Mol Biol Rev 1998 62 181 203 9529891
Koraimann G Lytic transglycosylases in macromolecular transport systems of Gram-negative bacteria Cell Mol Life Sci 2003 60 2371 2388 14625683 10.1007/s00018-003-3056-1
Shockman GD Höltje J-V Ghuysen J-M, Hakenbeck R Microbial peptidoglycan (murein) hydrolases New Comprehensive Biochemistry, Bacterial Cell Wall 1994 27 Amsterdam: Elsevier BV 131 166
Höltje J-V Mirelmen D Sharon N Schwarz U Novel type of murein transglycosylase in Escherichia coli J Bacteriol 1975 124 1067 1076 357
Clarke AJ Dupont C O-Acetylated peptidoglycan: Its occurrence, pathobiological significance and biosynthesis Can J Microbiol 1992 38 85 91 1521192
Clarke AJ Strating H Blackburn NT Doyle RJ Pathways for the O-acetylation of bacterial cell wall polymers Glycomicrobiology 2000 New York: Plenum Publishing Co. Ltd 187 212
Dupont C Clarke AJ Dependence of lysozyme-catalyzed solubilization of Proteus mirabilis peptidoglycan on the extent of O-acetylation Eur J Biochem 1991 195 763 769 1999194 10.1111/j.1432-1033.1991.tb15764.x
Swim SG Gfell MA Wilde CE IIIRosenthal RS Strain distribution in extents of lysozyme resistance and O-acetylation of gonocccal peptidoglycan determined by high-performance liquid chromatography Infect Immun 1983 42 446 452 6417014
Rosenthal RS Folkening WJ Miller DR Swim SC Resistance of O-acetylated gonococcal peptidoglycan to human peptidoglycan-degrading enzymes Infect Immun 1983 40 903 911 6406367
Rosenthal RS Blundell JK Perkins HR Strain-related differences in lysozyme sensitivity and extent of O-acetylation of gonococcal peptidoglycan Infect Immun 1982 37 826 829 6811442
Blundell JK Smith GJ Perkins HR The peptidoglycan of Neisseria gonorrhoeae: O-acetyl groups and lysozyme sensitivity FEMS Microbiol Lett 1980 9 259 261 10.1016/0378-1097(80)90028-2
Blackburn NT Clarke AJ Characterization of soluble and membrane-bound family 3 lytic transglycosylases from Pseudomonas aeruginosa Biochemistry 2002 41 1001 1013 11790124 10.1021/bi011833k
Payie KG Strating H Clarke AJ The role of O-acetylation in the metabolism of peptidoglycan in Providencia stuartii Microb Drug Resist 1996 2 135 140 9158736
Nivens DE Ohman DE Williams J Franklin MJ Role of alginate and its O acetylation in formation of Pseudomonas aeruginosa microcolonies and biofilms J Bacteriol 2001 183 1047 105 11208804 10.1128/JB.183.3.1047-1057.2001
Franklin MJ Ohman DE Identification of algI and algJ in the Pseudomonas aeruginosa alginate biosynthetic gene cluster which are required for alginate O-acetylation J Bacteriol 1996 178 2186 2195 8636017
Firmin J Wilson KE Carlson RW Davies AE Downie JA Resistance to nodulation of cv. Afghanistan peas is overcome by nodX, which mediates an O-acetylation of the Rhizobium leguminosarum lipo-oligosaccharide nodulation factor Mol Microbiol 1993 10 351 360 7934826
Bera A Herbert S Jakob A Vollmer W Gotz F Why are pathogenic staphylococci so lysozyme resistant? The peptidoglycan O-acetyltransferase OatA is the major determinant for lysozyme resistance of Staphylococcus aureus Mol Microbiol 2005 55 778 787 15661003 10.1111/j.1365-2958.2004.04446.x
Neisseria gonorrhoeae Genome Sequencing Strain FA 1090
CAZy – Carbohydrate-Active enZYmes
Clarke AJ Biodegradation of Cellulose: Enzymology and Biotechnology 1997 Boca Raton: CRC Press
Upton C Buckley JT A new family of lipolytic enzymes? Trends Biochem Sci 1995 20 178 179 7610479 10.1016/S0968-0004(00)89002-7
Vollmer W Tomasz A The pgdA gene encodes for a peptidoglycan N-acetylglucosamine deacetylase in Streptococcus pneumoniae J Biol Chem 2000 275 20496 20501 10781617 10.1074/jbc.M910189199
Fukushima T Kitajima T Sekiguchi J A polysaccharide deacetylase homologue, PdaA, in Bacillus subtilis acts as an N-acetylmuramic acid deacetylase in vitro J Bacteriol 2005 187 1287 1292 15687192 10.1128/JB.187.4.1287-1292.2005
Hofmann K A superfamily of membranebound O-acyltransferases with implications for Wnt signalling Trends Biochem Sci 2000 25 111 112 10694878 10.1016/S0968-0004(99)01539-X
Rajashankar KR Ray SS Bonanno JB Burley SK Crystal structure of penicillin-binding protein 4 (Pbp4) from Staphylococcus aureus PDB: 1TVF
Morlot C Pernot L Le Gouellec A Di Guilmi AM Vernet T Dideberg O Dessen A Crystal structure of a peptidoglycan synthesis regulatory factor (PBP3) from Streptococcus pneumoniae J Biol Chem 2005 280 15984 15991 15596446 10.1074/jbc.M408446200
Clarke AJ Extent of peptidoglycan O-acetylation in the tribe Proteeae J Bacteriol 1993 175 4550 4553 8331084
Alaedini A Day RA Identification of two penicillin-binding multienzyme complexes in Haemophilus influenzae Biochem Biophys Res Commun 1999 264 191 195 10527863 10.1006/bbrc.1999.1509
Vollmer W von Rechenberg M Höltje J-V Demonstration of molecular interactions between the murein polymerase PBP1B, the lytic transglycosylase MltA, and the scaffolding protein MipA of Escherichia coli J Biol Chem 1999 274 6726 6734 10037771 10.1074/jbc.274.10.6726
von Rechenberg M Ursinus A Höltje J-V Affinity chromatography as a means to study multienzyme complexes involved in murein synthesis Microb Drug Resist 1996 2 155 157 9158739
Blackburn NT Clarke AJ Identification of four families of peptidoglycan lytic tranglycosylases J Mol Evol 2002 52 78 84 11139297
Goffin C Ghuysen JM Multimodular penicillin-binding proteins: an enigmatic family of orthologs and paralogs Microbiol Mol Biol Rev 1998 62 1079 1093 9841666
Zipperle GF Ezzell JW Doyle RJ Glucosamine substitution and muramidase susceptibility in Bacillus anthracis Can J Microbiol 1984 30 553 559 6430537
Scott HW Dehority BA Vitamin requirements of several cellulolytic rumen bacteria J Bacteriol 1965 89 1169 1175 14292981
Vincent JM A Manual for the Practical Study of Root Nodulating Bacteria 1970 Oxford: Blackwell Scientific
Strating H Clarke AJ Differentiation of bacterial autolysins by zymogram analysis Anal Biochem 2001 291 149 154 11262168 10.1006/abio.2001.5007
Clarke AJ Compositional analysis of peptidoglycan by high-performance anion-exchange chromatography Anal Biochem 1993 212 344 350 8214575 10.1006/abio.1993.1339
SignalP 3.0 Server
PSORT Prediction
"DAS" – Transmembrane Prediction server
NCBI BLAST
ClustalW
THE MEME/MAST SYSTEM Motif Discovery and Search Version 3.0.14
GOR IV Secondary Structure Prediction Method
The PSIPRED Protein Structure Prediction Server
NCBI Conserved Domain Search
|
16111493
|
PMC1199599
|
CC BY
|
2021-01-04 16:03:39
|
no
|
BMC Microbiol. 2005 Aug 19; 5:49
|
utf-8
|
BMC Microbiol
| 2,005 |
10.1186/1471-2180-5-49
|
oa_comm
|
==== Front
BMC NeurosciBMC Neuroscience1471-2202BioMed Central London 1471-2202-6-521612238710.1186/1471-2202-6-52Research ArticleEvidence for proteolytic cleavage of brevican by the ADAMTSs in the dentate gyrus after excitotoxic lesion of the mouse entorhinal cortex Mayer Joanne [email protected] Michelle G [email protected] Paul E [email protected] University of South Florida College of Medicine, Department of Pharmacology and Therapeutics, 12901 Bruce B. Downs Blvd, Tampa, Florida 33612-4799, USA2005 25 8 2005 6 52 52 9 5 2005 25 8 2005 Copyright © 2005 Mayer et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Brevican is a member of the lectican family of aggregating extracellular matrix (ECM) proteoglycans that bear chondroitin sulfate (CS) chains. It is highly expressed in the central nervous system (CNS) and is thought to stabilize synapses and inhibit neural plasticity and as such, neuritic or synaptic remodeling would be less likely to occur in regions with intact and abundant, lectican-containing, ECM complexes. Neural plasticity may occur more readily when these ECM complexes are broken down by endogenous proteases, the ADAMTSs (adisintegrin and metalloproteinase with thrombospondin motifs), that selectively cleave the lecticans. The purpose of these experiments was to determine whether the production of brevican or the ADAMTS-cleaved fragments of brevican were altered after deafferentation and reinnervation of the dentate gyrus via entorhinal cortex lesion (ECL).
Results
In the C57Bl6J mouse, synaptic density in the molecular layer of the dentate gyrus, as measured by synaptophysin levels in ELISA, was significantly attenuated 2 days (nearly 50% of contralateral) and 7 days after lesion and returned to levels not different from the contralateral region at 30 days. Immunoreactive brevican in immunoblot was elevated 2 days after lesion, whereas there was a significant increase in the proteolytic product at 7, but not 30 days post-lesion. ADAMTS activity, estimated using the ratio of the specific ADAMTS-derived brevican fragment and intact brevican levels was increased at 7 days, but was not different from the contralateral side at 2 or 30 days after deafferentation.
Conclusion
These findings indicate that ADAMTS activity in the dentate outer molecular layer (OML) is elevated during the initial synaptic reinnervation period (7 days after lesion). Therefore, proteolytic processing of brevican appears to be a significant extracellular event in the remodeling of the dentate after EC lesion, and may modulate the process of sprouting and/or synaptogenesis.
==== Body
Background
Neurons of the entorhinal cortex (EC) send unidirectional, afferent projections to the hippocampus, where terminals synapse on granule cell dendrites in the outer molecular layer (OML) of the dentate gyrus [1]. Interruption of entorhinal input to the dentate gyrus, by chemical lesion or severing the afferent fibers, causes anterograde degeneration of the axon terminals and stimulates sprouting of viable fibers from this and other neuronal circuits [2-5]. The entorhinal cortex lesion (ECL) has been used as a model of neural plasticity for more than three decades [6]. To identify and innervate a target after injury or during development, growing neuron terminals must traverse through a complex extracellular milieu that consists of soluble factors, cell surface adhesive ligands and an extracellular matrix (ECM) along the way toward its target. The growing terminal samples this milieu, and appropriate protein-protein binding and activation regulates the direction and extent of growth, terminal sprouting, and likely synaptogenesis. For example, certain ECM molecules such as laminin are permissive toward neurite outgrowth, whereas others, such as the highly negatively charged, proteoglycans (PGs) substituted with chondroitin sulfate (CS) (ie. versican, neurocan, aggrecan and brevican, in general, lecticans), inhibit neurite outgrowth on permissive substrates [7-9]. Interestingly, the expression of several lecticans, including brevican is markedly up-regulated during the neural plasticity response that occurs following ECL [10].
The lectican PGs are the most abundant ECM molecules in the adult, uninjured central nervous system (CNS) and of these brevican is the most highly expressed [11,12]. Brevican and other lecticans are found in perineuronal nets and throughout the neuropil and are components of ECM aggregate complexes that are thought to stabilize synapses in neural networks and inhibit neurite outgrowth [13,14]. Most evidence suggests that the CS component in these complexes provides the inhibitory signal toward neurite outgrowth, although the protein core plays a role as well [8,15]. However, proteolytic cleavage of the brevican core protein may "loosen" the aggregated complexes and change the extracellular environment to one that is more permissive for neural plasticity to occur [14,16,17]. A significant proportion of brevican in the adult CNS exists as a fragment formed by proteolytic cleavage of the protein core, suggesting that this mechanism may play a role in experience-dependent and other forms of neural plasticity in the uninjured, adult. In addition, in disorders such as Alzheimer's disease, alterations in the ECM may be related to diminished synaptic plasticity and therefore play a role in cognitive dysfunction.
The metalloproteinases responsible for cleavage of lectican core proteins and the generation of fragments of aggrecan, brevican, and versican have been cloned and belong to a family of proteins, termed the ADAMTSs (adisintegrin and metalloproteinase with thrombospondin motifs), that include glutamyl endopeptidases. ADAMTS1 and ADAMTS4 are prominently expressed in rat brain [16] and cleave the 145 kD intact core protein of brevican into a 55 kD N-terminal, and an 80 kD C-terminal fragment [18,19] (see Fig. 2). The ADAMTS-derived 55 kD fragment may be distinguished from total cleaved fragment (cleaved by other proteinases) using a neoepitope antibody raised against the distinctive C-terminal sequence (EAMESE, murine sequence) of the N-terminal, 55 kD fragment (see Fig. 2, "C") that is generated by glutamyl endopeptidase cleavage. Using this antibody, in vivo ADAMTS activity may be estimated by expressing the amount of ADAMTS-derived brevican fragment as a proportion of total intact brevican.
Others have demonstrated changes in the levels and activity of matrix metalloproteinases (MMPs) in the deafferented neuropil after various lesions, however in these experiments, the substrate proteolytically processed by the MMPs was not identified [20,21]. Nonetheless, changes in structure and function that were associated with lesion-induced sprouting were reversed by MMP inhibitors [22]. Lesions produced by systemic injection of kainic acid produce wide-spread neuronal degeneration in the CNS, including the EC and stimulate expression of the MMPs and the ADAMTSs [16,23]. In response to this lesion, the abundance of the neoepitope fragment generated by ADAMTS-cleavage of brevican was markedly elevated in the OML of the dentate gyrus. This increase was preceded by elevated ADAMTS1 and ADAMTS4 mRNA expression in dentate granule neurons [16]. Although these results suggest that proteolytic processing of brevican appears to be a significant extracellular event in the remodeling of the dentate after ECL, because of the widespread neuronal death, it was difficult to associate these two particular endpoints as individual, yet associated events involved in neural plasticity. Thus, we decided to employ the classical ECL model in the C57Bl6 mouse, thereby discretely disrupting synaptic innervation of the molecular layer of the dentate gyrus. Our intention was to observe any altered expression of brevican and the ADAMTSs that may be associated with reinnervation of the injured area 2, 7 and 30 days after lesion.
Results
Expression of ADAMTS-derived, brevican fragment
Brevican is an abundantly expressed PG in the CNS that is secreted from astrocytes and neurons as a 145 kD core protein that bears up to three, covalently-linked, CS chains (Fig. 2A). It is also is secreted as a 145 kD core protein without CS chains (Fig. 2B). When cleaved by extracellular glutamyl endopeptidases, the ADAMTSs, a 55 kD N-terminal fragment is formed that contains the unique C-terminal epitope EAMESE (Fig. 2C). Each of these isoforms of brevican was expressed in numerous regions of the CNS. Figure 3A shows a brevican immunoblot of several brain regions from an untreated C57Bl6 mouse. The >145 kD (brevican core plus CS) intact isoform, the 145 kD (brevican core) intact core protein and the generalized 55 kD N-terminal fragment were easily differentiated from one another on Western blot (Fig. 3A) using an antibody that recognizes an epitope in the N-terminal globular domain (Fig. 2). In the mouse, all isoforms of brevican appeared to be more abundant in brain stem and cerebellum compared to other regions that were measured. Panel B shows a blot with the same extracts and probed using the anti-neoepitope EAMESE antibody. This antibody recognized the ADAMTS-derived, C-terminally truncated, 55 kD fragment with the novel C-terminal epitope EAMESE and did not recognize any of the intact, core brevican proteins (Fig. 3B). Thus, given the ability to quantitate the ADAMTS-derived brevican fragment and its intact substrates using immunoblot, regional in vivo ADAMTS activity may be estimated.
Deafferentation and neural plasticity in the dentate gyrus
To validate a complete deafferentation of this synapse of the perforant path, from the EC to the molecular layer of the dentate gyrus, animals were injected with ibotenic acid into the lateral EC, perfused with fixative, the brain sectioned, and immunostained for brevican. Brevican is a marker expressed by reactive glia in response to brain injury [24]. Thus, after lesion, increased deposition of brevican would be expected in the ipsilateral EC. Brevican immunoreactivity was elevated at the lesion site (Figure 4B) compared to the contralateral, non-lesioned EC (Figure 4A).
Since an endpoint in this experiment is the reinnervation and sprouting of neurites and the formation of new synapses in the OML of the dentate gyrus, synaptophysin immunoreactivity, a vesicular pre-synaptic marker, was examined two, seven, and thirty days post-lesion. On the contralateral side and in unlesioned mice, a tri-laminar synaptophysin staining pattern was observed that outlines the inner, middle and OMLs (Fig. 5B) of the dentate gyrus. Two days after ECL, there was a marked decline in synaptophysin immunoreactivity in the OML on the ipsilateral side of the dentate (Fig. 5A), that was reflected by a loss of the typical tri-laminar pattern of synaptophysin staining (Fig. 5B) with only the inner molecular layer clearly distinguishable. Seven days after ECL, synaptophysin immunoreactivity in the OML remained markedly reduced, although the pattern was similar to the contralateral side at 30 days (data not shown).
In response to injury and the deafferentation of synapses, there is an activation of astrocytes in the OML [25]. Tissue immunostained with the astrocytic marker GFAP (glial fibrillary astrocytic protein), showed increased staining selectively in the OML of the dentate seven days after ECL (Fig. 5C) compared to the contralateral side (Fig. 5D). Although these markers may be sufficient to identify some level of diminished innervation to the OML, a synaptophysin ELISA and a specific method to isolate the tissue of the dentate gyrus (Fig. 1) was developed to better quantitate this loss.
Two days after lesion, synaptophysin levels as detected by ELISA, declined by 46% in the ipsilateral dentate compared to the contralateral side (p = 0.039) (Fig. 5E). This trend was sustained at seven days post-lesion, and synaptophysin levels were 41% of the contralateral value (p = 0.033) (Fig. 5E). However, when tissue was examined thirty days post-lesion, synaptophysin concentrations were not different from the control side. This observation shows there is a complete reinnervation of the OML after ECL (Fig. 5E).
The decline in the level of synaptophysin concentrations in the tissue from the "dentate punch" on the ipsilateral side compared to the contralateral side indicates that there was denervation of cortical input to the dentate molecular layer. The question of interest was, however, were there associated changes in the abundance and proteolysis of brevican in this region of neural plasticity?
Abundance and proteolysis of brevican after ECL
Brevican and its various isoforms were identified in immunoblots from isolated dentate gyrus tissue obtained from mice two, seven and thirty days after ECL. At two days post-lesion (Figure 6A), there was a 3-fold increase in the glycosaminoglycan (GAG)-containing form of brevican (p = 0.047) in the dentate on the denervated side, and a trend for an increase in all the isoforms of brevican on this side. At seven days after insult, a time thought to be the initial synaptic reorganization period after ECL [26], intact brevican had returned to contralateral levels, but there was a significant, almost 2-fold elevation in the ADAMTS-derived fragment of brevican in the dentate (p = 0.030). In addition, there was a trend for an increase in the generalized 55 kD fragment of brevican (Fig. 6B) at this time point. Thirty days after undergoing surgery, all of the forms of brevican had returned to levels not different from the contralateral side; however, a slight increase in the GAG-containing, intact isoform was seen at this time point (Fig. 6C).
To estimate apparent ADAMTS activity in vivo, the density of the EAMESE (55 kD) immunoreactive fragment in Western blot was divided by the densitometric level of intact full-length and core protein brevican isoforms (>145 kD + 145 kD) (Fig. 7). Two days post-ECL there was a decrease in apparent ADAMTS activity in the ipsilateral dentate gyrus, however, at the seven day critical reinnervation period, ADAMTS activity was increased almost 50%. This observation is supported by the significant increase in the ADAMTS-derived, 55 kD fragment at this time point. By thirty days post-lesion, the activity was not different from the contralateral side. Thus, this data suggests that the ADAMTSs and the catabolism of brevican may play a prominent role in neural plasticity in the dentate gyrus after lesion of the EC in mice.
Discussion
The purpose of this study was to determine whether the catabolism of brevican is involved in mechanisms of neural plasticity in the hippocampus, and to accomplish this, synaptic input to the OML of the dentate gyrus was denervated by excitotoxic lesion in the lateral EC. Two days after lesion, synaptic input into the OML was significantly reduced and this was accompanied by an increase in the production of full length, intact brevican. At seven days, while brevican levels returned to baseline, a significant increase in the ADAMTS-derived, C-terminally truncated, brevican fragment was observed during this initial, sprouting and reinnervation period. This implies that there was an increase in ADAMTS activity in the OML during the highly plastic, regenerative phase. However at thirty days post-lesion, there was complete reinnervation of the OML on the ipsilateral side, as synaptic density, brevican and ADAMTS activity were not different from the contralateral side at this time point. These results indicate that the ADAMTSs and their substrate, brevican, that is abundant in the CNS, have a regulatory function in neural plasticity and support earlier data that had demonstrated important actions for the ADAMTSs in plasticity after seizure-induced hippocampal lesion [16].
Previous studies have examined the role of matrix-altering proteases in synaptic plasticity after CNS lesion. The expression of the matrix metalloproteinases (MMPs), MMP-9 and MMP-2, have been shown to be increased in various regions of the hippocampus after seizure-induced lesion [21,23,27,28]. MMP-3 concentrations were elevated in the molecular layer of the dentate after traumatic brain injury [29]. More specifically after ECL, administration of a non-selective MMP inhibitor was able to diminish sprouting and synaptogenesis in the dentate OML [22], suggesting a direct proteolytic role for the MMPs in this process. In adults, most MMP expression and activity is low and maintained throughout adulthood. After injury and during the recovery and regenerative phase, however, there is increased activity of MMPs derived from glia and neurons that is thought to facilitate axonal reinnervation, sprouting and/or synaptogenesis. Nonetheless, the mechanism(s) of action of the MMPs and the potential substrates on which they act to promote neural plasticity have yet to be determined in these models. More recently, the activity and expression of the PG-degrading, ADAMTSs have been shown to be elevated in the OML after kainite-induced lesion. In contrast to the absence of a defined substrate for the MMPs, a selective ADAMTS-derived, brevican fragment was localized to the OML after seizure-induced lesion in the rat [16]. In the present study, a similar ADAMTS-derived brevican fragment was localized to the OML of the mouse after discrete denervation of the perforant path, suggesting a critical role in neural plasticity for the proteolytic turnover of brevican. Thus, the ability to localize and quantitate the ADAMTS specific, proteolytic product of brevican provides a means to indirectly estimate ADAMTS activity during times of neural plasticity and synaptogenesis.
The expression of brevican was shown previously to be up-regulated in the OML, the area of denervation after ECL in the rat [10], however, in contrast to the transient production observed here in the mouse, expression of immunoreactive brevican remained elevated compared to the non-lesioned side for almost 6 months after injury. Neurocan is a lectican that is expressed at high levels during early development but it was found to be up-regulated and synthesized by astrocytes in the OML after ECL. It was suggested that neurocan and possibly brevican may act to maintain the boundary of the denervated dentate after ECL [30], yet these complex molecules may be multifunctional during periods of neural plasticity. Each of the lecticans, exhibit a characteristic pattern of expression during development, with neurocan and versican V1 highly expressed in the brain of the fetus and neonate, whereas aggrecan, versican V2 and brevican increase expression during the period of synaptic stabilization in the adult and expression remains high throughout adulthood [13,31]. Each of the lecticans is thought to bind to tenascin R and hyaluronic acid (with varying affinities) forming a multi-molecular lattice of ECM [14]. It may be that proteolytic cleavage of the lectican loosens the lattice to promote neurite growth and synaptogenesis. Classically, the highly negatively charged CS chains on the lecticans inhibits neurite outgrowth, but proteolytic cleavage of the core protein may allow more movement of these chains and actually promote plasticity of neurons. This is a testable notion and preliminary data indicates that the ADAMTSs promote neurite outgrowth and other measures of neural plasticity in vitro (our unpublished observations).
The projection from the EC to the hippocampus is called the perforant path [32], and is thought to be involved in long-term potentiation and learning and memory [33-35]. The ECL model to study neural and synaptic plasticity denervates up to 80% of the input to the outer two-thirds of the molecular layer of the dentate gyrus [36], and due to sprouting of surviving fiber systems will reinnervate nearly fully. This model was developed more than thirty years ago in the rat [37], yet it has been only relatively recently that the technique was employed in mice to take advantage of transgenic models [38,39]. Surprisingly, there are some differences in the projections from the EC to the hippocampal formation between rats and mice [40]. For example, input to the dentate molecular layer from the contralateral EC is absent in the mouse, yet these contralateral fibers are responsible for much of the sprouting after ECL in the rat. In addition, the width of the inner molecular layer, that contains associational-commissural fibers, is thinner in the mouse than in the rat, causing an increase in the relative width occupied by the middle and OMLs, layers innervated mainly by EC fibers. In the mouse, the middle and OML occupy closer to four-fifths of the total, rather than two-thirds as seen in the rat. Moreover, three layers can be clearly differentiated in an untreated mouse, but not in a rat, using synaptophysin immunohistochemistry [40], and following ECL this laminar feature is lost (Fig. 5B). Synaptophysin immunochemistry has been one of the more common techniques, among many, to quantitate the loss and reinnervation of input into the ipsilateral molecular layer of the dentate gyrus after unilateral ECL [41]. Optical density of the synaptophysin signal in the contralateral OML is measured using the sum (or average) of the gray levels of the pixels in this region, and this value is used as a "normal" value to the ipsilateral side. However, with this technique, the ipsilateral dentate usually shows only a 10–30% reduction in signal compared to the contralateral side at seven days after lesion, a time when sprouting has begun [42]. Clearly this absolute value does not reflect the extent to which fibers are actually lost in the OML after ECL. Thus, we decided to develop a fresh tissue, needle punch dissection technique that could be limited to the dentate gyrus in the mouse. This way, biochemical assays could be conducted on the tissue to measure overall synaptophysin immunoreactivity by ELISA. Using this method, at two and seven days after lesion, there was greater than a 40% reduction in synaptophysin levels in the OML, a value which at least may closer reflect the absolute loss of fibers after ECL. The major disadvantage of this method is that the dissected tissue also includes the granule layer and the hilus of the dentate, regions where input is not lost after ECL. At the same time using this technique, tissue containing the lesion itself may be collected and assayed biochemically or processed for histochemistry to monitor the extent of the lesion in the EC.
The present results suggest that the lecticans and the proteases that cleave the lecticans play a regulatory role in neural plasticity after ECL. There are several potential mechanisms by which this substrate – protease pair may modulate neural plasticity, one of which was described above. Significant changes were observed in the abundance of the different isoforms of brevican including the expression of the C-terminally-truncated ADAMTS-derived brevican fragment during plastic events in the hippocampus. A genetic approach to study the individual lecticans and ADAMTSs could reveal the individual contributions for each of the molecules involved in neural plasticity after ECL, however, there is considerable redundancy among these molecules. For example, there appears to be a compensatory increase in the expression of neurocan in the brain of the brevican null mouse. [43] In addition, several of the ADAMTSs exhibit proteoglycanase activity, and of these, at least ADAMTS1, 4 and 9 appear to be expressed in the nervous system (unpublished observations). Whether there are compensatory changes in the expression of any of these molecules in the brain in ADAMTS null mice remains to be determined. Nonetheless, significant protective effects toward arthritic changes were demonstrated just recently in a single mutant, the ADAMTS5 null mouse [44]. Should it turn out that these proteases play a significant role in plasticity related mechanisms in the nervous system, it will be interesting to examine how removing this regulatory action will impact development, sprouting after lesion, learning and memory and other plasticity related mechanisms in the adult.
Conclusion
Growth, sprouting and targeting in neural plasticity of an ECL model may involve extracellular cues whose expression and/or secretion is altered following the lesion. One of these cues is the extracellular PG, brevican. Here we have demonstrated changes in brevican expression and turnover after ECL that are associated with the loss and time-dependent reinnervation of the OML of the dentate gyrus.
Methods
Animals
All animal procedures described here were approved by the Institutional Animal Care and Use Committee at the University of South Florida. Sixty-two adult male C57Bl6 mice (23 g – 27 g; Harlan, Indianapolis, IN), 12 weeks of age were housed under a 12 h light cycle with regulated temperature and humidity. Mice were housed 3 to 4 per cage and had free access to food and water. Following ECL surgery, the animals were housed individually. Brains from control mice (n = 4) and lesioned mice surviving for 2 days (n = 6), 7 days (n = 5) and 30 days (n = 5) after the lesion were perfusion fixed and collected for immunohistochemistry. Tissue extracts of dentate gyrus and EC lesioned animals collected 2 days (n = 6), 7 days (n = 5) and 30 days (n = 6) after surgery were used in Western blot and biochemical immunoassays. For those animals that received a unilateral ECL, the contralateral, non-lesioned hemisphere was considered the control for immunohistochemistry and biochemical analysis.
Surgical procedures – the entorhinal cortex lesion (ECL)
Surgeries were performed using isofluorane/oxygen mixed gas anesthesia. Once deeply anesthetized, animals were placed into the stereotaxic apparatus. A hole was drilled in the skull of the right hemisphere to allow for needle penetration. The right, lateral EC of mice was unilaterally lesioned by lowering a needle attached to a Hamilton syringe (#701N) filled with ibotenic acid through the hole in the skull to the coordinates AP = 4.72 mm, L = 3.75 mm and DV = 4.70 mm using bregma as a reference and oriented 17° rostral-caudal [45]. One μl of the neurotoxin, ibotenic acid was injected into the lateral EC at a rate of 0.1 μl every 30 seconds, with the needle remaining in place for an additional minute at the end of the 5 min injection period to allow for complete diffusion of the drug into the lateral EC. Neurons of the lateral EC project preferentially to the septal dentate gyrus. The needle was removed, bone wax used to cover the skull hole, and the animal was allowed to recover on a heating pad after which the animal was returned to a new cage and housed individually. At 2, 7 and 30 days after lesion, mice were injected with an overdose of Nembutal (pentobarbital) for deep anesthesia, perfused transcardially with cold phosphate-buffered saline (pH 7.4) followed by cold 4% paraformaldehyde fixative diluted in 0.1 M phosphate buffer (pH 7.4). The brain was dissected from the skull, post-fixed in 4% paraformaldehyde overnight at 4°C, cryoprotected with consecutive solutions of 15 and 30% sucrose until completely infused, and the cryoprotected brain was sectioned on a cryostat at 30 μm. The extent and magnitude of the lesion in the EC was verified by cresyl violet and brevican staining. Synaptophysin and GFAP immunohistochemistry, using horseradish peroxidase amplification and diaminobenzidine as a chromogen, verified the loss of terminals in the molecular layer of the dentate gyrus.
Region isolation method
A method was developed to isolate dentate gyrus and EC tissue in mouse brain. This method eliminates the collection of much of the surrounding tissue that was unaffected by ECL. The dentate gyrus was subjected to synaptophysin ELISA as a measure of synaptic loss in the molecular layer. The EC underwent Western blot for PSD (post-synaptic density) 95 to quantitate the magnitude of the lesion. In addition, brevican and EAMESE Western blot was conducted on dentate gyrus tissue. Animals were completely anesthetized with a lethal dose of Nembutal (pentobarbital, Abbott Laboratories, North Chicago, IL), the brain was quickly removed and placed on a glass microscope slide, and a mm ruler was placed adjacent to the brain to assist in measuring the thickness of slices. Whole brain was dissected and frozen on a flat slab of dry ice for about 5 minutes (a point at which the brain surface is no longer shiny), being careful not to over-freeze the tissue, but to reach a temperature at which the tissue can be easily and efficiently cut into 2 mm coronal slabs. Two 2 mm coronal slabs are cut by referencing the most caudal of cerebral cortex as a land mark, making a coronal cut, moving rostrally 2 mm and making another cut to obtain the first slab for the EC isolation. A third cut was made 2 mm rostral from the previous slice and this slab contained the septal hippocampus and was used to isolate the dentate gyrus. The 2 mm hippocampal slab (outlined in Fig. 1C) was placed on a glass slide, and the dentate gyrus was localized using a stereomicroscope. The dentate was punctured with a blunt-ended 22 gauge needle (Fig. 1A) using the medial point of the wings as a landmark. A blunt-ended 18 gauge needle was used in the same manner to puncture and collect the EC from a 2 mm slab caudal to the slab used to collect the dentate gyrus (Fig. 1B). The frozen tissues were immediately expelled from the needle by attaching an air-loaded syringe to the needle and forcing the tissue into the bottom of a microfuge tube. The dentate gyrus and EC tissues were then homogenized in extraction buffer (20 nM Tris-HCL pH = 7.4, 5 mM EDTA, 1% Triton-X-100, & 1:100 protease inhibitor; 20 μl for dentate gyrus and 40 μl for EC) using three cycles of 2 min of 4°C incubation and 30 sec of vortex. The solubilized extract was centrifuged at 8000 rpm in a refrigerated microfuge for 3 min, the supernatant collected and frozen for later use in Western blotting and ELISA analysis.
Immunohistochemistry
Mice were anesthetized with an overdose of Nembutal (pentobarbital, Abbott Laboratories, North Chicago, IL) and their brains were fixed via standard cardiac perfusion methods. Briefly, the animal was cleared with phosphate buffered saline (PBS; pH 7.4) and fixed with 4% paraformaldehyde in 0.1 M phosphate buffer (PB; pH 7.4). The brains were post-fixed overnight in the same fixative; cryoprotected with 15% sucrose (in 0.1 M PB) followed by a 30% sucrose solution for 24 h each. Whole brain was frozen with mounting medium and cut into 30 μm coronal sections using a cryostat. Free-floating sections were washed with PBS, placed in a blocking/permeabilization solution (10% normal goat serum; 3% 1 M Lysine, 3% Triton-X) for 1 h, washed with PBS and incubated with primary antibody overnight at 4°C. For detection of the antigen, sections were washed and incubated in secondary antibody solution (anti-rabbit or anti-mouse IgG conjugated to Alexa-Fluor 488 or 594, Molecular Probes, Eugene, OR) for 1 h. The sections were washed for 15 minutes, wet mounted onto Fisher SuperFrost Plus glass sides and coverslipped with VectaShield mounting medium (Vector Labs, Burlingame, CA).
The primary antibodies or probes used in these experiments were: mouse anti-brevican, raised against the G1 domain of brevican (BD Transduction Labs, San Jose, CA); rabbit anti-EAMESE, the ADAMTS-derived neoepitope of brevican developed in our lab; rabbit anti-glial fibrillary acidic protein (GFAP, DAKO, Carpinteria, CA); and rabbit anti-synaptophysin (DAKO).
Western blotting
Dentate gryus or EC extracts were loaded (equal amounts of protein and 2 × sample buffer) on to 4–20% polyacrylamide gels (Invitrogen, Carlsbad, CA) and subjected to SDS-PAGE. Protein was transferred to a polyvinylidine difluoride membrane (PVDF, Immobilon, Millipore, Billerica, MA) and the membrane was blocked with 5% milk in PBS. Membranes were probed with primary antibodies against brevican (1:1000), EAM (1:500), and secondary anti-rabbit or anti-mouse IgG conjugated to horse radish peroxidase (Chemicon, Temecula, CA). Antigens were visualized using a chemiluminescence developing system (SuperSignal, Pierce, Rockford, IL).
Antibody generation
A rabbit antibody raised against the brevican neoepitope on the 55 kD N-terminal fragment derived from glutamyl endopeptidase activity of the ADAMTSs was generated by Sigma-Genosys (St. Louis, MO) and purified in our laboratory. The novel C-terminal sequence "QEAMESE" from the mouse was the neoepitope and the peptide used for antibody generation contained a glycine spacer "GGGQEAMESE". This peptide was synthesized by Sigma-Genosys, conjugated to keyhole limpet hemocyanin at the N-terminus, and rabbits were subjected to standard immunization protocols. Serum collected after the 5th booster was titered against the peptide using a solid-phase system and specific antibody was purified using peptide affinity chromatography. Interestingly, antibody raised against the rat epitope "QEAVESE" did not recognize the mouse, ADAMTS-derived, "QEAMESE" epitope. On Western blot, the antibody against the mouse fragment recognized a single band at 55 kD in extracts from mouse brain and did not cross react with the intact brevican core protein.
Authors' contributions
JM performed the surgical procedures and carried out the immunoassays, statistical analysis and drafted the manuscript. MH was involved in the development of the synaptophysin ELISA. PG conceived of the study, participated in its design and coordination, performed statistical analysis and helped draft the manuscript. All authors read and approved the final manuscript.
Acknowledgements
The authors would like to acknowledge that this work was supported in part by Alzheimer's Association Grant IIRG-02-3758 and a grant from the Shriners Hospitals (research grant #8850). Dr. John Sandy (Shriners Hospital, Tampa, FL) was instrumental in the conceptual development of the neoepitope antibody.
Figures and Tables
Figure 1 Schematic representation of the micro-dissection of dentate gyrus and entorhinal cortex: This procedure was developed to isolate dentate gyrus and entorhinal cortex tissue from different rostral-caudal slabs in fresh-frozen mouse brain. A: Coronal section of mouse brain showing bore holes left from blunt-ended, 22 gauge needle punch (green) in the ipsilateral and contralateral dentate gyrus. B: Coronal section of mouse brain (2 mm caudal from section in "A") showing bore holes left from blunt-ended, 18 gauge needle punch (green) in the ipsilateral and contralateral entorhinal cortex. C: Sagittal view of 2 mm slices used for the collection of dentate gyrus (A) and entorhinal cortex (B). Diagrams adapted from 'The Mouse Brain' [46].
Figure 2 Schematic representation of brevican cleavage by the glutamyl-endopepidases, the ADAMTSs: A: Secreted brevican core protein that bears chondroitin sulfate chains (MW > 145 kD). B: Secreted brevican core protein without chondroitin sulfate side chains (MW = 145 kD). C: When cleaved by extracellular glutamyl endopeptidases, the ADAMTSs, an N-terminal fragment of brevican (MW = 55 kD) is formed that contains the unique C-terminal epitope EAMESE. The anti-brevican antibody recognizes all three isoforms of brevican, whereas anti-EAMESE antibody selectively recognizes the ADAMTS-derived proteolytic fragment of brevican and not the parent protein or other isoforms.
Figure 3 Brevican and EAMESE Western blotting in regions of mouse brain: Panel A shows an immunoblot containing extracts from various regions of mouse brain and probed with anti-brevican antibody. Immunoreactive bands are present at >145 kD (high molecular weight smear), 145 kD core protein, and the generalized 55 kD fragment. Panel B contains a membrane with the same samples probed with anti-EAMESE. This antibody recognized the specific ADAMTS-derived 55 kD fragment of brevican. CB = cerebellum, BS = brain stem, DE = diencephalon, HC = hippocampus, FC = frontal cortex, TL = temporal lobe, and OT = olfactory tubercle.
Figure 4 Alterations in brevican levels in lesioned entorhinal cortex: Horizontal sections of entorhinal cortex immunostained with anti-brevican antibody collected two days after entorhinal cortex lesion. A: Contralateral entorhinal cortex showed little immunoreactive signal. B: Ipsilateral entorhinal cortex showed abundant anti-brevican staining in the lateral region.
Figure 5 Denervation of the outer molecular layer of the dentate gyrus after entorhinal cortex lesion: A and B: Synaptophysin immunoreactivity in the molecular layer of the dentate gyrus two days after ECL A: ipsilateral dentate, B: contralateral dentate. Note diminished immunostaining and loss of tri-laminar pattern in the OML after ECL. C and D: GFAP immunoreactivity in the molecular layer of the dentate gyrus seven days after ECL. Note increased astroglial expression of GFAP observed after deafferentation of the ipsilateral (C), but not contralateral (D), OML. E: Synaptophysin levels in extracts from fresh micro-dissected (needle punch, see Fig. 1) dentate gyrus tissue as measured by ELISA. Synaptophysin immunoreactivity was reduced two and seven days post-lesion. The ipsilateral side was not different from contralateral dentate thirty days after lesion. oml = outer molecular layer, mml = middle molecular layer, iml = inner molecular layer, and gcl = granule cell layer.
Figure 6 Brevican immunoreactivity in the dentate gyrus after entorhinal cortex lesion: Brevican immunoreactivity in the contralateral (CL) and ipsilateral (IL) dentate gyrus of mice that had undergone ECL two, seven, and thirty days earlier. Optical density was measured on the Western blots and data from the ipsilateral side was expressed as a fraction of the contralateral dentate. A: At two days, the CS-containing, >145 kD brevican was significantly elevated on the lesion side compared to the contralateral side although all fragments showed a trend for increase. B: At seven days, there was a significant elevation of the ADAMTS-derived EAMESE fragment of brevican and a trend for an increase in 55 kD fragment. C: At thirty days, brevican isoforms were not different from the control side, except that the >145 kD core protein was slightly elevated.
Figure 7 Apparent ADAMTS activity: Apparent activity in the dentate gyrus of mice after entorhinal cortex lesion as measured by the ratio of the optical density of the ADAMTS-derived EAMESE fragment and the sum of the densities of the brevican core protein isoforms (>145 kD + 145 kD). Mean level of apparent activity was calculated at two, seven and thirty days after lesion. At two days after lesion, ADAMTS activity declined, whereas at seven days a significant 47% increase in ADAMTS activity was observed compared to the contralateral ADAMTS activity. Levels on the lesioned side were not different from the non-lesioned dentate in tissue collected thirty days after lesion.
==== Refs
van Groen T Miettinen P Kadish I The entorhinal cortex of the mouse: organization of the projection to the hippocampal formation Hippocampus 2003 13 133 149 12625464 10.1002/hipo.10037
Steward O Cholinergic sprouting is blocked by repeated induction of electroconvulsive seizures, a manipulation that induces a persistent reactive state in astrocytes Exp Neurol 1994 129 103 111 7925832 10.1006/exnr.1994.1151
Frotscher M Deller T Heimrich B Forster E Haas C Naumann T Survival, regeneration and sprouting of central neurons: the rat septohippocampal projection as a model Anat Anz 1996 178 311 315
Deller T Haas CA Frotscher M Sprouting in the hippocampus after entorhinal cortex lesion is layer- specific but not translaminar: which molecules may be involved? Restor Neurol Neurosci 2001 19 159 167 12082219
Del Turco D Woods AG Gebhardt C Phinney AL Jucker M Frotscher M Deller T Comparison of commissural sprouting in the mouse and rat fascia dentata after entorhinal cortex lesion Hippocampus 2003 13 685 699 12962314 10.1002/hipo.10118
Lynch G Matthews DA Mosko S Parks T Cotman C Induced acetylcholinesterase-rich layer in rat dentate gyrus following entorhinal lesions Brain Res 1972 42 311 318 4115093 10.1016/0006-8993(72)90533-1
Snow DM Lemmon V Carrino DA Caplan AI Silver J Sulfated proteoglycans in astroglial barriers inhibit neurite outgrowth in vitro Exp Neurol 1990 109 111 130 2141574 10.1016/S0014-4886(05)80013-5
Bandtlow CE Zimmermann DR Proteoglycans in the developing brain: new conceptual insights for old proteins Physiol Rev 2000 80 1267 1290 11015614
Rauch U Extracellular matrix components associated with remodeling processes in brain Cell Mol Life Sci 2004 61 2031 2045 15316653 10.1007/s00018-004-4043-x
Thon N Haas CA Rauch U Merten T Fassler R Frotscher M Deller T The chondroitin sulphate proteoglycan brevican is upregulated by astrocytes after entorhinal cortex lesions in adult rats Eur J Neurosci 2000 12 2547 2558 10947829 10.1046/j.1460-9568.2000.00109.x
Yamaguchi Y Brevican: a major proteoglycan in adult brain Perspect Dev Neurobiol 1996 3 307 317 9117262
Bruckner G Grosche J Hartlage-Rubsamen M Schmidt S Schachner M Region and lamina-specific distribution of extracellular matrix proteoglycans, hyaluronan and tenascin-R in the mouse hippocampal formation J Chem Neuroanat 2003 26 37 50 12954529 10.1016/S0891-0618(03)00036-X
Hockfield S Kalb RG Zaremba S Fryer H Expression of neural proteoglycans correlates with the acquisition of mature neuronal properties in the mammalian brain Cold Spring Harb Symp Quant Biol 1990 55 505 514 2132834
Yamaguchi Y Lecticans: organizers of the brain extracellular matrix Cell Mol Life Sci 2000 57 276 289 10766023
Snow DM Letourneau PC Neurite outgrowth on a step gradient of chondroitin sulfate proteoglycan (CS-PG) J Neurobiol 1992 23 322 336 1624935 10.1002/neu.480230311
Yuan W Matthews RT Sandy JD Gottschall PE Association between protease-specific proteolytic cleavage of brevican and synaptic loss in the dentate gyrus of kainate-treated rats Neuroscience 2002 114 1091 1101 12379262
Gottschall PE Sandy JD Zimmermann DR Conant K and Gottschall PE Matrix metalloproteinases in the Central Nervous System 2005 Singapore, World Scientific Publishing Co.
Matthews RT Gary SC Zerillo C Pratta M Solomon K Arner EC Hockfield S Brain-enriched hyaluronan binding (BEHAB)/brevican cleavage in a glioma cell line is mediated by a disintegrin and metalloproteinase with thrombospondin motifs (ADAMTS) family member J Biol Chem 2000 275 22695 22703 10801887 10.1074/jbc.M909764199
Nakamura H Fujii Y Inoki I Sugimoto K Tanzawa K Matsuki H Miura R Yamaguchi Y Okada Y Brevican is degraded by matrix metalloproteinases and aggrecanase-1 (ADAMTS4) at different sites J Biol Chem 2000 275 38885 38890 10986281 10.1074/jbc.M003875200
Phillips LL Reeves TM Interactive pathology following traumatic brain injury modifies hippocampal plasticity Restor Neurol Neurosci 2001 19 213 235 12082223
Szklarczyk A Lapinska J Rylski M McKay RD Kaczmarek L Matrix metalloproteinase-9 undergoes expression and activation during dendritic remodeling in adult hippocampus J Neurosci 2002 22 920 930 11826121
Reeves TM Prins ML Zhu J Povlishock JT Phillips LL Matrix metalloproteinase inhibition alters functional and structural correlates of deafferentation-induced sprouting in the dentate gyrus J Neurosci 2003 23 10182 10189 14614076
Zhang X Gelowitz DL Lai CT Boulton AA Yu PH Gradation of kainic acid-induced rat limbic seizures and expression of hippocampal heat shock protein-70 Eur J Neurosci 1997 9 760 769 9153582
Jaworski DM Kelly GM Hockfield S Intracranial injury acutely induces the expression of the secreted isoform of the CNS-specific hyaluronan-binding protein BEHAB/brevican Exp Neurol 1999 157 327 337 10364444 10.1006/exnr.1999.7062
Eng LF Ghirnikar RS GFAP and astrogliosis Brain Pathol 1994 4 229 237 7952264
Steward O Loesche J Quantitative autoradiographic analysis of the time course of proliferation of contralateral entorhinal efferents in the dentate gyrus denervated by ipsilateral entorhinal lesions Brain Res 1977 125 11 21 851867 10.1016/0006-8993(77)90356-0
Zhang JW Deb S Gottschall PE Regional and age-related expression of gelatinases in the brains of young and old rats after treatment with kainic acid Neurosci Lett 2000 295 9 12 11078924 10.1016/S0304-3940(00)01582-2
Jourquin J Tremblay E Decanis N Charton G Hanessian S Chollet AM Le Diguardher T Khrestchatisky M Rivera S Neuronal activity-dependent increase of net matrix metalloproteinase activity is associated with MMP-9 neurotoxicity after kainate Eur J Neurosci 2003 18 1507 1517 14511330 10.1046/j.1460-9568.2003.02876.x
Kim HJ Fillmore HL Reeves TM Phillips LL Elevation of hippocampal MMP-3 expression and activity during trauma-induced synaptogenesis Exp Neurol 2005 192 60 72 15698619 10.1016/j.expneurol.2004.10.014
Haas CA Rauch U Thon N Merten T Deller T Entorhinal cortex lesion in adult rats induces the expression of the neuronal chondroitin sulfate proteoglycan neurocan in reactive astrocytes J Neurosci 1999 19 9953 9963 10559403
Milev P Maurel P Chiba A Mevissen M Popp S Yamaguchi Y Margolis RK Margolis RU Differential regulation of expression of hyaluronan-binding proteoglycans in developing brain: aggrecan, versican, neurocan, and brevican Biochem Biophys Res Commun 1998 247 207 212 9642104 10.1006/bbrc.1998.8759
Hjorth-Simonsen A Jeune B Origin and termination of the hippocampal perforant path in the rat studied by silver impregnation J Comp Neurol 1972 144 215 232 4112908 10.1002/cne.901440206
Liu P Bilkey DK Direct connection between perirhinal cortex and hippocampus is a major constituent of the lateral perforant path Hippocampus 1996 6 125 135 8797014 10.1002/(SICI)1098-1063(1996)6:2<125::AID-HIPO4>3.0.CO;2-O
Sutula T Steward O Facilitation of kindling by prior induction of long-term potentiation in the perforant path Brain Res 1987 420 109 117 3676745 10.1016/0006-8993(87)90245-9
Ruthrich H Dorochow W Pohle W Ruthrich HL Matthies H Colchicine-induced lesion of rat hippocampal granular cells prevents conditioned active avoidance with perforant path stimulation as conditioned stimulus, but not conditioned emotion Physiol Behav 1987 40 147 154 3628522 10.1016/0031-9384(87)90200-9
Scheff SW Dekosky ST Glucocorticoid suppression of lesion-induced synaptogenesis: effect of temporal manipulation of steroid treatment Exp Neurol 1989 105 260 264 2767199 10.1016/0014-4886(89)90128-3
Steward O Cotman CW Lynch GS Re-establishment of electrophysiologically functional entorhinal cortical input to the dentate gyrus deafferented by ipsilateral entorhinal lesions: innervation by the contralateral entorhinal cortex Exp Brain Res 1973 18 396 414 4778785 10.1007/BF00239108
Shi B Stanfield BB Differential sprouting responses in axonal fiber systems in the dentate gyrus following lesions of the perforant path in WLDs mutant mice Brain Res 1996 740 89 101 8973802 10.1016/S0006-8993(96)00849-9
Destrade C Gauthier M Cazala P Caudarella M Lesion of the temporo-ammonic perforant path facilitates self-stimulation of the lateral entorhinal cortex in mice Brain Res 1985 344 377 381 3840048 10.1016/0006-8993(85)90819-4
van Groen T Kadish I Wyss JM Species differences in the projections from the entorhinal cortex to the hippocampus Brain Res Bull 2002 57 553 556 11923027 10.1016/S0361-9230(01)00683-9
Deller T Frotscher M Lesion-induced plasticity of central neurons: sprouting of single fibres in the rat hippocampus after unilateral entorhinal cortex lesion Prog Neurobiol 1997 53 687 727 9447617 10.1016/S0301-0082(97)00044-0
Wilson RC Changes in translation of synaptic excitation to dentate granule cell discharge accompanying long-term potentiation. I. Differences between normal and reinnervated dentate gyrus J Neurophysiol 1981 46 324 338 6267216
Brakebusch C Seidenbecher CI Asztely F Rauch U Matthies H Meyer H Krug M Bockers TM Zhou X Kreutz MR Montag D Gundelfinger ED Fassler R Brevican-deficient mice display impaired hippocampal CA1 long-term potentiation but show no obvious deficits in learning and memory Mol Cell Biol 2002 22 7417 7427 12370289 10.1128/MCB.22.21.7417-7427.2002
Stanton H Rogerson FM East CJ Golub SB Lawlor KE Meeker CT Little CB Last K Farmer PJ Campbell IK Fourie AM Fosang AJ ADAMTS5 is the major aggrecanase in mouse cartilage in vivo and in vitro Nature 2005 434 648 652 15800625 10.1038/nature03417
White F Nicoll JA Horsburgh K Alterations in ApoE and ApoJ in relation to degeneration and regeneration in a mouse model of entorhinal cortex lesion Exp Neurol 2001 169 307 318 11358444 10.1006/exnr.2001.7655
Paxinos Franklin The Mouse Brain: in stereotaxic coordinates 2001 San Diego, Academic Press
|
16122387
|
PMC1199600
|
CC BY
|
2021-01-04 16:03:48
|
no
|
BMC Neurosci. 2005 Aug 25; 6:52
|
utf-8
|
BMC Neurosci
| 2,005 |
10.1186/1471-2202-6-52
|
oa_comm
|
==== Front
BMC PediatrBMC Pediatrics1471-2431BioMed Central London 1471-2431-5-231601416610.1186/1471-2431-5-23Research ArticleAssociation between congenital toxoplasmosis and parent-reported developmental outcomes, concerns, and impairments, in 3 year old children Freeman Katherine [email protected] Alison [email protected] Andrea [email protected] Gunilla [email protected] Nicole [email protected] Wilma [email protected] Dorthe [email protected] Hooi Kuan [email protected] Ruth E [email protected] European Multicentre Study on Congenital Toxoplasmosis (EMSCOT) [email protected] Albert Einstein College of Medicine, Department of Epidemiology and Population Health, New York, U.S.A2 The Neurodisability Service, Great Ormond Street Hospital for Children and Institute of Child Health, London, UK3 Department of Pediatrics, Division of Neonatology and Intensive Care, Medical University of Vienna, Austria4 Karolinska University Hospital, Huddinge, Stockholm, Sweden5 CHU de NICE, Service Parasitologie – Mycologie, Hopital L'Archet II, BP 3079, 06202 NICE Cedex 3, France6 Perinatal Infection Unit, Dept of Pediatrics, University of Naples Federico II, Naples, Italy7 Department of Parasitology, Staten Seruminstitut, Copenhagen, Denmark8 Centre for Paediatric Epidemiology and Biostatistics, Institute of Child Health, London, UK2005 13 7 2005 5 23 23 27 10 2004 13 7 2005 Copyright © 2005 Freeman et al; licensee BioMed Central Ltd.2005Freeman et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Information is lacking on the effects of congenital toxoplasmosis on development, behavior, and impairment in later childhood, as well as on parental concerns and anxiety. This information is important for counselling parents about the prognosis for an infected child and for policy decisions on screening.
Methods
We prospectively studied a cohort of children identified by screening for toxoplasmosis in pregnant women or neonates between 1996 and 2000 in ten European centers. At 3 years of age, parents of children with and without congenital toxoplasmosis were surveyed about their child's development, behavior, and impairment, and about parental concerns and anxiety, using a postal questionnaire.
Results
Parents of 178/223 (80%) infected, and 527/821 (64%) uninfected children responded. We found no evidence that impaired development or behavior were more common in infected children, or that any potential effect of congenital toxoplasmosis was masked by prenatal treatment. Parents of infected children were significantly more anxious and reported more visual problems in their children.
Conclusion
On average, children aged three to four years with congenital toxoplasmosis identified by screening and treated during infancy in this European setting had risks of abnormal development and behavior similar to uninfected children. Parental anxiety about infected children needs to be addressed by clinicians. Future studies with longer follow up and clinician-administered assessments may be better able to detect any subtle differences in child outcomes.
==== Body
Background
Congenital toxoplasmosis is associated with a wide spectrum of clinical signs and symptoms. At its most severe, congenital toxoplasmosis causes death or severe disability in 1–4% of infants identified by prenatal or neonatal screening[1]. Although the remaining 96% of infected infants appear clinically normal in infancy, one in six have intracranial and/or ocular lesions. Clinicians lack information for counselling parents about their child's functional abilities in later childhood, and whether intracranial or ocular lesions predict adverse functional outcomes. In addition, policy makers need to know what proportion of children with congenital toxoplasmosis suffer long term functional impairment. However, a systematic search of the literature found only one study of children identified by screening in which school performance in 11 infected children was compared with their peers at 7 years[2]. No difference was found but this may be due to the sample size and/or the insensitivity of the outcome measure. Other studies of developmental outcomes have been based on case series of referred and usually symptomatic children with congenital toxoplasmosis, and have not included an appropriate comparison group[3,4].
We wanted to know whether 3-year-old children with congenital toxoplasmosis are more at risk of adverse developmental or behavioral outcomes than uninfected children. We also investigated parental concerns and anxiety as clinicians caring for infected children highlighted parental anxiety as a common clinical problem. We conducted a prospective cohort study of children identified by prenatal or neonatal screening for toxoplasmosis. Infected and uninfected children born to infected women were followed up during infancy and then surveyed using a parent-completed questionnaire to assess development and other outcomes when the child turned 3 years. This design aimed to ensure that, apart from congenital infection status, the experience of screening, and follow up in early infancy, was similar. However, only infected children received prolonged postnatal treatment and follow up.
Methods
Study population
We compared children, with and without congenital toxoplasmosis, born to women identified by prenatal screening for maternal toxoplasmosis between 1996 and 2000 in eight centers (Lyon, Paris, Marseille, Toulouse, Nice, Grenoble, Vienna, and Naples), and by neonatal screening in two centers (Stockholm, Poznan). One other center, Denmark, was excluded from the analyses as no uninfected children were recruited. Details of the methods have been reported elsewhere [5,6]. In brief, 91% of the women in the prenatal screening centers[6], but none in the neonatal screening centers, received anti-toxoplasma treatment before birth. In Poznan (Poland), children were identified by universal neonatal screening for specific IgM in filter paper blood spots from the Guthrie card and uninfected controls were selected as the next six children with a negative screening test born after each infected child[5].
Clinical follow up
Women suspected to have acquired toxoplasma infection during pregnancy and infected infants identified by neonatal screening were enrolled prospectively, prior to the collection of follow up data. We used a standard questionnaire to record clinical findings during pregnancy, at pediatric examinations in the neonatal period, at six and 12 months, and at ophthalmoscopy before four months and at 12 months of age. Cranial ultrasound was performed within the first four months of life. The number of examinations (pediatric, ophthalmic, and cranial ultrasound) was similar in infected and uninfected children up to 4 months of age.
Confirmation of congenital infection status was based on the persistence of IgG antibodies after 11.5 months of age (infected), or undetectable specific IgG antibodies in the absence of treatment, which usually occurred between 8 and 12 months[6]. To avoid potential biases due to exclusion of 15% of infants who did not meet these confirmatory criteria, we used probability estimations of their congenital infection status based on PCR analysis of amniotic fluid, specific IgM or IgA in the infant, last available IgG results, and the weeks of gestation at maternal seroconversion[6]. All infected children received treatment from early infancy for 12 to 24 months, depending on the centre, based on the results of prenatal diagnosis, and/or a positive IgM/IgA test, or a lack of decline in specific IgG titers. The median age at the start of postnatal treatment in prenatal centers was 2 days (IQR: 0, 14), and in neonatal screening centers 26 days (IQR: 22, 33). Few uninfected children were treated postnatally. As prenatal treatment after fetal diagnosis, and postnatal treatment, were given predominantly to those with evidence of congenital infection, postnatal treatment could not be investigated in this analysis. In Poznan, the uninfected children had no involvement in the cohort until their parents were sent a questionnaire when they turned 3 years. As no clinical follow up information was available for these children, they are excluded from analyses of children with no signs or symptoms in infancy.
We included all children identified by prenatal or neonatal screening in the parent-questionnaire study at 3 years, with some exceptions. First, in Austria and Italy, we randomly selected 4 uninfected children for each infected child, stratified by year of birth, to avoid surveying approximately 9 uninfected children for every infected child in these centers. Second, to minimize potential selection biases, we excluded children with non-sequential dates for the detection of maternal infection, fetal infection or abnormality, or, in the neonatal centre, for screening and confirmatory tests, as these women or children may have been referred. Third, 21 infected children known to be lost to follow up were not surveyed (4 died, but no serious outcomes were documented by paediatricians in the remainder).[7] Fourth, two centers participating in the initial cohort[6] did not participate in the 3 year follow up (Reims, Milan), and one did not enroll uninfected controls (Copenhagen). A detailed description and evaluation of questionnaire response has been reported elsewhere.[7] In brief, the questionnaire was composed of separate assessment tools (in total or in part) for behavior, speech and language, cognition, and motor skills, that have been previously validated against clinician assessments[7]. To-date the questionnaire as a whole has not been validated with standardized clinician assessments. The postal questionnaire, with a stamped addressed envelope for reply, and crayons for the child, was sent to parents by the local study centre when the child turned 3 years. Non-responders were sent two reminders at 2-monthly intervals. Research Ethics approval was obtained for all participating centres.
Outcomes
The results for the effect of congenital toxoplasmosis and potential confounders are presented for three groups of outcomes. The first comprised developmental outcomes (gross motor, speech and language, and cognitive), and abnormal behavior. These were measured by questions derived from standardized tools that had been validated against clinician assessments[7]. The questionnaire also included two child-completed sections that assessed cognitive and fine motor development. Children were asked to copy a line, circle and cross, drawn by their parent, and secondly, to draw a man[8]. Child-completed measures were analyzed separately from parent-completed measures because not all children participated, and we could not standardize the degree of assistance given to the child.
The second group of outcomes measured parental concerns about development, learning, behavior, and speech and language, specialist referral, and asked parents to rate how worried they were about their child's general health at the time and in the future[9]. Parental concerns about specific areas of development, or specialist referral may be proxy markers of abnormal development[10,11]. More general anxiety about the child's health now and in the future may reflect anxiety generated by diagnosis, treatment and follow up[9], but may also be a marker for abnormality. The third group of outcomes comprised specific impairments that parents were asked about, including difficulty hearing or seeing, and whether the child had cerebral palsy or seizures. The primary outcomes were the scores for motor development, speech and language, cognition, and behaviour, and parental anxiety.
Analyses
We aimed to compare infected and uninfected children, stratified by centre because of differences in screening programs and treatment regimens, and possible differences in expectations and attitudes to congenital toxoplasmosis. There was also significant variation among centers for exposure and outcome variables. We therefore present the preliminary bivariate analyses to show the effect of centre on outcomes, and adjust all subsequent analyses for center, nested within country, and congenital infection status. Stockholm and Naples were grouped as one centre, due to small numbers. Outcome variables were defined by a score measured on an interval scale (further details reported elsewhere[6]. Abnormality was defined by a cut-point approximating the least able 10% of controls, or for behavior, using an established cut point for abnormality (equivalent to 10% in a large community sample in which the questionnaire was validated[12]). Missing data for mother's age was imputed using a procedure for both continuous and categorical variables[13]
Multivariate analyses of the effect of congenital toxoplasmosis included centre nested within country, maternal education, and child's age at questionnaire completion, in every model based on evidence of confounding in some comparisons and consensus among investigators as to their clinical relevance. We included other potential confounders with p values = 0.20 from the bivariate analyses[14], but retained those variables with p-values <0.05 in the final model. Where the effect of congenital toxoplasmosis was significant (= 0.05), we tested for an interaction between centre and infection status. We used a hierarchical generalized linear model to account for heterogeneity among centres within France and between centres inside and outside France[15] A generalized estimating equation (SAS Version 9.1 PROC GENMOD with the ASSESS options to assess fit of the model) with centers nested within country was derived to determine characteristics associated with each dichotomized outcome. Odds ratios were derived by exponentiating parameter estimates, as well as lower and upper bounds for corresponding 95% confidence intervals. For logistic regression, Wald estimates were used.
We performed several sensitivity analyses for the association between congenital toxoplasmosis and development and behavior, and parental concerns and anxiety, and specialist referral. First, to avoid spurious associations due to dichotomization of outcomes, we repeated the main analyses using ordinal logistic regression for all outcomes measured using an ordinal scale (motor, speech and language, and cognitive development, behavior, child completed drawings, impact of behavior on family, and parental anxiety). Second, for the main model with dichotomized outcomes primary analyses using country as a fixed effect, were repeated using a generalized mixed model approach with country as a random effect Second, to test whether the effect of congenital toxoplasmosis on outcomes differed across countries, we derived separate multivariate models for each country. Third, to minimize the confounding effect of child's age at questionnaire completion on developmental outcomes, we restricted analyses to children aged 36 to 40 months. Fourth, to minimize confounding due to clinical manifestations detected in early infancy, we restricted analyses to children who had no clinical signs identified by the first 4 months of life. Signs were defined as intracranial calcification or ventricular dilatation on ultrasound, retinochoroiditis, lymphadenopathy or hepatosplenomegaly, microphthalmia, microcephaly, seizures, or an abnormal neurological examination. These analyses excluded Poland, where uninfected children were not assessed by the study during infancy. Fifth, to determine whether the duration of prenatal treatment may have masked an effect of congenital toxoplasmosis, we analyzed the effect of congenital toxoplasmosis, adjusted for prenatal treatment duration, and gestational age at maternal seroconversion, in French women who seroconverted during pregnancy. In these women, the gestational age at seroconversion could be estimated fairly precisely based on the midpoint between a negative and positive specific IgM test (usually a one month interval), or 14 days before a positive specific IgM result and negative IgG result[16]. Finally, to detect any adverse effects of prenatal treatment, we confined analyses to uninfected children, using the same approach as for the main model.
Without taking into account adjustment for confounders, the study was designed to have 80% power to detect a minimum difference of 0.22 standard deviations, assuming 70% response in the controls, 80% response in infected children, and an alpha error of 0.05.
Results
Study population
Of the 1044 families sent a questionnaire in 10 centers, parents of 178/223 (80%) infected, and 527/821 (64%) uninfected children responded. The distribution of infected and uninfected children by centre was: Lyon (40, 94), Paris (41,50), Marseille (18, 51), Nice (5, 28), Toulouse (18, 26), Vienna (18, 116), Stockholm (2, 6), Naples (10, 40), Poznan (23, 111). Details about factors associated with response, reasons for non-response, and exclusions due to death or potential referral bias, are reported elsewhere7. Most parents completed the entire questionnaire, as shown by the denominator for each outcome in Table 1. However, far fewer children completed the drawings: 70% (493/705) completed the draw a man test; and 95% (670/705) copied a line, circle or cross. Few parents reported neurological problems or impaired mobility (N = 6), but visual impairment (n = 10), and hearing loss was more common (N = 37).
Table 1 Number of children with an adverse outcome according to congenital infection status (%)
1) DEVELOPMENT AND BEHAVIOUR
PARENT COMPLETED
Infection status Motor1 Speech & language1 Behaviour2 Cognition1
CT+ (n = 178) 24/178 (13.5) 20/177 (11.3) 34/176 (19.3) 9/178 (5.1)
CT- (n = 527) 58/525 (11.0) 50/524 (9.5) 97/527 (18.4) 31/525 (5.9)
CHILD COMPLETED
Infection status 'Draw a Man'1 Line, circle, cross1
CT+ (n = 178) 17/121(14.0) 13/170(7.6)
CT- (n = 527) 54/372(14.5) 38/500(7.4)
2) PARENTAL CONCERNS, SPECIALIST REFERRAL, AND PARENTAL ANXIETY
CONCERNS
Learning/ development3 Speech & language3 ANY3
CT+ (n = 178) 24/176(13.6) 16/177(9.0) 34/178(19.1)
CT- (n = 527) 53/522(10.2) 55/522(10.5) 79/527(15.0)
REFERRALS
Learning/ development Speech & language ANY Behaviour impact 2 Parental anxiety1
CT+ (n = 178) 10/178(5.6) 4/178(2.3) 11/178(6.2) 3/173 (1.7) 41/176 (23.3)
CT- (n = 527) 13/527(2.5) 14/527(2.7) 20/527(3.8) 16/509 (3.1) 61/521 (11.7)
3) PARENT-REPORTED IMPAIRMENT
VISUAL IMPAIRMENT
Glasses only Strabismus/ blind/limited vision ANY Hearing loss4 Neurological/mobility impairment5
CT+ (n = 178) 5 10 15/177 (8.5) 7/166 (4.2) 2/175 (1.1)
CT- (n = 527) 11 13 24/527 (4.6) 30/498(6.0) 4/522 (0.8)
Full questionnaire available from the EMSCOT website [22]
CT = congenital toxoplasmosis
1Adverse outcome defined by score corresponding most closely to the least able/most worried 10% in the uninfected children
2Adverse outcome defined by developers of SDQ questionnaire. Equivalent to lowest 10% in community sample of 10,000 examined in evaluation study. [12]
3Defined by answering 'yes' or 'a little' to question(s) about concerns.
4Parents reported intermittent or permanent hearing loss
5Parents reported seizures requiring treatment, cerebral palsy, and/or impaired mobility
Potential confounders
The results for the associations between exposures and the three groups of outcomes are shown in Tables 2,3 and 4, adjusted for center and congenital infection status. Maternal age and education level, and child's age at questionnaire completion had significant effects on parent-reported developmental outcomes (Table 2), but not on parental concerns and anxiety (Table 3), or specific impairments (Table 4). Adverse cognitive development was reported more commonly in Grenoble compared with Lyon, and behavior problems were reported more commonly in Marseille. There were significant associations between congenital infection status and parental anxiety (Table 3). Parents in Poznan, reported more anxiety than in Lyon, and children in Poznan, and Stockholm/Naples were more likely to be referred to a specialist. There were more parental concerns about children in Poznan than in Lyon, and about girls than boys. Table 4 shows that visual impairment was more common in infected children, and was reduced in children with longer duration of prenatal treatment. Hearing loss appeared to be less common with increasing parity.
Table 2 Factors associated with development and behaviour at 3 years, adjusted for centre and congenital infection status (Total N = 705)
EXPOSURES Infection status Odds ratio for abnormal developmental outcome (95% confidence interval)
Parent completed Child completed
Before birth CT+ CT- Motor Speech & language Behaviour Cognition 'Draw a Man' Line, circle, cross
1) Congenital toxoplasmosis (reference = uninfected) 178 525 1.35
(0.79, 2.30) 1.32
(0.75, 2.33) 1.10
(0.70, 1.73) 0.88
(0.37, 2.08) 0.94
(0.51, 1.74) 1.05
(0.53, 2.07)
2) Maternal age (yrs) 1*
<25 40 86 1.07
(1.02, 1.12) 0.93
(0.88, 0.99) 0.92
(0.87, 0.97) 0.97
(0.90, 1.05) 1.04
(0.99, 1.09) 1.00
(0.93, 1.07)
25–35 106 288
>35 27 35
3) Parity1*
0 40 179 1.00
(0.80, 1.24) 1.18
(0.94, 1.48) 0.97
(0.74, 1.27) 1.22
(0.93, 1.60) 0.96
(0.70, 1.31) 1.02
(0.74, 1.40)
= 1 72 212
Missing 66 136
4) Maternal education1
Primary 14 48 1.15
(0.89, 1.48) 0.58
(0.44, 0.77) 0.77
(0.63, 0.94) 0.74
(0.52, 1.05) 1.20
(0.87, 1.66) 1.01
(0.73, 1.38)
Secondary Lower 47 181
Secondary Upper 47 135
Further Education 66 152
Missing 4 11
5) Mother born outside country2
Yes 25 60 1.45
(0.74, 2.84) 2.30
(1.14, 4.63) 1.65
(0.92, 2.96) 2.25
(0.88, 5.78) 0.89
(0.39, 2.01) 1.66
(0.72, 3.83)
no (reference) 152 462
Missing 1 5
6) Child's gender2
Male (reference) 95 284 1.45
(0.74, 2.84) 0.81
(0.49, 1.36) 1.65
(0.92, 2.96) 2.25
(0.88, 5.78) 0.89
(0.39, 2.01) 1.66
(0.72, 3.83)
Female 83 232
Missing 0 11
7) Duration of prenatal treatment1 (mean wks, sd) 6.3
(6.5) 15.1
(12.2) 1.00
(0.98, 1.02) 1.02
(0.99, 1.05) 1.00
(0.98, 1.02) 0.99
(0.96, 1.02) 0.98
(0.96, 1.01) 1.00
(0.98, 1.03)
Missing 0 0
Exposures after birth
8) Child's age at questionnaire1 (months) Median, IQR 38
(37,40) 39
(38, 41) 0.81
(0.71, 0.93) 0.87
(0.77, 0.98) 0.98
(0.93, 1.04) 0.73
(0.56, 0.97) 0.88
(0.80, 0.97) 0.73
(0.60, 0.88)
Missing 0 1
1Odds ratio expresses risk of adverse developmental/behavioural outcome per unit increase in exposure, adjusted for centre and congenital infection status.
2Odds ratio expresses risk of adverse developmental/behavioural outcome, adjusted for centre and congenital infection status
Table 3 Factors associated with parental concerns, specialist referral, impact of behaviour on the family, and parental anxiety, at 3 years, adjusted for centre and congenital infection status(Total N = 705)
Odds ratio for outcome (95% confidence interval)
EXPOSURES Any concerns5 Any referral Behaviour impact4 Parental anxiety3
Before birth
1) Congenital toxoplasmosis (reference = uninfected) 0.98 (0.95, 1.01) 2.00 (0.90, 4.45) 0.55 (0.22, 1.39) 3.01 (1.84, 4.94)
2) Maternal Age1 0.97 (0.92, 1.03) 0.96 (0.89, 1.05) 0.98 (0.90, 1.07) 0.98 (0.93, 1.03)
3) Parity1 0.96 (0.81, 1.15) 1.05 (0.72, 1.54) 0.98 (0.66, 1.44) 1.26 (1.06, 1.50)
4) Maternal Education1 0.91 (0.73, 1.14) 0.77 (0.51, 1.16) 1.18 (0.81, 1.74) 0.82 (0.65, 1.04)
5) Mother born outside country2 (no = reference) 1.59 (0.84, 3.01) 0.82 (0.18, 3.79) 2.20 (0.86, 5.60) 1.06 (0.51, 2.21)
6) Child's gender (male = reference) 1.89 (1.22, 2.93) 0.83 (0.39, 1.73) 0.99 (0.50, 1.96) 1.26 (0.81, 1.94)
7) Duration of prenatal treatment (wks) 0.98 (0.96, 1.00) 1.00 (0.96, 1.04) 1.01 (0.97, 1.05) 0.97 (0.94, 0.99)
After birth
8) Child's age at Questionnaire1 (month) 0.97 (0.90, 1.04) 1.07 (0.97, 1.19) 1.02 (0.93, 1.13) 0.97 (0.90, 1.04)
1Odds ratio expresses risk of adverse developmental/behavioural outcome per unit increase in exposure, adjusted for centre and congenital infection status
2Odds ratio expresses risk of adverse developmental/behavioural outcome, adjusted for centre and congenital infection status
3Adverse outcome defined by score corresponding most closely to the least able/most worried 10% in the uninfected children
4Adverse outcome defined by developers of SDQ questionnaire.
5Defined by answering 'yes' or 'a little' to question(s) about concerns.
Table 4 Factors associated with parent-reported impairment at 3 years adjusted for centre and congenital infection status (Total N = 705)
Odds ratio for adverse outcome (95% confidence interval)
EXPOSURES Any visual impairment3 Hearing loss4 Neurological/mobility impairment5
Before birth
1) Congenital toxoplasmosis 2.22 (1.10, 4.49) 0.61 (0.26, 1.46) 1.88 (0.32, 11.04)
2) Maternal Age1 0.99 (0.92, 1.07) 0.91 (0.92, 1.07) 1.15 (0.94, 1.42)
3) Parity1 1.08 (0.82, 1.43) 0.77 (0.84, 0.99) 1.63 (1.01, 2.63)
4) Maternal Education1 1.13 (0.77, 1.65) 0.96 (0.67, 1.37) 0.70 (0.27, 1.79)
5) Mother born outside country2 (no = reference) 0.35 (0.08, 1.53) 0.54 (0.16, 1.87) Not estimable
6) Child's gender (male = reference) 1.76 (0.88, 3.51) 1.27 (0.64, 2.50) 0.78 (0.15, 3.98)
7) Duration of prenatal treatment (wks) 0.95 (0.92, 0.99) 1.01 (0.97, 1.04) 0.96 (0.87, 1.06)
After birth
8) Child's age at Questionnaire1 (month) 1.02 (0.93, 1.11) 0.90 (0.79, 1.03) 1.05 (0.86, 1.29)
1Odds ratio for effect of exposure on outcome in all subjects per additional unit increase in exposure variable, adjusted for centre 2Odds ratio for effect of exposure on outcome, adjusted for centre 3Defined as any visual impairment, including wearing glasses 4Defined as any intermittent or permanent hearing loss 5Defined as seizures requiring treatment, cerebral palsy, and/or impaired
Information on clinical signs before four months of age was available for 571 children (excluding 23 infected and 111 uninfected children in Poznan). 131/155 (85%) infected children and 407/414 (98%) uninfected children had no clinical signs. In the remainder, clinical manifestations were detected before four months of age in order of increasing severity: lymphadenopathy or hepatosplenomegaly (3 infected, 2 uninfected); retinochoroiditis alone (6 infected, 0 uninfected); intracranial lesions, with or without retinochoroiditis (13 infected, 3 uninfected); and neurological impairment, with or without ocular or intracranial lesions and including seizures, microcephaly, microphthalmia, or abnormal neurological examination (2 infected, 4 uninfected).
Multivariable analyses
Development and behavior
There was no significant association between congenital infection status and abnormal development or behavior (Table 5). Similar results were obtained in the sensitivity analyses. In analyses confined to France, there was no evidence that an adverse effect of congenital infection status was masked by the duration of prenatal treatment, or that prenatal treatment had an adverse effect in uninfected children (results not shown).
Table 5 Association between congenital toxoplasmosis and developmental outcomes: multivariable analyses
Developmental outcome (Odds ratio and 95% confidence interval)
PARENT COMPLETED
Motor Speech & language Behaviour Cognition
All centres (N = 705) 1.30 (0.75, 2.25)2 1.10 (0.60, 2.02) 3,4 1.05 (0.66, 1.67) 2 0.85 (0.35, 2.09)
Sensitivity analyses
a) France only (N = 379)1 1.22 (0.61, 2.42)2 0.98 (0.45,2.14) 1.15 (0.64, 2.09) 0.55 (0.17, 1.76)2
b) Poland only (N = 134) 1.10 (0.30, 4.06) 0.54 (0.13, 2.26) 0.94 (0.30, 2.92) 1.02 (0.16, 6.62)
c) Children aged 36 to <41 months (N = 553) 1.18 (0.68, 2.05) 1.04 (0.54, 2.00) 1.25 (0.70, 2.22) 0.61 (0.25, 1.50)
d) Children with no signs before 4 months (N = 538 Poland excluded) 1.08 (0.57, 2.07) 1.10 (0.51, 2.36) 1.02 (0.57, 1.82)2 0.35 (0.10, 1.24)
CHILD COMPLETED
'Draw a Man' Line, circle, cross
All centres (N = 705) 1.27 (0.66, 2.44)5 1.13 (0.56, 2.25)
Sensitivity analyses
a) France only (N = 379)1 0.77 (0.35, 1.67) 0.93 (0.40, 2.18)
b) Poland only (N = 134) 1.00 (0.17, 5.77) 0.76 (0.13, 4.41)
c) Children aged 36 to <41 months (N = 553) 0.79 (0.39, 1.63)5 0.96 (0.48, 1.95)
d) Children with no signs before 4 months (N = 538 Poland excluded) 0.96 (0.46, 2.02) 1.03 (0.45, 2.36)
All models include centre, maternal education, and child's age at completion of questionnaire
1adjusts for variability among centres relative to Lyon
2adjusted for maternal age
3adjusted for born outside the country
4adjusted for maternal age
5adjusted for gender, and maternal age squared
Parental concerns, specialist referral, and parental anxiety
Significantly more anxiety was reported in parents of infected than uninfected children. Overall the risk of having a high anxiety score was more than doubled for parents of infected children in all centers. Not surprisingly, the risk was more than trebled in Poland, where parents of infected children were compared with parents from the general population who had not been identified by screening. The association between congenital infection status and parental anxiety was attenuated and no longer significant in analyses confined to France. None of these results changed appreciably when analyses were repeated using ordinal regression. There was no evidence that duration of prenatal treatment either masked an effect of congenital infection status on parental concerns, referrals, or anxiety, or had an adverse effect on these outcomes in uninfected children (results not shown, but available from authors).
Parent- reported impairment
The risk of visual impairment was doubled in infected vs uninfected children (p = .024). This was mainly due to an increased risk of limited vision, affecting 7 infected and 3 uninfected children. There was no significant difference in the proportion of children wearing glasses at 3 years (6.8% infected, 4.2% uninfected; p = 0.17). Hearing impairment was more common in uninfected children but this association was not significant. The risk of neurological or mobility problems was higher in infected children, but not significant at the 5% level as this outcome was reported for only 6 children.
Discussion
Congenital toxoplasmosis was associated with increased parental anxiety about the child's health now and in the future, and with an increased risk of visual impairment. The magnitude of the effect of congenital toxoplasmosis on parental anxiety was reduced in France. We found no evidence for an association between congenital toxoplasmosis and other markers of adverse development or behavior at three years.
These results concur with clinicians' experience, voiced during the design of the study, that parental anxiety was one of the most common problems encountered. An alternative interpretation, is that the association was a chance finding, arising because of the multiple comparisons performed. However, the consistency of the finding in sensitivity analyses, and with prior observation, make this explanation unlikely.
Other studies of children identified by screening have similarly reported that infected children appear to be developmentally normal, but no studies have systematically assessed development[2,17]. In contrast, the risk of neurological impairment is much higher in case series of referred children identified because of symptoms or abnormalities. For example, in the Chicago study, 13/45 (29%) children had an IQ/DQ score less than 85 (equivalent to one standard deviation below the mean) at 1 or 3 years old[4]. In a further case series reported by Wilson et al, 8/13 referred children had low (<50) or declining IQ/DQ scores at a mean age of 5.5 years[3].
A strength of the study is the minimization of selection bias due to referral of affected fetuses or children, by requiring that screening tests predated investigations for abnormality. As this study was prospective, data on clinical manifestations in infancy, could not be biased by the child's condition at 3 years. Conversely, we investigated whether parents that responded to the questionnaire were more or less likely to have a child with clinical manifestations detected before 4 months of age (defined as microphthalmia, microcephaly, seizures, abnormal or suspicious neurological examination requiring referral to a specialist, ventricular dilatation, or intracranial calcification) and found no significant association (reported elsewhere[7]). The association between neurological findings and abnormal developmental and impairment scores will be the subject of a further report. Nevertheless, a weakness of the study is the response rate, which was lower for uninfected than infected children.
Our analysis of reasons for non-response, found organizational attributes to be important determinants of response. Centers that provided follow up themselves, or had access to an address register, had better response rates and, apart from congenital infection status, we found no evidence of differential response rates according to other patient characteristics.[7] Uninfected children were more difficult to trace than infected children as they were discharged from follow up in late infancy, whereas infected children were followed indefinitely. However, we cannot exclude the possibility that response was associated with neurological problems more subtle than those detected in early infancy, and that the direction of this effect may have varied in infected and uninfected children.
A further weakness of the study is that we used only part of the standardized assessment tool, or a modified version of the tool, for all outcomes except behavior. This was done to enhance response by reducing the questionnaire to a reasonable length and average completion time of 21 minutes. Use of the full assessment tools, which would have required considerable time to complete, may have been more sensitive, but is unlikely to have been acceptable to parents.[18] A further compromise was to assess children at 3 years, rather than after school entry, when mild to moderate impairment may be more readily detected. Three years was decided because tracing addresses would have become increasingly difficult with age.
Our findings are consistent with the hypothesis that congenital toxoplasmosis either causes overt clinical damage, or no abnormality, rather than a spectrum of neurological impairment. However, further, more detailed assessments at older ages, including clinician assessment designed to detect mild or moderate impairment, are required to confirm or refute this hypothesis. An alternative interpretation is that the similarity in developmental outcomes in infected and uninfected children is due to anti-toxoplasma treatment. However, we found no evidence that the duration of prenatal treatment masked an adverse effect of congenital toxoplasmosis, nor that treatment caused adverse effects in uninfected children. We were unable to explore the effect of postnatal treatment in this analysis, although this will be investigated in a separate report confined to infected children.
We found an increased risk of parent-reported visual problems, which was largely due to reports of limited vision. This difference may be overestimated as parents of uninfected children were less likely to respond to this (overall response rate 92%, only 1 infected child had no response), and parents of infected children are likely to be more concerned about vision. For the seven congenitally infected children with limited vision, follow up by an ophthalmologist at 3 or more years found normal vision in 1 child, unilateral impaired vision in 5 children, and bilateral blindness in one child (results to be reported in detail elsewhere). Nevertheless, this finding concurs with a long term follow up study (median age 6 years) of 327 infected children in Lyon. None of 79 infected children with retinochoroiditis had bilateral visual impairment[19].
Generalisability of our findings should be made with caution. Our cohort was from a European setting where severely affected fetuses may be terminated. In our original cohort, there were 21 terminations: 17 were for toxoplasmosis, and 8 of these occurred after a positive prenatal diagnosis; 5 of the 8 underwent autopsy and 4 had intracranial lesions or signs of disseminated infection.[6] It is unlikely that, if these four babies had survived to three years, the results for developmental and behavioural outcomes would have been substantially altered. A further possibility is that congenital toxoplasmosis may be more severe in settings where the infecting organism load is high, as has been hypothesized for postnatal acquired infection[10]. For example, postnatal acquired ocular toxoplasmosis is both more common and more severe in Brazil, and may be associated with acquisition of infection from oocysts in unfiltered water[20]. Although a Brazilian study found a similar risk of intracranial or ocular lesions (17% of 47 infected children) to our European cohort (20%: 51/255 infected children; personal communication, R Gilbert on behalf of EMSCOT), the study was confined to women able to afford neonatal screening and may not be representative of the risk in less affluent communities[21].
Finally, the increased risk of anxiety may vary in different settings. Parents of infected children are told that eye lesions can appear at any age, and that such lesions may cause loss of vision. How this message is presented, may differ among clinicians. The effect of congenital infection status on parental anxiety was most marked in Poland, where the controls had not experienced screening, and where there is no nationwide screening program. In contrast, the risk of anxiety was lowest in France. This may be because parents are more familiar with congenital toxoplasmosis, and its limited clinical effects. Alternatively, it may be that anxiety persists in parents of uninfected children in France, thereby attenuating any difference. Marteau has reported long term effects of false positive results from other prenatal screening programs, highlighting persisting perceptions among parents that 'something must still be wrong'[9].
Conclusion
The purpose of prenatal and neonatal screening is to reduce the risk of functional impairment. Apart from visual impairment, we found no evidence for adverse effects of congenital toxoplasmosis on developmental outcomes, behavior, or specific impairments by three years of age. On average, infected children were no worse off, and there was no evidence that this was because they received prenatal treatment. However, we could not examine the possibility that postnatal treatment ameliorated any differences in function as all infected children were treated in infancy. In addition, we cannot rule out the possibility that we failed to detect more subtle differences that would become apparent later in childhood. Finally, clinicians need to explore ways for reducing the adverse effects of diagnosis and follow up on parental anxiety. One approach is to review whether they are being sufficiently optimistic when counseling parents about prognosis.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
*Members of the European Multicentre Study on Congenital Toxoplasmosis (EMSCOT)
Coordinating Committee: M-H Bessieres, W Buffolano, H Dumon, R Gilbert, E Petersen (chairperson), A Pollak, P Thulliez, M Wallon.
Writing Committee: Freeman K, Salt A, Prusa A, Malm G, Ferret N, Buffolano W, Petersen E, Gilbert RE (coordinator).
Centres contributing data (number of patients contributed to this report): M Paul (134; University Medical Sciences, Poznan), A Prusa, M Hayde, A Pollak (134; University Children's Hospital, Vienna), M Wallon, F Peyron (134; Hôpital de la Croix Rousse, Lyon), S Romand, P Thulliez (91; Institut de Puericulture, Paris), J Franck, H Dumon, P Bastien, E Issert (69; Hôpital de la Timone, Marseille; CHU de Montpellier), W Buffolano (50; Universita di Napoli, Naples), M-H Bessieres (44; Hôpital de Rangueil, Toulouse), N Ferret, P Marty (33; Hôpital de l'Archet, Nice), H Pelloux, H Fricker-Hidalgo, C Bost-Bru (8; Centre Hospitalier Universitaire de Grenoble), G Malm, B Evengard (8; Huddinge Hospital, Stockholm), E Petersen (0; Statenseruminstitut, Copenhagen), C Chemla, (0; Hôpital Maison Blanche, Reims), E Semprini, V Savasi (0; Milan).
Study design and coordination: R Gilbert (Principal Investigator), L Gras, Hooi Kuan Tan, J Rickett, A Salt, L Valenti (Institute of Child Health, London)
Statistical analysis: Freeman K, Gras L (Institute of Child Health, London).
AS developed the questionnaires, participated in analyses and wrote the paper. KF did the statistical analyses and wrote the paper. RG had the idea for the study, obtained the funding, designed the study, developed the questionnaires, coordinated the study, and wrote the paper. All authors contributed to the design of the study and the writing of the paper. All authors read and approved the final manuscript.
Table 6 Association between congenital toxoplasmosis and parental concerns, specialist referral, and anxiety: multivariable analyses
Parental concerns and anxiety (odds ratio and 95% confidence interval)
Any concerns Any referral Behaviour impact Parental anxiety
All centres (N = 705) 1.22 (0.65, 2.29)1 0.79 (0.25, 2.55)2 0.57 (0.23, 1.43) 2.59 (1.41, 4.79)3
Sensitivity analyses
a) France only (N = 379) 1.41 (0.73, 2.73) 0.25 (0.03, 2.09) 0.40 (0.11, 1.46) 1.63 (0.80, 3.34)
b) Poland only (N = 134) 1.06 (0.36, 3.05) 3.97 (0.90, 17.59) 0.55 (0.06, 5.40) 3.50 (1.24, 9.84)
c) Children aged 36 to <41 months (N = 553) 1.27 (0.73, 2.22) 1.50 (0.54, 4.13) 0.44 (0.14, 1.34) 2.63 (1.47, 4.68)
d) Children with no signs before 4 months (N = 538-Poland excluded) 1.00 (0.98, 1.02) 0.46 (0.10, 2.16) 0.39 (0.11, 1.38) 2.17 (1.14, 4.15)
All models include country, maternal education, and child's age at completion of questionnaire
1adjusted for gender and prenatal treatment
2adjusted for interaction between congenital infection status and Poland
3adjusted for parity
4adjusted for gender
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The research was part of the European multicentre study on congenital toxoplasmosis, funded by the European Commission (BIOMED II No. BMH4-CT98-3927 and QLG5-CT-2000-00846). Additional support has been provided by the National Eye Institute, grant code R03 EY015287-01 and the "Verein unser_kind, Verein zur Durchfuehrung der wissenschaftlichen Forschung auf dem Gebiet der Neonatologie und Kinderintensivmedizin" in Vienna.
==== Refs
Gilbert RE Peckham CS Congenital toxoplasmosis in the United Kingdom: to screen or not to screen? J Med Screen 2002 9 135 141 12370327 10.1136/jms.9.3.135
Koppe JG Loewer Sieger DH de Roever Bonnet H Results of 20-year follow-up of congenital toxoplasmosis Lancet 1986 1 254 256 2868264 10.1016/S0140-6736(86)90785-3
Wilson CB Remington JS Stagno S Reynolds DW Development of adverse sequelae in children born with subclinical congenital Toxoplasma infection Pediatrics 1980 66 767 774 7432882
Roizen N Swisher CN Stein MA Hopkins J Boyer KM Holfels E Mets MB Stein L Patel D Meier P al Neurologic and developmental outcome in treated congenital toxoplasmosis Pediatrics 1995 95 11 20 7770286
Paul M Petersen E Pawlowski ZS Szczapa J Neonatal screening for congenital toxoplasmosis in the Poznan region of Poland by analysis of Toxoplasma gondii-specific IgM antibodies eluted from filter paper blood spots Pediatr Infect Dis J 2000 19 30 36 10643847 10.1097/00006454-200001000-00007
Gilbert R Gras L Effect of timing and type of treatment on the risk of mother to child transmission of Toxoplasma gondii BJOG 2003 110 112 120 12618153
Salt A Freeman K Prusa A Ferret N Buffolano W Malm G Petersen E Gilbert RE Determinants of response to a parent questionnaire about development and behavior in 3 year olds: European multicenter study of congenital toxoplasmosis. BMC Pediat 2005 5 21 15998464 10.1186/1471-2431-5-21
Harris B Dale and Harris B Requirements for scoring drawings, item by item. Children's drawings as measures of intellectual maturity 1963 1st Harcourt, Brace and World, Inc 246 263
Marteau TM Cook R Kidd J Michie S Johnston M Slack J Shaw R The psychological effects of false positive results in prenatal screening for fetal abnormality: a prospective study Prenat Diagn 1992 12 205 214 1375378
Holland GN Ocular toxoplasmosis: a global reassessment. Part II: disease manifestations and management Am J Ophthalmol 2004 137 1 17 14700638 10.1016/j.ajo.2003.10.032
Glascoe FP Parents' evaluation of developmental status: how well do parents' concerns identify children with behavioral and emotional problems? Clin Pediatr (Phila) 2003 42 133 138 12659386
Goodman R Ford T Simmons H Gatward R Meltzer H Using the Strengths and Difficulties Questionnaire (SDQ) to screen for child psychiatric disorders in a community sample Br J Psychiatry 2000 177 534 539 11102329 10.1192/bjp.177.6.534
Schafer JL Analysis of Incomplete Multivariate Data Monographs on Statistics and Applied Probability 2004 London, Chapman & Hall
Hosmer DW Lemeshow SI Hosmer DW and Lemeshow SI Model building strategies and methods for logistic regression Applied logistic regression 2000 New York, John Wiley and Sons 82 89
Agresti A Agresti A Categorical data analysis 2002 New York, John Wiley and Sons
Freeman K Oakley L Pollak A Buffolano W Petersen E Semprini AE Salt A Gilbert RE Association between congenital toxoplasmosis and preterm birth, low birth weight, and small for gestational age birth BJOG 2005 112 31 37 15663394
Berrebi A Kobuch WE Bessieres MH Bloom MC Rolland M Sarramon MF Roques C Fournie A Termination of pregnancy for maternal toxoplasmosis Lancet 1994 344 36 39 7912304 10.1016/S0140-6736(94)91054-5
Edwards P Roberts I Clarke M DiGuiseppi C Pratap S Wentz R Kwan I Increasing response rates to postal questionnaires: systematic review BMJ 2002 324 1183 12016181 10.1136/bmj.324.7347.1183
Binquet C Wallon M Quantin C Kodjikian L Garweg J Fleury J Peyron F Abrahamowicz M Prognostic factors for the long-term development of ocular lesions in 327 children with congenital toxoplasmosis Epidemiol Infect 2003 131 1157 1168 14959784 10.1017/S0950268803001316
Bahia-Oliveira LM Jones JL Azevedo-Silva J Alves CC Orefice F Addiss DG Highly endemic, waterborne toxoplasmosis in north rio de janeiro state, Brazil Emerging Infect Dis 2003 9 55 62 12533282
Neto EC Anele E Rubim R Brites A Schulte J Becker D Tuuminen T High prevalence of congenital toxoplasmosis in Brazil estimated in a 3-year prospective neonatal screening study Int J Epidemiol 2000 29 941 947 11034982 10.1093/ije/29.5.941
EMSCOT website
|
16014166
|
PMC1199601
|
CC BY
|
2021-01-04 16:31:06
|
no
|
BMC Pediatr. 2005 Jul 13; 5:23
|
utf-8
|
BMC Pediatr
| 2,005 |
10.1186/1471-2431-5-23
|
oa_comm
|
==== Front
BMC PediatrBMC Pediatrics1471-2431BioMed Central London 1471-2431-5-261603364210.1186/1471-2431-5-26Research ArticleCatch-up growth up to ten years of age in children born very preterm or with very low birth weight Knops Noël BB [email protected] Kommer CA [email protected] Ronald [email protected] Elysee TM [email protected] Ouden A Lya [email protected] Jan-Maarten [email protected] S Pauline [email protected] Department of Paediatrics; Leiden University Medical Center, J6S; P.O. Box 2600; 2300 RC Leiden; The Netherlands2 TNO Prevention and Health, Child Health Division, Leiden; The Netherlands3 Department of Medical Statistics, Leiden University Medical Center; The Netherlands2005 20 7 2005 5 26 26 22 1 2005 20 7 2005 Copyright © 2005 Knops et al; licensee BioMed Central Ltd.2005Knops et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Improved survival due to advances in neonatal care has brought issues such as postnatal growth and development more to the focus of our attention. Most studies report stunting in children born very preterm and/or small for gestational age. In this article we study the growth pattern of these children and aim to identify factors associated with postnatal catch-up growth.
Methods
1338 children born with a gestational age <32 weeks and/or a birth weight of <1500 grams were followed during a Dutch nationwide prospective study (POPS). Subgroups were classified as appropriate for gestational age and <32 weeks (AGA) or small for gestational age (<32 wks SGA and ≥32 wks SGA). Data were collected at different intervals from birth until 10 years for the 962 survivors and compared to reference values. The correlation between several factors and growth was analysed.
Results
At 10 years the AGA children had attained normal height, whereas the SGA group demonstrated stunting, even after correction for target height (AGA: 0.0 SDS; SGA <32 wks: -0.29SDS and ≥32 wks: -0.13SDS). Catch-up growth was especially seen in the SGA children with a fast initial weight gain. BMI was approximately 1 SD below the population reference mean.
Conclusion
At 10 years of age, children born very preterm AGA show no stunting. However, many children born SGA, especially the very preterm, show persistent stunting. Early weight gain seems an important prognostic factor in predicting childhood growth.
==== Body
Background
Advances in neonatal care in the past two decades have improved survival of very preterm and low birth weight infants dramatically. This has brought issues such as developmental outcome and growth of these 'survivors' to the focus of our attention [1]. In 1989 Barker et al demonstrated an increase in mortality from coronary heart disease in adulthood in subjects who had low birth weight [2], and this observation was followed by studies of many others on long term adverse effects of intra-uterine growth retardation [3]. Furthermore not only the appropriateness of weight for gestational age but the course of postnatal growth seems to predict later neurodevelopmental outcome in children with low birthweight [4]. Follow-up of these children is an important tool in learning to understand and tackling problems that can occur.
Several studies have studied physical growth during early childhood in children born preterm, small for gestational age (SGA) or appropriate for gestational age (AGA). They show stunting for all groups with limited to no catch-up growth especially in SGA group [5-11]. This could not be explained by factors known to influence postnatal growth such as: hospitalisation, interventional status, bronchopulmonary dysplasia, socio-economic status, parental level of education, neonatal thyroxine or thyroxine stimulating hormone levels [8,12].
Most data reporting on further growth into adolescence show incomplete catch up growth and therefore persistent stunting, especially in those born SGA [13-16]. Those who report attainment of normal height studied mainly children born AGA [17,18]. Only the female SGA cohort followed by Hack et al reached normal adult height but their male counterparts did not [19]. Thus stunting is primarily seen in the infant with low birth weight for gestational age. These findings suggest a long-term relationship between intrauterine growth retardation and growth into adolescence. Determining the factors that cause growth impairment in these children may enable us to find therapeutic or preventive options.
Following an earlier report on height achievement at five years of a large cohort of children born very preterm or VLBW [12], we studied the growth pattern up to 10 years of age in the same children. The data were collected in a nationwide effort of Dutch pediatricians: the " Project On Preterm and Small for Gestational Age Infants in The Netherlands" (POPS) [20]. We aimed to identify factors associated with catch-up growth between 5 and 10 years of age. Predictive values and sensitivity rates of short stature at younger age for having short stature at 10 years of age were determined. Finally, Body Mass Index was calculated to assess nutritional status in our study population.
Methods
The POPS study was started in 1983 to investigate relationships between perinatal factors, mortality and morbidity in very preterm and VLBW infants. The protocol was approved by the medical ethics committees of the participating hospitals and parents gave informed consent. The study population consisted of live-born infants in The Netherlands from January 1 to December 31, 1983, with a gestational age of <32 weeks and / or a birth weight of <1500 g. Follow-up examinations were performed at the approximate corrected ages of: 3, 6, 12 and 24 months and again at the approximate chronological age of 5 and 10 years. Of the original 1338 children, 962 were alive at 10 years of age.
Paediatricians performed length/height measurements at 3 and 6 months and at 1, 2 and 5 years. At 10 years, the parents reported height as part of an extensive questionnaire. Recent growth curves were used as reference values for body length/height measurements and body mass index [21,22]. These measurements were collected in 1996–1997 during a nationwide growth study on individuals of Dutch origin aged 0 to 21 years. Based on these reference values, for each child the standard deviation score (SDS) was calculated using corrected age (height minus age/sex-specific mean height in the reference population divided by age/sex-specific SD in the reference population). A child was considered having "short stature" at a specific age if his/her height was below the 10th percentile (being equivalent to approximately -1.3 SDS) of the reference population.
Since in our earlier report [12] reference values were obtained from the Dutch growth curves of 1980 [23], the SD-scores employed in the current study were also compared to those based on the 1980 reference data. Only marginal differences were observed, with the SD-scores based on the 1996–1997 reference values being on average 0.04 SD lower than those based on the 1980 reference data (range -0.37 to 0.34).
Classification of AGA/SGA (appropriate/small-for-gestational-age) infants was based on measurements of birth weight relative to gestational age, described by Kloosterman [24,25]. An SGA infant has a birth weight less than the 10th percentile for gestational age, gender and parity.
For the present study, non-Caucasian children and children with congenital malformations were excluded, since the reference data cannot be used for these children. Hence, 753 of the 962 surviving children were included in the analysis.
Statistical analyses included the following steps. The growth pattern of the children over time was examined by comparing mean length/height SDS, median percentiles and percentages of children below the 10th percentile (<P10) at the various ages. Considering possible bias due to differences in genetic growth potential, target height (TH) was calculated for each individual using parental height data compared to Dutch national growth statistics. Target height was defined as the mean of parental height corrected for the mean difference in height between sexes, plus the mean increase in height per generation (according to Dutch reference values [23]:((height father + height mother plus or minus 12 cm) divided by 2 + 3 cm). Using individual target height, target height SD-scores (TH-SDS) and height SDS corrected for TH (HSDScorr) were assessed (TH-SDS: (TH minus mean adult height (male or female)) divided by the sex-specific standard deviation for adult height) and HSDScorr: height-SDS minus TH-SDS). Separate analysis of growth patterns was performed among three subgroups of children: very preterm but appropriate-for-gestational age (<32 wks/AGA), very preterm and small-for-gestational age (<32 wks/SGA) and gestational age of 32 weeks or over who were small-for-gestational age (≥32 wks/SGA). The clinical characteristics of these subgroups are shown in table 1. The relationship between several factors and catch-up growth between 5 and 10 years of age was explored by means of regression analysis. Differences between the 5- and 10-years height SDS were calculated and compared between the three defined subgroups, as well as for single/multiple birth, sex, socio-economic status, maternal and paternal height. Cross-tabulations were used to examine the predictive values and sensitivity rates of height <P10 (and <P3) at various ages for height <P10 (and <P3) at 10 years of age. Follow-up data on weight were used to study the effect of early postnatal growth on height at 10 years and to assess BMI.
Table 1 Clinical characteristics of the different subgroups according to gestational age and weight of the entire study population, of the group lost to follow up at age10 years and of the children with available data on weight increase during the first three months.
Sex Mean Gestational Age Mean Birth-weight Multiple Birth Respiratory support >1 week Intra Cranial Haemorrhage Necrotising Entero Colitis Mean Hospital stay
n (% males) (wk) (gram) (%) (%) (%) (%) days
<32 wks/AGA 445 53.1 29.7 1412 26.7 27.8 21.7 4.5 64
<32 wks/SGA 86 51.2 30.3 968 11.6 23.3 14.0 8.1 86
≥32 wks/SGA 222 49.5 34.5 1278 16.7 2.5 5.1 5.6 59
Lost to follow up at age 10 years
243 49.2 31.1 1326 23.0 18.4 18.3 5.8 63
Weight Increase at 3 months
<3500 gr 257 45.3 32.1 1275 21.7 20.6 14.9 6.2 69
≥3500 gr 389 54.6 30.1 1349 21.2 19.3 16.3 4.6 62
Results
Of the 753 children in the POPS cohort who met our selection criteria, length/height measurements were available for 649 to 721 children at various ages up to 5 years of age (86% to 96%). Table 1 shows the clinical characteristics of the survivors according to subgroup. At 10 years, height data were obtained for 510 children (68%).
Table 2 shows the mean length/height data expressed as SDS, median percentiles and proportions of children <P10 for the total and subgroups. In the total group there was a substantial increase in mean SDS from -1.28 at 3 months to -0.18 at 10 years of age, similarly reflected in an increase in median percentile and decrease in percentage of children <P10. Both the median percentile and the percentage <P10 at 10 years are similar to those at 5 years of age. The 17% of children <P10 at 10 years of age were on average 13 cm shorter (range 8 to 25 cm) than sex/age-specific reference values (131 cm versus 144 cm).
Table 2 Growth measurements uncorrected and corrected for target height (HSDScorr) at various ages for subgroups based on weight for gestational age
Length/height-SDS Percentiles HSDScorr
Approximate age n mean <P10 n SDS
Total group
3 months 659 -1.28 48% 633 -1.11
6 months 649 -0.95 39% 626 -0.76
1 year 670 -0.76 32% 649 -0.58
2 years 651 -0.64 28% 628 -0.44
5 years 721 -0.18 18% 719 0.03
10 years (observed)1 510 -0.18 17% 509 -0.07
10 years (estimated)2 723 -0.24 17% 721 -0.09
<32 wks/AGA
3 months 388 -0.68 29% 381 -0.58
6 months 383 -0.49 24% 382 -0.39
1 year 387 -0.43 22% 386 -0.34
2 years 378 -0.33 20% 374 -0.21
5 years 422 0.11 7% 420 0.18
10 years (observed)1 298 -0.04 15% 298 0.00
10 years (estimated)2 423 -0.06 15% 422 0.01
<32 wks/SGA
3 months 78 -2.22 82% 78 -1.90
6 months 72 -1.72 64% 72 -1.39
1 year 78 -1.38 46% 78 -1.08
2 years 78 -1.33 45% 78 -1.02
5 years 85 -0.65 24% 85 -0.37
10 years (observed)1 64 -0.55 25% 64 -0.29
10 years (estimated)2 85 -0.64 24% 85 -0.33
≥32 wks/SGA
3 months 193 -2.19 78% 174 -1.90
6 months 194 -1.60 72% 172 -1.31
1 year 205 -1.16 48% 185 -0.87
2 years 195 -0.95 36% 176 -0.67
5 years 214 -0.62 25% 214 -0.35
10 years (observed)1 148 -0.35 17% 147 -0.13
10 years (estimated)2 215 -0.48 16% 214 -0.20
1 observed length for availabe cases
2 observed length for availabe cases + estimated length (based on length at 5 years) for missing cases
To examine whether the missing data pattern at 10 years was random or not, mean SD-scores at 5 years (n = 721) were compared between the 508 children for whom height measurements were available at both 5 and 10 years and the 213 children without a 10-years height measurement. A statistically significant difference was noted (p = 0.02), with mean SD-scores of -0.18 and -0.63 for those with and without 10-years data, respectively. To adjust for possible bias due to drop-out of shorter children, SD-scores for the 213 children without height measurements at 10 years but for whom height at 5 years was available were estimated by means of a linear regression equation. This equation was based on regression of 10-years SDS on those at 5 years in the subgroup of 508 children with both measurements available: 10-years SDS = 0.006 + (0.832 × 5-years SDS); explained variance R2 = 0.58. As shown in Table 2, this calculation yielded a mean height SDS of -0.24 (instead of -0.18 for available measurements).
For the <32 wks/AGA children moderate stunting was noted with catch-up growth up to 5 years of age but none between age 5 and 10 years. For both the <32 wks/SGA and the ≥32 wks/SGA children, more serious stunting was noted with continuing catch-up growth up to 10 years.
The <32 wks/AGA group had a normal target height (-0.01 SDS). However, both subgroups born SGA had lower target height of -0.29 SDS. Therefore, in the SGA group length/height-SDS corrected for TH (HSDScorr) is substantially higher than uncorrected length/height-SDS (mean difference HSDScorr - length/height-SDS: + 0.30 SDS in the <32 weeks and +0.27 SDS in the ≥32 weeks group). Correction in the <32 wks/AGA group was small (mean difference: +0.10 SDS).
The relationship between several other factors and catch-up growth between 5 and 10 years of age was examined among those children for whom height measurements were available at both ages. The change between the 5- and 10-years SD-scores for height was on average 0.07 SDS, a large range was noted at the individual level (-2.96 to 3.74). Although parental height and, to a smaller extent, multiple birth and socio-economic status were significantly associated with height at 5 and 10 years, these factors were not associated with catch-up growth in that period (respectively p = 0.14 for multiple birth and p = 0.3 for socio-economic status). Regression analysis showed that, in addition to the observed differences in catch-up growth between the three subgroups (p = 0.001), the level of catch-up growth was independently associated with sex (p = 0.03). Table 3 shows that more catch-up growth was observed for boys than for girls. Regarding the different subgroups, more substantial catch-up growth was observed for the SGA boys, especially for those of ≥32 weeks (mean SD-score change of 0.37). Only marginal changes in 5- and 10-years mean SD-scores were seen for girls in all three subgroups, as well as for the <32 wks/AGA boys. In these children, the percentage of children <P10 at 10 years of age were even slightly higher than at 5 years of age. The highest percentage <P10 at 10 years of age was observed for <32 wks/SGA boys, in spite of a significant decrease in the last 5 years (from 49% to 36%).
Table 3 Comparison of growth patterns in boys and girls in the different subgroups
Mean length/height SD-scores Proportions < P10
n 5 yr 10 yr change 5 yr 10 yr
<32 wks/AGA
girls 133 -0.05 -0.15 -0.10 9% 14%
boys 164 -0.24 -0.21 0.03 17% 20%
<32 wks/SGA
girls 31 -0.57 -0.50 0.07 19% 26%
boys 33 -1.14 -0.89 0.25 49% 36%
≥32 wks/SGA
girls 66 -0.69 -0.52 0.17 21% 24%
boys 63 -0.72 -0.35 0.37 25% 19%
The predictive value of height <P10 at various ages for height <P10 at 10 years was examined by cross-tabulation of available measurements. The predictive value at 5 years of age was 69% (Table 4). Lowering the cut-off point from P10 to P3, a stricter criterion for "short stature", did not increase the predictive value.
Table 4 Relation between short stature at age 5 and 10 years
n At 10 years
<P10 ≥P10
At 5 years
<P10 95 66 (69%) 29 (31%)*
≥P10 413 37 (9%) 376 (91%)
<P3 ≥P3
At 5 years
<P3 44 29 (66%) 15 (34%)†
≥P3 464 18 (4%) 446 (96%)
* median percentile: 23 (16 children < P25)
† median percentile: 14 (12 children < P25)
To examine the effect of early postnatal growth on height at 10 years we looked at the average increase in weight during the first three months, approximately 3500 grams, and used this as a cut-off point defining two subgroups (fast and slow initial growth rate). Clinical characteristics and HSDScorr for these groups are given in table 1 and 5, respectively. Children with fast initial growth already attained normal height for TH SDS at 5 years, while those with slow initial growth showed persistent stunting. The fast subgroup of <32 wks/AGA attained full catch-up growth at 2 years of age. The fast subgroup of the ≥32/SGA weeks did so at age 5 year. All slow equivalents show persistent stunting. Despite initial high catch-up growth in the fast subgroup of the very preterm SGA they remain stunted at age 10 years.
Table 5 Height corrected for target height (HSDScorr) in children with fast and slow initial growth for the different subgroups based on weight for gestational age.
HSDScorr
Weight increase at 3 months
<3500 gram ≥3500 gram
Approximate age n mean n mean
Total group
3 months 257 -1.85 389 -0.59
6 months 252 -1.41 371 -0.32
1 year 263 -0.99 370 -0.27
2 years 243 -0.85 367 -0.18
5 years 271 -0.61 391 -0.08
10 years (observed) 189 -0.37 282 -0.08
10 years (estimated) 270 -0.42 392 -0.08
<32 wks/AGA
3 months 100 -1.29 278 -0.32
6 months 101 -1.05 263 -0.14
1 year 104 -0.75 262 -0.19
2 years 98 -0.65 258 -0.08
5 years 108 -0.41 278 -0.04
10 years (observed) 77 -0.35 199 -0.07
10 years (estimated) 109 -0.33 279 -0.07
<32 wks/SGA
3 months 37 -2.47 41 -1.38
6 months 34 -2.03 37 -0.82
1 year 38 -1.36 38 -0.81
2 years 38 -1.34 38 -0.64
5 years 41 -0.91 41 -0.37
10 years (observed) 29 -0.46 32 -0.37
10 years (estimated) 41 -0.60 41 -0.33
≥32 wks/SGA
3 months 120 -2.20 70 -1.30
6 months 117 -1.57 71 -0.78
1 year 121 -1.11 70 -0.31
2 years 107 -0.85 71 -0.28
5 years 122 -0.73 72 -0.06
10 years (observed) 83 -0.38 51 -0.06
10 years (estimated) 120 -0.44 72 -0.04
Body Mass Index (BMI) and BMI-SDS for the total group and different subgroups was calculated. BMI for all the children was approximately 1 kg/m2 below Dutch reference values [22]. However, the overall trend is similar to Dutch reference values. Mean BMI-SDS in the total group was -0.73 at 10 years. The SGA group had the lowest BMI (mean BMI-SDS: -1.02 and -0.96 for the <32 weeks and ≥32 weeks, respectively). The <32 wks/AGA had the highest BMI (-0.56 SDS). BMI-SDS was lower in early childhood in both the SGA and AGA group.
Discussion
Our results demonstrate, despite catch-up growth, persistent stunting until the age of 10 year in children born SGA, both before and after 32 weeks. The <32 wks/AGA showed little to no stunting and no catch-up growth from 5 to 10 yr. The improvement of SDS does not necessarily implicate a decrease in the absolute difference of a subject's height versus the average for age so that catch-up growth can remain unnoticed for the families. We are aware that the reporting by the parents of their child's height at 10 yrs could be imprecise, however this will be the same for all groups and have therefore no effect on the final conclusion of this study. Correction for Target Height suggests that the low height SDS in SGA children is partially associated with genetic factors. Failing to correct for target height leads to an overestimating of stunting in especially the SGA children (±0.3 SDS). This point was also made by Ford et al [26]. These data show that intrauterine growth retardation has more important long-term impact on growth than gestational age, suggesting an intrinsic lesser growth potential in children born small for date or a persistent effect of growth retardation in utero. A study by Sung et al showed more stunting in the SGA at 4 yr. compared to weight matched children born AGA (mean gestational age: SGA: 29 wk; AGA 26 wk) although the latter had more complications in early infancy [10]. Peralta-Carcelen et al demonstrated persistent stunting in adolescence in extremely low birth weight children without major handicaps [16]. Also in our study there was no significant difference in the incidence of neonatal morbidity between the subgroups (data published earlier [27]). Further analysis of the extent of each complication as stated in table 1 was not performed in this study but could be important for growth in individual cases (for example gut resection after NEC). Thus, intrauterine growth retardation seems a more important factor in determining childhood growth than neonatal complications associated with long-term physical and mental impairments.
Early recognition of persistent stunting can be of value for better treatment and preventive measures. In our study the predictive value of height <P10 at various ages for height at 10 years increased with age until 69% at 5 years. Lowering the cut-off point to <P3 did not improve the predictive value much, probably because of continuing catch-up growth. If in the future these cut-off points are used as a criterion to start growth enhancing treatment at an age of 5 years, the results of treatment should be compared with these figures on spontaneous growth.
Early postnatal (catch-up) growth possibly provides a useful tool for predicting height at 10 years. Because of difficulty in the accurate measurement of length at birth we did not register length at birth in our study and expressed early catch-up growth in weight gain. The arbitrary cut-off point we chose of the mean weight increase in the first 3 months (approximately 3500 grams) could result in wrongfully selecting children to a group with a different growth potential compared to their own. This would result in pulling height outcome of the different groups toward each other (regression to the mean). However our data show a clear difference between the subgroups, supporting the possibility of an inherently different growth potential. Children with an initial high catch-up growth generally attain normal height but the slow starters and the very preterm SGA do not. A similar observation was made by others [28,29]. Fewtrell et al suggested a distinction in children with an intrinsic or extrinsic disturbed intra-uterine growth potential after the observation that preterm children whose mothers had hypertension or toxemia showed more catch-up growth and less stunting at 12 year compared to those whose mothers were unaffected [30].
Variation in early catch-up could be caused by a disturbance in hormonal response. Albertsson-Wikland et al demonstrated low growth hormone, IGF-I, IGFBP-3 and leptin secretion in the subgroup of term SGA children with little catch-up growth at two months and short final height versus term SGA children with good catch-up and no stunting [31]. This could be due to rare IGF-I polymorphisms more frequently seen in the SGA and is associated with low bone mineral density [32]. This again could be a possible explanation for the high alkaline phosphatase levels in especially SGA preterm infants who remain short at age 12 yr, demonstrated by Fewtrell et al who suggested early metabolic bone disease as a cause for stunting in these children [30].
Preterm boys demonstrated more stunting at age 5 and 10 yr. but also showed more catch-up growth than girls. Only the ≥32 wks/SGA boys demonstrated less stunting than the girls at age 10 years. Other studies also demonstrated more stunting in SGA boys [19,33]. Possible explanations for the higher catch-up growth in boys in our study include a higher degree of stunting compared to the girls (which leaves them more to catch-up), a higher morbidity in the neonatal period and later life, and a different time of onset of puberty. Girls born SGA were shown to have an exaggerated adrenarche by Ibanez et al [34,35] but no difference has been demonstrated in the onset of puberty between very preterm SGA and a control population [14,26,36]. Nonetheless an advanced bone age has been reported despite similar sexual maturation in adolescents of very preterm SGA origin [16,37]. Advanced bone age could further compromise final height in these children. Unfortunately, we have no data on sexual maturity in our population.
Our study population demonstrated a low mean BMI compared to their peers. However at age 10 years there was a slight increase in BMI-SDS in relation to the previous ages. Although BMI-SDS is lowest in the SGA, a similar trend is observed in all groups. Saigal et al [14] demonstrated a 1.9 SD lower BMI in their very preterm/ SGA group vs. controls at age 8. At adolescence there remained a difference of -1.5 SD in BMI. A cohort study from the United States performed by Hack et al showed eventual normal adult BMI in SGA girls (0.4 SDS) but not in boys (-0.3 SDS) [19]. Hediger et al demonstrated lower mid-upper arm circumference in SGA children (-0.5 SDS) [38]. These data suggest a diminished nutritional status again especially in the SGA group when compared to their peers. This does not necessarily implicate a higher incidence of wasting in the study population but could also be explained by the increased incidence of obesity in the Dutch children and higher BMI reference values of 1997 compared to 1980.
Conclusion
In this study we show that children born very preterm with an appropriate weight for gestational age show little or no stunting at the age of 10 yr. However children born small for gestational age, especially those born <32 weeks, show persistent stunting at age 10 yr notwithstanding considerable catch-up growth in many of them. Early weight gain seems an important factor in predicting catch-up growth. Knowledge about spontaneous growth is useful if one considers offering growth-enhancing treatment to low birth weight children with short stature in early childhood.
List of abbreviations
SGA Small for Gestational Age
AGA Appropriate for Gestational Age
SDS Standard Deviation Score
TH Target Height
HSDScorr Height corrected for Target Height
BMI Body Mass Index
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
NK and KS were involved in the design of the analysis and in writing the manuscript. RB performed the statistical analyses. EH and LdO provided the data of the POPS cohort. JMW supervised the design and writing process. SVV is projectleader of POPS. All authors contributed to the preparation of the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We thank all the children and parents for their friendly cooperation and all the contributing pediatricians for their kind help in the data collection.
==== Refs
Hack M Flannery D Schluchter M Cartar L Borawski E Klein N Outcomes in young adulthood for very-low-birth-weight infants NEJM 2002 346 149 157 11796848 10.1056/NEJMoa010856
Barker D Osmond C Winter P Margretts B Simmonds S Weight in infancy and death from ischaemic heart disease Lancet 1989 2 577 580 2570282 10.1016/S0140-6736(89)90710-1
Jaquet D Czernichow P Born Small for Gestational Age: Increased Risk of Type 2 Diabetes, Hypertension and Hyperlipidaemia in Adulthood Hormone Research 2003 59 131 137 12638526 10.1159/000067848
Latal-Hajnal B von Siebenthal K Kovari H Bucher H Largo R Postnatal Growth in VLBW Infants: Significant Association with Neurodevelopmental Outcome J Pediatr 2003 143 163 170 12970627
Sann L Darre E Lasne Y Bourgeois J Bethenod M Effects on prematurity and dysmaturity on growth at age 5 years J Pediatr 1986 109 681 686 3761087
Casey P Kraemer H Bernbaum J Yogman M Sells J Growth status and growth rates of a varied sample of low birth weight, preterm infants: A longitudinal cohort from birth to three years of age J Pediatr 1991 119 599 605 1919893
Casey P Kraemer H Bernbaum J Tyson J Sells J Yogman M Bauer C Growth patterns of low birth weight preterm infants: A longitudinal analysis of a large, varied sample J Pediatr 1990 117 298 307 2199644
Strauss R Dietz W Effects of intrauterine growth retardation in premature infants on early childhood growth J Pediatr 1997 130 95 102 9003857
Barros F Huttly S Victoria C Kirkwood B Vaughan J Comparison of the causes and consequences of prematurity and intrauterine growth retardation: A longitudinal study in southern Brazil Pediatrics 1992 90 238 244 1641289
Sung I Vohr B Oh W Growth and neurodevelopmental outcome of very low birth weight infants with intrauterine growth retardation: Comparison with control subjects matched by birthweight and gestational age J Pediatr 1993 123 618 624 7692029
Gutbrod T Wolke D Soehne B Ohrt B Riegel K Effects of gestation and birth weight on the growth and development of very low birthweight small for gestational age infants: a matched group comparison Arch Dis Child (Fetal Neonatal Ed) 2000 82 F208 214 10794788 10.1136/fn.82.3.F208
Qvigstad E Verloove-Vanhorick S Ens-Dokkum M Schreuder A Veen S Brand R Oostdijk W Ruys J Prediction of height achievement at five years of age in children born very preterm or with very low birth weight: continuation of catch-up growth after two years of age Acta Paediatr 1993 82 444 448 8518520
Rantakallio P A 14 year follow-up of children with normal and abnormal birth weight for their gestational age. A popular study Acta Paediatr Scand 1985 74 62 69 3984729
Saigal S Stoskopf B Burrows A Rosenbaum P Physical growth and current health status of extremely low weight infants and controls in adolescence (Abstract) Pediatrics 2001 108 407 415 11483807 10.1542/peds.108.2.407
Ericson A Kallen B Very low birthweight boys at the age of 19 Arch Dis Child (Fetal Neonatal Ed) 1998 78 F171 F174 9713026
Peralta-Carcelen M Jackson D Goran M Royal S Mayo M Nelson K Growth of adolescents who were born at extremely low birth weight without major disability J Pediatr 2000 136 633 640 10802496 10.1067/mpd.2000.104291
Hirata T Bosque E When they grow up: The growth of extremely low birth weight (1000 gm) infants at adolescence J Pediatr 1998 132 1033 1035 9627599
Niklasson A Engstrom E Hard A Albertsson-Wikland K Hellstrom A Growth in Very Preterm Children: A Longitudinal Study Ped Res 2003 54 1 7 10.1203/01.PDR.0000066622.02736.A8
Hack M Schluchter M Cartar L Rahman M Cuttler L Borawski E Growth of Very Low Brith Weight Infants to Age 20 Years Pediatrics 2003 112 e30 e38 12837903 10.1542/peds.112.1.e30
Verloove-Vanhorick S Verwey R Brand R Bennebroek Gravenhorst J Keirse M Ruys J Neonatal mortality risk in relation to gestational age and birthweight: Results of a national survey of preterm and very-low-birthweight infants in the Netherlands Lancet 1986 1 55 57 2867312 10.1016/S0140-6736(86)90713-0
Fredriks A Van Buuren S Burgmeijer R Meulmeester J Beuker R Brugman E Roede M Verloove-Vanhorick S Wit J Continuing positive secular growth change in the Netherlands 1955–1997 Pediatr Res 2000 47 316 323 10709729
Fredriks A Van Buuren S Hirasing R Wit J Verloove-Vanhorick S Comparison of the Quetelet index (body mass index) in Dutch children in 1997 and 1980: new growth references for the detection of underweight, overweight and obesity Ned Tijdschr Geneesk 2001 145 1296 1303
Roede M Van Wieringen J Growth diagrams 1980: Netherlands third nation-wide survey Tijdschr Soc Gezondheidsz 1985 63 1 34
Kloosterman G Over Intra-uteriene groei en de intra-uteriene groeicurve Medschr v Kindergeneesk 1969 37 209 225
Kloosterman G On intrauterine growth Int J Gynaecol Obstet 1970 8 895 912
Ford G Doyle L Davis N Callanan C Very Low Birth Weight and growth into adolescence Arch Pediatr Adolesc Med 2000 154 778 784 10922273
Kok J den Ouden L Verloove-Vanhorick P Brand R Outcome of very preterm small for gestational age infants: the first nine years of life Br J Obstet Gynaecol 1998 105 162 168 9501780
Hokken-Koelega A De Ridder M Lemmen R Den Hartog H De Muinck Keizer-Schrama S Drop S Children born small for gestational age: Do they catch up? Pediatr Res 1995 38 267 271 7478827
Karlberg J Albertsson-Wikland K Growth in full-term small-for-gestational-age infants: from birth to final height Pediatr Res 1995 38 733 739 8552442
Fewtrell M Cole T Bishop N Lucas A Neonatal factors predicting childhood height in preterm infants: Evidence for a persisting effect of early metabolic bone disease? J Pediatr 2000 137 668 673 11060533 10.1067/mpd.2000.108953
Albertsson-Wikland K Boguszweski M Karlberg J Children born small-for-gestational age: postnatal growth and hormonal status Horm Res 1998 49 7 13 10.1159/000053080
Deal C Polymorphisms and Mutations in the GH-IGF Axis in Very Short Children Born SGA due to IUGR, and the Implications for the Patients in Childhood and Adult life Hormone Research 2003 59 130 12638525 10.1159/000067847
Albertsson-Wikland K Wennergren G Wennergren M Vilbergsson G Rosberg S Longitudinal follow-up of growth in children born small for gestational age Acta Paediatric 1993 82 428 443
Ibanez L Potau N Francois I de Zegher F Precocious Pubarche, Hyperinsulinism, and Ovarian Hyperandrogenism in Girls: Relation to reduced Fetal Growth J Clin Endocrinol Metab 1998 83 3558 3562 9768664 10.1210/jc.83.10.3558
Ibanez L Potau N Marcos M de Zegher F Exaggerated Adrenarche and Hyperinsulinism in adolescent Girls Born Small for Gestational Age J Clin Endocrinol Metab 1999 84 4739 4741 10599744 10.1210/jc.84.12.4739
Zegher de F Albertssson-Wikland K Wollmann H Chatelain P Chaussain J Lofstrom A Jonsson B Rosenfeld R Growth Hormone Treatment of Short Children Born Small for Gestational Age: Growth Responses with Continious and Discontinious Regimens over 6 Years J Clin Endocrinol Metab 2000 85 2816 2821 10946888 10.1210/jc.85.8.2816
Powls A Botting N Cooke R Pilling D Marlow N Growth impairment in very low birthweight children at 12 years: correlation with perinatal and outcome variables Arch Dis Child Fetal Neonatal Ed 1996 75 F152 F157 8976679
Hediger M Overpeck M McGlyn A Kuczmarski R Maurer K Davis W Growth and Fatness at three to six years of age of children born Small- or Large for Gestational Age Pediatrics 1999 104 1 6 10390252 10.1542/peds.104.3.e33
|
16033642
|
PMC1199602
|
CC BY
|
2021-01-04 16:31:06
|
no
|
BMC Pediatr. 2005 Jul 20; 5:26
|
utf-8
|
BMC Pediatr
| 2,005 |
10.1186/1471-2431-5-26
|
oa_comm
|
==== Front
BMC PediatrBMC Pediatrics1471-2431BioMed Central London 1471-2431-5-291608079810.1186/1471-2431-5-29Research ArticleIron absorption and oxidant stress during erythropoietin therapy in very low birth weight premature infants: a cohort study Friel James K [email protected] Khalid [email protected] Wayne L [email protected] Robert E [email protected] Department of Human Nutritional Sciences, University of Manitoba, Winnipeg, Manitoba, R3T 2N2, Canada2 Department of Biochemistry, Memorial University, St. John's, Newfoundland, A1B 3X9, Canada3 Department of Pediatrics, Memorial University, St. John's, Newfoundland, A1B 3X9, Canada4 Department of Preventive Medicine and Community Health, The University of Texas Medical Branch at Galveston, Galveston, Texas, 77555-1109, USA2005 5 8 2005 5 29 29 4 3 2005 5 8 2005 Copyright © 2005 Friel et al; licensee BioMed Central Ltd.2005Friel et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Iron supplementation may be associated with oxidative stress particularly in premature infants. Our purpose was to examine 1) early supplemental iron during treatment with erythropoietin (EPO) and oxidative stress; 2) enhanced iron absorption during EPO in those infants receiving human milk. Therefore, we determined the effect of erythropoietin plus supplemental iron intakes (4 mg/kg/d) on antioxidant status and iron incorporation.
Methods
Ten very-low-birth-weight infants who were enterally fed and receiving either human milk or formula were followed for 4 weeks during erythropoietin therapy; blood and urine were collected at 3 times; baseline, 2 and 4 weeks later. Once oral feeds commenced the study protocol was initiated. After baseline blood collection, a dose of Fe57 was administered. Two weeks later, a dose of Fe58 was administered as ferrous chloride to determine the effect of human-milk or formula on iron incorporation into RBCs.
Results
Infants started the study at 35 ± 13 days. Incorporation of isotope into RBCs did not differ between formula fed for Fe57 (mean incorporation 8 ± 2.9 n = 3) compared to human-milk fed infants (8.7 ± 5 n = 7) nor for Fe58 (6 ± 2.7 n = 3 vs. 8.6 ± 5 n = 7). Tissue damage measured by malondialdehyde in plasma and F-2 – isoprostanes in urine, did not differ by feed or over time. Neither ability to resist oxidative stress/nor RBC superoxide dismutase differed according to feed or over time.
Conclusion
Data suggest that during erythropoietin therapy antioxidant defence in VLBW infants are capable of dealing with early supplemental iron during treatment with EPO.
==== Body
Background
Infant formulas (F) have been the prevalent feed for very low birthweight (VLBW, < 1500 g birthweight) infants. More recently, neonatal units have successfully implemented human milk (HM) feeding programs with support groups so that human milk is now routinely fed to the majority of these infants. Rates vary across Canada from ~ 50% in St. John's NFLD to > 90 % in Victoria BC (Derek Matthews, Wayne Andrews, personal communication). Human milk alone may not meet all the nutritional needs of the growing VLBW infant during the first 2 months of life. Therefore, human milk is supplemented with appropriate nutrients [1,2]. Few studies [3,4] have examined iron intakes in VLBW infants receiving human milk. Therefore, the iron needs of this group are poorly defined.
A prevalent condition in VLBW infants during the first months of life is anemia. This "physiologic" anemia may be compounded by non-physiologic mechanisms such as red blood cell loss due to bleeding, hemolysis etc., and frequent blood sampling for clinical purposes [5]. Because the premature infant has a diminished erythropoietin (EPO) response to anemia, the administration of exogenous recombinant EPO has been proposed as a promising therapy. Early trials with EPO alone, reported variable results, which may have been due to limited availability of iron for hemoglobin formation [6-8]. While iron is now commonly provided with EPO, the amounts of iron supplements provided with EPO are not consistent and range from 0–12 mg/kg/d [6,9]. There is little experience with iron administered at these levels in the early life of the neonate and in only one of the trials with EPO (given intravenously as well as orally) has the safety of supplemental intakes of iron been examined [9]. Further, 1) little data is available on iron absorption in these infants 2) no distinction is made between breast-fed and formula fed infants in terms of possible differences in iron absorption related to the large amounts of iron provided as supplements during EPO. If iron absorption is indeed greater in human milk fed than formula-fed infants [10,11], then human milk fed VLBW infants may be at increased risk for problems associated with iron excess, particularly production of free radicals [12].
We used erythrocyte incorporation of an iron stable isotope as a surrogate for iron absorption in the intestinal tract (6, 9, 10). We assume that in identical conditions prompt erythrocyte incorporation of iron would account for the same percentage of newly absorbed iron by all infants in the study.
Our hypotheses were: 1) intakes of early supplemental iron during treatment with EPO would stress antioxidant defenses and lead to oxidant damage 2) iron absorption during EPO therapy plus iron supplements would be elevated with HM feeding. This study was necessary as the effect on oxidative stress during EPO therapy has not been reported nor has the possible increase in absorption of iron in human milk feeding been examined during EPO. We did not find any effect of HM on iron absorption nor an increase in oxidative stress during EPO.
Methods
The study included 10 VLBW infants who received EPO plus iron (2–4 mg/kg/d) as part of their normal clinical care. The study protocol was approved by the Memorial University Ethics Committee and informed written permission was obtained from parents/guardians. The baseline period was that time before the administration of EPO and iron isotopes. EPO treatment began when infants were tolerating 75% of feeds (non-iron fortified formula or human milk) by the enteral route. On two separate occasions, before commencement of EPO treatment, and again after 2 weeks of EPO treatment stable non-radioactive isotopes of iron (57Fe, 58Fe) were administered enterally. At baseline and again at 2 and 4 weeks of treatment, a blood sample was collected by venipuncture and a 12-hour urine sample was collected. We thereby tested the two hypotheses that firstly, early supplemental iron intakes (2–4 mg/kg/d) result in oxidant stress, and that secondly, incorporation of ingested iron is affected by type of feed (human milk or formula) that the infant receives.
Subjects and feedings
Eligibility required birth weight < 1500 g, < 34 weeks gestation receiving oral or gastric feeds (75%), stable respiratory status as defined by FIO2 ≤ 30%, and a mean airway pressure of 8 cm of water or less, if on assisted ventilation. Infants were not eligible if they received transfusions within the previous week or during the 4 week study period, had any major congenital defect, liver disease, necrotizing enterocolitis, grade III or IV intracranial hemorrhage or culture-proven sepsis.
All infants received parenteral nutrition with no added iron before study entry. During the study, 3 infants received formula (Similac Special Care, Ross Products Division, Abbott Laboratories, Columbus, OH) and the remaining 7 infants received their own mother's milk fortified with Enfamil Human Milk Fortifier (Mead Johnson Nutritionals, Evansville, IN). No iron or vitamin supplementation were given on the day of isotope administration, otherwise management of nutrition was under the direction of the medical team.
Beginning at study entry, r-HuEPO (R.W. Johnson Pharmaceutical Research Institute, Raritan, NJ) was administered S.C. in a dose of 200 U/kg 3 times a week for six weeks. Iron was given concomitantly in graded doses of 2 mg/kg/d wks 1 and 2 and 4 mg/kg/d, wks 3 and 4.
Administration of 57Fe and 58Fe test doses
At baseline (57Fe dosing) and again on d 14 (58Fe dosing) each infant received one dose with a regular feed (as described previously). A 24 hour equilibration time was used and is sufficient to allow isotope to mix with native iron in human milk [10]. The average daily volume of milk at baseline was 148 ± 20 mL/kg and at week 2, 158 ± 18 mL/kg (x ± SD).
Enriched 57Fe (95 weight % 57Fe, Atomergix Inc. Farmingdale, NY), and 58Fe (93.05 weight % 58Fe, Cambridge Isotope Labs, Andover, MA) were converted to ferrous chloride as described previously [13]. Enriched oxides were dissolved in a small amount of Aqua Regia (HCl: HN03, 3:1), diluted, pH adjusted to 5 with NaOH, brought up to a volume of 20 mL with saline and filtered through a 0.2 um filter and stored in glass vials until use. Before use, each vial was checked for purity from pyrogens with a test kit (Limulus Ambeocyte Lysate Test Kit: Whittaker Bioproducts, Walkersville, MD).
To prepare the enriched formulas, an amount of solution which provided approximately 1800 ug 57Fe/kg or 450 ug 58Fe/kg was added directly to the milk using a pre-weighed syringe. 57Fe was equilibrated with vitamin C overnight in milk alone. 58Fe was mixed in milk with vitamin C and a solution of natural abundance ferrous sulfate (Fer-In-Sol, Mead Johnson, Inc., Evansville, IN). The final iron content of the first feed with 57Fe was 0.2 mg/kg. The final iron content of the latter feed was set to 4 mg/kg made up from 58Fe, endogenous iron in the formula as labeled, (human milk was considered to have negligible iron) with the majority of iron from ferrous sulfate. This final dosage matched the usual treatment dose (4 mg/kg/d) provided with EPO at this time period. Milk administered by bottle was flushed with 5 mL of a 10% glucose solution.
Laboratory methods
Hemoglobin (Hgb) concentration was determined in whole blood by Coulter Counter (Coulter Electronics, Inc., Hialeah, FL). Whole blood was centrifuged and plasma removed. Red blood cells (RBCs) were washed 3 times in isotonic saline and both plasma and RBCs frozen at -80°C until analysis.
For isotopic analyses, iron was obtained from RBCs according to Fomon et al. [14]. Two week study periods were used as 2 weeks is the time required for enrichment of RBCs (14). For each of 10 infants, 3 separate samples were analyzed for a total of 20 estimations of iron absorption, plus 10 baseline measurements. Briefly, after microwave digestion with concentrated nitric acid (HNO3), in closed vessels, iron was selectively extracted with xylene that contained thenoyltrifluoroacetone and processed as reported previously [15]. Samples were analysed for 58Fe/54Fe and 57Fe/54Fe ratios by inductively coupled plasma mass spectrometry (ICP-MS, Plasma Quad3, Thermo Elemental, Franklin, MA US).
The quantity of administered 57Fe and 58Fe label incorporated into erythrocytes 14 days after each dose was calculated as described previously [14];
Total circulating iron was estimated as; Fe circ = BV * Hgb * 3.47. BV is blood volume, assumed to be 0.085 L/kg [16], Hgb is hemoglobin concentration in g/L, and 3.47 is the mg/g concentration of iron in Hgb. The following assumption was made; of the isotropic iron that is absorbed, by 14 d, 80% is incorporated into erythrocytes [14].
All glass and plastic beakers, flasks and pipettes were acid washed and only distilled deionised water was used throughout as previously reported [17]. Iron concentrations were determined by atomic absorption spectrometry (17). NIST standard reference material spectrophotometric grade iron was run as a reference standard and to determine a normalization factor applied to each sample. To measure oxidant stress in plasma, malondialdehyde (MDA) was measured by HPLC [18]. Total radical antioxidant protection (TRAP) as a measure of the ability to resist oxidative stress was measured using an oxygen electrode (YSI Model 5331) according to Wayner et al. [19]. In erythrocytes, superoxide dismutase (SOD) and catalase (CAT) were determined spectrophotometrically [20,21]. F2-Iso-prostanes as a measure of oxidant stress in urine were measured using a commercially available ELISA kits (Oxford Biomedical Research Inc., Oxford, MI). Information on oxygen exposure, medications, transfusions daily nutrient intakes and other data was recorded from nursing records.
Differences in descriptive data for study subjects were assessed by Student's t tests (SPSS- version 10, SPSS Inc., Chicago, IL). Pearson Correlation Coefficients were determined between iron absorption and study weight, birthweight, gender and gestational age. Differences in iron incorporation by type of feed and oxidative stress were measured by both parametric and nonparametric tests with consistent results.
Sample size was determined from measurements taken from previous studies in similar infants (13, 17) based on iron incorporation. We expected absorption in the F group to be 10 % with an SD of 5 %. In the HM fed group we expected absorption of 30% with an SD of 5 based on our earlier work (13, 17). Thus with an expected difference of 20 %, a power of 0.8 and an alpha of 0.05 we would need 5 subjects in each group. Significance for all assays was assigned to p < 0.05 (two-tailed).
Results
Infants weighed 1046 ± 212 g at birth and were 28 ± 2 weeks gestational age (x ± SD: n = 10). All infants gained approximately 30 g/d in weight during the 4 week study period. For these infants: age at study entry was 35 ± 13 days (range 22–68 days); weight at study entry was 1363 ± 217 g; week 2, 1708 ± 312 g; and week 4, 2167 ± 283 g. HGB at baseline (87 ± 12 g/L) week 2 (89 v 11 g/L) and week 4 (99 ± 12 g/L) increased significantly between weeks 2 and 4. There was no difference in birth weight, gestational age or weight at baseline between HM (n = 7) fed infants and F (n = 3) fed infants.
Data for each subject is presented in Table 1. Incorporation into RBCs of 57Fe after the first 14 days of the study period was 8.7 ± 5 % for HM fed infants and 8 ± 3 % for F fed infants and did not differ. Incorporation of 58Fe from days 14 to 28 of the study period into RBCs was 8.6 ± 5 % for HM fed infants and 6.0 ± 2.7 % for F fed infants and did not differ. Mean percent relative standard deviation for all 30 samples including baseline that were analyzed for isotopic enrichment was 0.62 ± 0.10 for 57Fe/54Fe ratios and 0.77 ± 0.29% for 58Fe/54Fe ratios.
Table 1 Oxidant status and iron incorporation in VLBW Infants during treatment with EPO plus iron
Baseline Week 2 Week 4
PLASMA TRAP (umol/L 492 ± 3451 510 ± 316 438 ± 293
PLASMA MDA (umol/L) 0.35 ± 0.47 0.33 ± 0.24 0.42 ± 0.48
URINE F2 -Isoprostanes (ng/mL) 19 ± 14 29 ± 30 30 ± 19
RBC SOD (U/mL) 3.1 ± 2 4.0 ± 2.3 3.8 ± 2.1
RBC CAT (U/mL) 62 ± 19 68 ± 25 75 ± 25
INCORP (%) ______ 8.5 ± 4.2 7.8 ± 4.4
1means ± S.D.
Energy and protein intake and volume of milk fed were at base line; 103 ± 18 kcals/kg, 2.6 ± 0.6 g/kg and 139 ± 17 ml/kg. At week 2, energy intake was 134 = 30 kcal/kg, protein intake was 3.1 ± 0.7 g kg and volume of milk fed was 158 ± 10 ml/kg.
Data for plasma TRAP, MDA, urinary F-2-isoprostanes, RBC SOD and CAT are presented in Table 1. As there was no difference between feeding groups, data were pooled. Control values for healthy adults and premature infants of similar gestational ages not receiving EPO from our laboratory are: for TRAP 964 ± 240, umol/L (adult, n = 23); MDA 0.35 ± 0.17 umol/L (premature infant n = 36) 0.38 ± 0.16 (adult, n = 13); urinary F2 isoprostanes 28 ± 13 ng/mL (premature infant, n = 24); RBC SOD, 3.2 ± 1.8 U/ml (premature infant, n = 33) 1.2 ± 0.3 (adult n = 10); RBC CAT 68 ± 14 U/mL (premature infant, n = 30) 12 ± 20 u/mL (adult, n = 20). As well, each infant at baseline served as their own control for later analyses.
Measures of antioxidant status and oxidant stress did not correlate with birth weight, gestational age, Apgar scores, and volume of feed nor days of oxygen exposure.
Discussion
Iron isotopes
While our sample size is small (n = 10 infants), we were able to obtain 3 measurements from each infant serving as their own control. As well, the sample size is similar to isotope studies reported in other infants (6, 10). We had hoped to enrol at least 5 infants in each group as our power analysis suggested, but this did not occur, thereby increasing the chance of a Type II error. From this data, we did not find any difference in incorporation of iron isotopes into RBCs between VLBW infants who were receiving either human milk or premature formula (Table 1). In a previous experiment [13] we found that isotope incorporation could be affected by zinc intakes in these infants in a non-meal situation. We chose however to examine iron absorption as it would occur naturally in the NICU, which is during the fed state. However, we are left too speculate on the comparisons between feeding groups because of the small sample size. We believe there is merit in presenting this data as we originally hypothesized large differences which clearly did not occur.
While it is known that food components can enhance iron absorption, in general iron absorption decreases with food intake [22]. In-between feedings iron absorption has been shown to range from 32 to 42% exhibiting wide variation [4,23,24]. With meals, absorption of stable and radioactive iron isotopes has been reported to be from 2.8 to 15.3%, a range that our results fall within [4,25-27]. It has been suggested that iron absorption by premature infants is unregulated by the level of iron intake [4].
Fomon [14] found that the mean RBC Fe58 incorporation was 7.8% for HM fed infants compared to 4.4% for formula fed infants even when the isotope was fed between meals. He suggested that his difference was due to the greater quantities of inhibitors present in the intestine of formula fed infants. Whether or not the assumed greater absorption of Fe in HM is due to enhancement by HM or lack of inhibitors present in formula feeding is still unresolved. This latter report was one of the studies that generated our hypothesis however; we did not find the same results as Fomon [14] as we studied VLBW infants who received 4 times the iron dose as did full-term infants in the latter study. As well, Fomon [14] suggested that greater iron incorporation into RBCs in HM fed infants made up for lower Fe content in HM whereas in our study during EPO treatment, the quantity of exogenous iron added to feedings was so large as to mask any contribution from milk feedings. The finding could also be a Type II error due to the uneven sample size.
However the important finding in the current study is that RBC incorporation of iron isotope does not appear to be different in HM fed infants as compared to formula fed infants in any significant way. Earlier reports suggest an order of magnitude difference between HM and formula fed infants (10, 11, 13) particularly in our earlier study where one HM fed infants had an absorption of 35% of the isotopic dose. Thus even the 1–2% possible difference in our study that may have been found with a larger sample size does not indicate a concern for over absorption of iron in HM fed premature infants when larger does of iron are given with EPO therapy. Therefore concern of potential iron excess or increased free radical generation with EPO plus iron in HM fed infants does not appear warranted.
Oxidant stress
There has only been one study to date examining the effects of EPO on potential free radical damage [9] in VLBW infants and one report in adults [28]. Pollack [9] reported a transient rise in MDA after intravenous iron infusion. The present study with a more comprehensive assessment did not find similar results either in MDA or F2 isoprostane levels. It is well known that iron may cause oxidative damage however much of the data is from in vitro assessment. As there is a substantial increase in iron utilization during treatment with EPO, it may be that iron is sequestered and not available for free radical chemistry adults, EPO therapy was studied in hemodialysed patients with and without iron [28]. EPO plus iron did not lead to increased MDA levels however SOD, GHSPx and CAT increased in this group of patients. In the present study, we found a trend (p = 0.07) to increased CAT levels. The authors in the latter study suggested that increased enzyme activity levels might be due to an increased number of reticulocytes or a stimulation of enzyme synthesis in young RBC by reactive oxygen species. Differences in these latter studies may be due to different populations and the much longer time of treatment (>6 months) in the adult population. In the current study, lack of evidence of lipid peroxidation suggests that increased CAT values are physiological and well within the infants' ability to handle the current amounts of iron given with EPO.
Recent work has also suggested that EPO plus early iron does not affect anthropometry, need for rehospitilization, transfusion after discharge nor developmental outcome (29). Thus longer term data supports our short term data that at least from the point of oxidative stress premature infants can cope with early supplemental iron plus EPO.
Conclusion
a) When EPO therapy is used, particularly with iron intakes up to 4 mg/kg/d, our data suggests that, in regards to oxidative stress, the process appears to be safe.
b) Human milk feeding during early supplemental iron during treatment with EPO does not enhance iron absorption. Increased iron absorption would have been a concern in the VLBW population.
Abbreviations
CAT Catalase
EPO Erythropoietin
F Formulas
57Fe Iron stable isotope mass 57
58Fe Iron stable isotope mass 58
HCL Hydrochloric acid
Hgb Hemoglobin
HM Human milk
HNO3 Nitric acid
ICP-MS Inductively coupled mass spectrometry
INCORP Iron Incorporation (%)
MDA Malondialdehyde
NIST National Institute of Standards and Technology
RBC Red blood cell
SOD Superoxide dismutase
TRAP Total Radical Trapping Antioxidant Parameter
VLBW Very low birth weight
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
RES was responsible for isotopic analyses. KA and WLA were in addition responsible for patient recruitment. JKF originated the study, obtained funding and was responsible for day to day maintenance of the study protocol. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We acknowledge financial support from the Janeway Research Foundation and Canadian Institutes of Health Research grant # 146833. The technical help of Claude Mercer and Allison McDonald is gratefully acknowledged. As well, the parents of infants and the nurses in the NICU contributed greatly to this work. The authors declare that they have no competing interests.
==== Refs
Atkinson SA Jensen R Effect of gestational stage at delivery on human milk components Handbook of Milk Composition 1995 San Diego: Academic Press 222 234
Nutrient needs and feeding of premature infants. Nutrition Committee, Canadian Pediatric Society C M A J 1995 152 1765 1785
Shaw JC Iron absorption by the premature infant. The effect of transfusion and iron supplements on the serum ferritin levels Acta Paediatr Scand Suppl 1982
Dauncey MJ Davies CG Shaw JC Urman J The effect of iron supplements and blood transfusion on iron absorption by low birth weight infants fed pasteurized human breast milk Pediatr Res 1987 12 899 904 714536
Strauss RG Erythropoietin in the pathogenesis and treatment of neonatal anemia Tranfusion 1995 35 68 73 10.1046/j.1537-2995.1995.35195090667.x
Widness JA Lombard KA Ziegler EE Serfass RE Carlson SJ Johnson KJ Miller JE Erythrocyte incorporation and absorption of 58Fe in premature infants treated with erythropoietin Pediatr Res 1997 41 1 8 8979282
Gallagher PG Ehrenkranz RA Erythropoietin therapy for anemia of prematurity Clin Perinatol 1993 20 169 91 8458164
Widness JA Seward VJ Kromer IJ Burmeister LF Straus RG Changing patterns of red blood cell transfusions in very low birthweight infants J Pediatr 1996 129 680 687 8917234
Pollak A Hayde M Hayn M Herkner K Lombard KA Lubec G Effect of intravenous iron supplementation on erythropoiesis in erythropoietin-treated premature infants Pediatrics 2001 107 78 85 11134438 10.1542/peds.107.1.78
Davidsson L Kastenmayer P Yuen M Lonnerdal B Hurrell RF Influence of lactoferrin on iron absorption from human milk in infants Pediatr Res 1994 35 117 124 8134189
Saarinen UM Siimes MA Dallman PR Iron absorption in infants: high bioavailability of breast milk iron as indicated by the extrinsic tag method of iron absorption and by concentration of serum ferritin J Pediatr 1977 91 36 9 577504
Berger HM Mumby S Gutteridge JC Ferrous ions detected in iron-overloaded cord blood plasma from preterm and term babies: implications for oxidative stress Free Radic Res 1995 22 555 9 7543335
Friel JK Serfass RE Fennessey PV Miller LV Andrews WL Simmons BS Downton GF Kwa PG Elevated intakes of zinc in infant formulas do not interfere with iron absorption in premature infants J Pediatr Gastroenterol Nutr 1998 27 312 316 9740203 10.1097/00005176-199809000-00008
Fomon SJ Ziegler EE Nelson SE Erythrocyte Incorporation of 58Fe by 56-day-old breast-fed and formula-fed infants Pediatr Res 1993 33 573 576 8378114
Moore FL Selective liquid-liquid extraction of iron with 2-Thenoyltrifloracetone-Xylene Analytical Chemistry 1959 31 1148 51 10.1021/ac60151a023
Lentner C Geigy Scientific Tables 1984 Chapter 3 Ciba-Geigy, Switzerland: 65 7
Friel JK Andrews WL Long DR Hall M Rodway M Meeting the iron requirements of parenterally fed very low birth weight infants JPEN 1995 19 302 310
Lepage G Munoz G Champagne J Roy CC Preparative Steps Necessary for the Accurate Measurement of Malondialdehyde by HPLC Anal Biochem 1991 197 277 83 1785679 10.1016/0003-2697(91)90392-7
Wayner DDM Burton GW Ingold KU Locke S Quantitative measurement of the total peroxyl radical-trapping antioxidant capability of human blood plasma by controlled oxidation Fed Eur Bio Soc 1985 87 33 7
Oberley LW Spitz DR Packer L Assay of superoxide dismutase in tumour tissue Methods in Enzymology 1984 Academic Press 457 469 6547201
Claiborne A Greenwald RA Catalase activity CRC Handbook of Methods for Oxygen Radical Research 1985 Boca Raton, Fl: CRC Press, Inc 283 4
Fairweather-Tait SJ The concept of bioavailability as it relates to iron nutrition Nutrition Research 1987 7 319 325
Ehrenkranz RA Gettner PA Nelli C Sherwonit EA Williams JE Pearson HA Ting BT Jangherbani M Iron absorption and incorporation into red blood cells by very low birth weight infants: Studies with the stable isotope 58Fe J Pediatr Gastroenterol Nutr 1982 15 270 8 1432464
Gorten MK Cross ER Iron metabolism in prematures J Pediatrics 1964 64 509 20
Zlotkin S Lay D Kjarsgard J Longley T Determination ofIron absorption in very low birth weight premature infants using two isotopes of iron (57Fe and 58Fe) J Pediatr Gastroenterol Nutr 1995 21 190 199 7472906
Ehrenkranz RA Tsang RC, Lucas A, Uauy R, Zlotkin S Iron, folic acid and vitamin B12 Nutritional Needs of the Preterm Infant 1993 Williams and Wilkins Baltimore 177 94
Oettinger L JrMills WB Hahn PF Iron absorption in premature and full-term infants J Pediatr 1964 45 302 306 13202028
Delmas-Beauvieux MC Combe C Peuchant E Carbonneau MA Dubourg L de Precigout V Evaluation of red blood cell lipoperoxidation in hemodialysed patients during erythropoietin therapy supplemented or not with iron Nephron 1995 69 404 10 7777104
Ohls RK Ehrenkranz RA Das A Dusick AM Yolton K Romano E Delaney-Black V Papile LA Simon NP Steichen JJ Lee KG National Institute of Child Health and Human Development Neonatal Research Network Neurodevelopmental outcome and growth at 18 to 22 months' corrected age in extremely low birth weight infants treated with early erythropoietin and iron Pediatrics 2004 114 1287 91 15520109 10.1542/peds.2003-1129-L
|
16080798
|
PMC1199603
|
CC BY
|
2021-01-04 16:31:06
|
no
|
BMC Pediatr. 2005 Aug 5; 5:29
|
utf-8
|
BMC Pediatr
| 2,005 |
10.1186/1471-2431-5-29
|
oa_comm
|
==== Front
BMC PediatrBMC Pediatrics1471-2431BioMed Central London 1471-2431-5-311610915810.1186/1471-2431-5-31Research ArticleAcute bronchiolitis in infancy as risk factor for wheezing and reduced pulmonary function by seven years in Akershus County, Norway Fjærli Hans-Olav [email protected] Teresa [email protected]ød Gisle [email protected] Gunn Kristin [email protected] Pål [email protected] Britt [email protected] Faculty Division Akershus University Hospital, University of Oslo, 1474 Nordbyhagen, Norway2 Department of Paediatrics, Akershus University Hospital, 1478 Lørenskog, Norway3 Voksentoppen Centre for Asthma and Allergy in Children, Rikshospitalet, 0791 Oslo, Norway2005 18 8 2005 5 31 31 17 3 2005 18 8 2005 Copyright © 2005 Fjærli et al; licensee BioMed Central Ltd.2005Fjærli et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Acute viral bronchiolitis is one of the most common causes of hospitalisation during infancy in our region with respiratory syncytial virus (RSV) historically being the major causative agent. Many infants with early-life RSV bronchiolitis have sustained bronchial hyperreactivity for many years after hospitalisation and the reasons for this are probably multifactorial. The principal aim of the present study was to investigate if children hospitalised for any acute viral bronchiolitis during infancy in our region, and not only those due to RSV, had more episodes of subsequent wheezing up to age seven years and reduced lung function at that age compared to children not hospitalised for acute bronchiolitis during infancy. A secondary aim was to compare the hospitalised infants with proven RSV bronchiolitis (RS+) to the hospitalised infants with non-RSV bronchiolitis (RS-) according to the same endpoints.
Methods
57 infants hospitalised at least once with acute viral bronchiolitis during two consecutive winter seasons in 1993–1994 were examined at age seven years. An age-matched control group of 64 children, who had not been hospitalised for acute viral bronchiolitis during infancy, were recruited from a local primary school. Epidemiological and clinical data were collected retrospectively from hospital discharge records and through structured clinical interviews and physical examinations at the follow-up visit.
Results
The children hospitalised for bronchiolitis during infancy had decreased lung function, more often wheezing episodes, current medication and follow-up for asthma at age seven years than did the age matched controls. They also had lower average birth weight and more often first order family members with asthma. We did not find significant differences between the RSV+ and RSV- groups.
Conclusion
Children hospitalised for early-life bronchiolitis are susceptible to recurrent wheezing and reduced pulmonary function by seven years compared to age-matched children not hospitalised for early-life bronchiolitis. We propose that prolonged bronchial hyperreactivity could follow early-life RSV negative as well as RSV positive bronchiolitis.
==== Body
Background
Acute viral bronchiolitis is one of the most common causes of hospitalisation during infancy in Akershus County, Norway, with respiratory syncytial virus (RSV) historically being the major causative agent. RSV causes respiratory disease in young children worldwide and by the age of two years most children have been infected. In temperate climates the infection occurs as yearly winter epidemics and the impact of RSV on human health is demonstrated annually when infants are admitted to hospitals in large numbers. Symptoms vary from a mild upper respiratory tract infection to severe bronchiolitis with hyperinflated lungs and hypoxemia [1]. The first infection is usually the most severe but milder reinfections are common throughout life. The risk of severe illness is highest in infants born prematurely and in those with chronic lung disease, certain congenital heart defects and immunodeficiency disorders [2,3]. Many infants with severe RSV bronchiolitis experience recurrent wheezing in later childhood and there is growing evidence that early-life RSV bronchiolitis may predispose some infants to the development of childhood asthma [4-6]. The genetic background of the infant, intermittent changes in host cellular immune responses and neural control leading to sustained bronchial hyperreactivity and recurrent wheezing, timing of RSV infection with respect to allergen exposure, environmental conditions and exposure to endotoxin are all factors suggested to contribute to RSV induced asthma [7-9].
Human metapneumovirus (hMPV) is now recognised as one of several other viral pathogens that can cause acute bronchiolitis, the remaining being mainly rhinovirus, parainfluenzavirus, influenzavirus and adenovirus [10]. hMPV have recently been found to be an important cause of acute bronchiolitis in infants and children. However, the long-term effect of this and other viral agents on lung function and symptoms in later childhood is not yet fully investigated [11,12].
The primary aim of this study was to compare the airway symptoms and lung function of children hospitalised for early-life bronchiolitis with age matched controls without such a history at age seven years. We also aimed to describe possible differences between those with proven RSV bronchiolitis and those with non-RSV bronchiolitis regarding the same endpoints.
Methods
Study population
During two consecutive years in 1993–1994 a sample of nasopharyngeal aspirate (NPA) was collected from infants hospitalised with acute respiratory disease. The doctors on call conducted, as part of the admission routine, clinical interviews using structured questionnaires and performed physical examination of all infants admitted with suspected acute bronchiolitis.
In the present follow-up study at age seven years hospital records, questionnaires and microbiological results of all children hospitalised during infancy with acute bronchiolitis were analysed in retrospect. 109 infants presenting with respiratory insufficiency such as tachypnoe, intercostal retractions, increased mucus production and soar coughing were eligible for follow-up. However, we excluded 11 infants who were born before completed 37 wGA (weeks of gestational age), 11 infants who had divergent results in two different microbiological tests, 22 infants who had insufficient NPA material for testing, two children who had died from unrelated causes and six children that did not want to participate or had moved far away from our region.
The remaining 57 children were willing and able to take part in the study and came for a follow-up visit at age seven years. Children hospitalised during infancy with verified RSV bronchiolitis (RS+) had to be positive for RSV in two different microbiological tests, and children hospitalised with non-RSV bronchiolitis (RS-) had to be negative for RSV in the same two different microbiological tests.
Microbiological tests
Two different tests for detection of RSV in NPA were used. These were our own in-house Abbott TestPack RSV enzyme immunoassay (EIA) and immunofluorescens (IF) staining performed in our reference laboratory at the Department of Microbiology, Rikshospitalet, Oslo [13,14].
Structured clinical interview at the follow-up visit
A clinical interview based on the structured questionnaire used when the children with bronchiolitis were admitted to hospital were performed by three of the authors in the study (H.O.F, T.F., G.R.) to give information on episodes of wheezing (defined by us as episodes of difficult breathing accompanied by a whistling noise in the chest during expiration), number of visits to a family physician, practicing paediatrician and resubmissions to any hospital for wheezing or asthma (as diagnosed by a medical doctor) during the 6–7 years that had passed since the hospitalisation in infancy. Hospital records confirmed number of hospitalisations, whereas the parents gave information regarding visits to a family physician or practicing paediatrician. Information on failure to thrive, regular medications and physical activity was given in the interviews, as was information on parental smoking habits, number of preschool (<6 years old) siblings and prevalence of allergy, eczema and asthma in first order family members (mother, father or siblings) that had been diagnosed by a medical doctor.
Physical examination and pulmonary function tests at the follow-up visit
The medical doctors who conducted the structured clinical interviews performed weight and height measurements and a general clinical examination including inspection and auscultation at the follow-up visit. All children performed a standard pulmonary function test (PFT) under the supervision and guidance of a skilled paediatric nurse (G.K.U.). Forced vital capacity (FVC), forced expiratory volume in one second (FEV1) and forced expiratory flow rate at 50% of FVC (FEF50) were measured with the Vmax 20 equipment (SensorMedics). The equipment was calibrated daily according to the specifications of the producer. Measurements of pulmonary function were performed both before and 10 minutes after inhalation of 0,2 mg salbutamol (Airomir Autohaler) and the one attempt (out of three) with the best technical profile both before and after salbutamol was used for analysis.
The control group
An age-matched control group of seven years old children were recruited from a nearby primary school. A letter was distributed by the school principal to the parents of 92 children (three school classes) providing information about the study and explaining the criteria set for participation. All children had to be healthy term infants born at completed 37 wGA and without history of hospitalisation for respiratory problems during the first year of life. 64 children fulfilled the inclusion criteria, gave informed consents and took part in the study. At the follow-up visit the inclusion criteria were confirmed, an identical structured clinical interview with the parents of the control children were undertaken and the children performed the same PFT as the other children in the study.
Statistics
For comparison of continuous data between two groups the independent samples T-test was used and for comparison of ordinal and nominal data between two groups we used chi square tests. A p-value <0.05 was considered statistically significant and continuous data were given with 95% confidence intervals (CI). The statistical programme SPSS, version 12.0.1, was used in all calculations.
Ethics
The regional ethics committee for medical research approved the study and the parents gave oral and written informed consent.
Results
Structured clinical interviews
Among the 57 children hospitalised with acute bronchiolitis in infancy, 35 children (61%) tested double RSV positive while 22 (39%) tested double negative for RSV in NPA. Mean birth weight, mean age at hospitalisation and mean length of hospitalisation including admission day were not significantly different between these groups. According to data collected at admission, seven (32%) in the RS- and 15 (43%) in the RS+ group had experienced one or more episodes of a possible lower respiratory tract infection prior to the first hospitalisation for bronchiolitis.
There were not significant differences between the RS- and RS+ group or between these two groups combined and the control group regarding gender, weight or height at the follow-up visit. Mean birth weight was significantly lower in the hospitalised group compared to the control group (tab 1). First order family members of the hospitalised children had significantly higher prevalence of asthma than first order family members of the control group (39% vs 19%, p = 0.015), while no significant differences were observed for other risk factors (tab 1). Within the hospitalised groups no significant differences were found for any of the risk factors (tab 2).
Table 1 Follow-up of children hospitalised or not with acute bronchiolitis in infancy
Reported at child age seven years Hospitalised group Not hospitalised group
RS- (n = 22) RS+ (n = 35) 95% CIa Both (n = 57) Controls (n = 64) 95% CIa
Age, mean (range) at hospitalisation, months 5.9 (0–11) 5.1 (0–9) -0.8 2.5
Length, mean (range) of hospitalisation, days 4.6 (2–20) 4.8 (2–14) -2.1 1.8
Birth weight, mean (range), gramsb 3507 (2420–4300) 3506 (2500–4045) -225.6 228.5 3506 (2420–4300) 3733 (2100–5220) -425.2 -27.7
Weight, mean (range) at follow-up visit, kg 27.5 (23–42) 28.3 (22–53) -4.0 2.4 28.0 (22–53) 29.2 (20–45) -0.8 3.2
Height, mean (range) at follow-up visit, cm 127.7 (118–139) 127.3 (117–146) -2.8 3.7 127.4 (117–146) 128.6 (115–145) -1.0 3.4
Age, mean (range) at follow-up visit, months 93.6 (86–101) 92.8 (86–102) -1.2 2.9 93.1 (86–102) 94.0 (88–99) -0.4 2.2
No. boys (%) 12 (54) 20 (57) 32 (56) 36 (56)
a 95% confidence interval (CI) of difference between means
b P = 0.023 with independent samples T-test between the hospitalised and not hospitalised group
Table 2 Risk factors in family members of children hospitalised or not with acute bronchiolitis in infancy
Reported at child age seven years Hospitalised group Not hospitalised group
No. families with risk factorsa RS- (n = 22) RS+ (n = 35) Both (n = 57) Controls (n = 64)
Allergy before seven years 8 13
Current allergy 12 16 28 33
Eczema before seven years 6 7
Current eczema 8 14 22 28
Asthma before seven years 10 8
Current asthmab 10 12 22 12
Siblings <6 years before seven years 12 22
Current siblings <6 years 6 11 17 26
Parental smoking before seven years 10 16
Current parental smoking 10 19 29 28
a Risk factors in first order family members (mother, father, siblings) as reported by parents
b P = 0.015 with chi square test between the hospitalised group and not hospitalised group
In the group of children hospitalised with acute bronchiolitis in infancy 32 subsequent hospitalisations in 24 children were noted with peak incidence in the first year of life. Only one child in the control group needed hospitalisation for wheezing during the study period (fig 1).
Figure 1 Subsequent readmissions after hospitalisation for acute viral bronchiolitis during infancy.
29 children in the hospitalised group were reported with ≥3 episodes of wheezing during the follow-up period compared to 9 children in the control group (51% vs 14%, p < 0.001). No significant differences were found between the RS+ and RS- groups in this respect (tab 3).
Table 3 Wheezinga ever and current asthma in children hospitalised or not with acute bronchiolitis in infancy
Reported at child age seven years Hospitalised group Not hospitalised group
RS- (n = 22) RS+ (n = 35) Both (n = 57) Controls (n = 64)
≥3 episodes of wheezing ever during follow-upd 12 17 29 9
Current follow-up for asthma by a medical doctorb,e 10 21 31 5
Current medication for asthmac,e 8 12 20 5
a Wheezing ever in childhood treated by a family physician or a practicing paediatrician
b Current follow-up for asthma by a family physician or a practicing paediatrician
c Current asthma medication like β-2 agonists or inhaled corticosteroids
d P < 0.001 with chi square test between the hospitalised group and not hospitalised group
e P < 0.001 with chi square test between the hospitalised group and not hospitalised group
At the follow-up visit 31 children in the hospitalised group were currently under supervision for asthma by a practicing paediatrician or a family physician, in contrast to five children in the control group (54% vs 8%, p < 0.001). More children in the hospitalised group than in the control group used asthma medication currently (35% vs 8%, p < 0.01). No statistical differences were found between the RS+ and RS- groups for these parameters (tab 3).
Pulmonary function tests
No differences were found between the RS+ and RS- groups for FVC, FEV1 and FEF50 when comparing predicted values and test results before and after inhalation of salbutamol (tab 4). The test results for FVC after salbutamol, FEV1 before salbutamol and FEF50 before and after salbutamol were significantly worse in the combined hospitalised group than in the control group (tab 4). We did not find differences between groups for reversibility when measured as percent improvement from baseline after inhalation of salbutamol for any of the test parameters (tab 4).
Table 4 Spirometry in children hospitalised or not with acute bronchiolitis in infancy
Age seven years Hospitalised group Not hospitalised group
FVC (liters) RS- (n = 22) RS+ (n = 35) 95% CIe Both (n = 57) Ctr (n = 64) 95% CIe
Predicteda 1.78 1.76 -0.12 0.15 1.77 1.82 -0.04 0.14
Before salbutamol 1.65 1.65 -0.17 0.17 1.65 1.75 -0.02 0.19
After salbutamolc 1.69 1.69 -0.16 0.17 1.69 1.81 0.004 0.22
Reversibilityb 2.5 1.9 -2.3 3.6 2.1 3.3 -9.3 3.2
FEV1 (liters)
Predicteda 1.51 1.50 -0.10 0.12 1.51 1.54 -0.04 0.11
Before salbutamolc 1.50 1.51 -0.16 0.13 1.50 1.64 0.05 0.23
After salbutamol 1.57 1.59 -0.16 0.11 1.59 1.70 0.03 0.20
Reversibilityb 4.9 5.6 -4.0 2.6 5.3 5.6 -3.5 3.9
FEF50 (liters/min)
Predicteda 2.34 2.33 -0.12 0.14 2.33 2.36 -0.07 0.13
Before salbutamold 1.92 1.99 -0.35 0.20 1.96 2.28 0.15 0.49
After salbutamold 2.17 2.34 -0.43 0.10 2.28 2.56 0.11 0.46
Reversibilityb 11.7 14.5 -10.0 4.3 13.4 9.6 -9.2 1.6
a According to European Respiratory Society 1993 Update
b Percent improvement 10 minutes after inhalation of 0.2 mg salbutamol
c P < 0.05 with independent samples T-test between the hospitalised group and not hospitalised group
d P < 0.01 with independent samples T-test between the hospitalised group and not hospitalised group
e 95% confidence interval (CI) of difference between means
We also performed all analyses excluding the children with prior infection, first from the RS- group, then from both groups. There were not any substantial changes in observations or conclusions. These data are not shown.
Discussion
The results of this follow-up study show that children hospitalised with acute bronchiolitis during infancy in our region are at increased risk for subsequent childhood wheezing and for reduced pulmonary function, at least up to age seven years. The children were all admitted to the same hospital and from two consecutive winter seasons.
Mean length of hospitalisation including admission day is in accordance with other studies [15]. Mean age at first hospitalisation is higher than in other studies, and 39% of the children had experienced episodes of possible RSV infections before hospitalisation [16,17]. These children were not unevenly distributed on the RS+ and RS- groups. We were not able to observe substantial changes in the results when excluding children with prior infections. This is not surprising. First, since the number of observations is reduced by this procedure, larger differences are necessary to provide significant changes. Secondly, none of these prior infections were severe enough to require admission to hospital. Accordingly, they did probably not meet the criteria of Ruuskanen and Ogra [18] or Court [19], and we assume it rather unlikely that they were due to RSV. Nevertheless, there is a small possibility that we have underestimated differences between the RS negative and positive groups.
We were surprised to find that many of the hospitalised children were deemed RSV negative in NPA, since they were examined in the middle of an RSV epidemic. Obviously, the two microbiological techniques used in our study may not have been sensitive and specific enough due to properties of the tests themselves [20]. In 1993–1994 we did not have access to more sensitive microbiological tests, such as detection of nucleic acid by reverse-transcriptase polymerase chain reaction (PCR) [21]. However, using new knowledge about the performance of the tests we used compared to PCR [22,23], we have calculated that the overall probability of misclassification in our study is 4%, restricted to one or two false negatives. Thus, we find it unlikely that the results in the RSV negative group are contaminated by misclassification of the microbiological agent. For all variables studied, the results in these two groups are so similar, that even a much larger study would not be likely to detect differences.
Several studies have shown that hospitalisation for acute RSV bronchiolitis in infancy is an important risk factor for subsequent wheezing [24,25]. We show that hospitalisation for RSV negative bronchiolitis is an equal important risk factor for subsequent wheezing as is RSV positive bronchiolitis. The intensity of the host inflammatory response, and not only the virus itself, is probably a major determinant to whether or not bronchiolitis develops at the time of infection. This has been well documented in a recent study comparing infants with bronchiolitis due to RSV and influenzavirus [26]. Also, a high frequency of dual infection by RSV and hMPV has been reported in infants with severe bronchiolitis receiving mechanical ventilation [27].
The incidence of childhood asthma, when defined as three or more episodes of wheezing ever during follow-up, was in our study 51% in the hospitalised group and 14% in the control group. Correspondingly, as reported by parents, 54% in the hospitalised group and 8% in the control group were at age seven years consulting for asthma with a family physician or a practicing paediatrician. Our findings further support that hospitalisation for early-life bronchiolitis, not only those due to RSV, is associated with increased prevalence of childhood asthma, at least up to age seven years [28,29].
The percentage of smokers among parents of the hospitalised children in our study was high. However, a high percentage of parents currently smoking was also recorded in the control group. Exposure to tobacco smoke seemed not to be an independent risk factor for childhood wheezing in our study. Several studies have found parental smoking to be an important risk factor for wheezing in early childhood but not in older children. In the Tucson Children's Respiratory Study the greatest impact was found in children under three years of age when mothers were smoking during pregnancy [30].
No statistically significant differences were found between the hospitalised group and the control group in current prevalence of allergy or eczema in first order family members and number of families with preschool siblings. Heredity for asthma in the hospitalised group seemed to be the only significant risk factor in first order family members for childhood wheezing in our study. Several other studies have shown that the prevalence of asthma among family members of infants with acute viral bronchiolitis is slightly higher than in the general population [31].
Early-life RSV bronchiolitis has in many studies been associated with sustained impaired lung function. Our study supports this by demonstrating significantly reduced test results for several parameters in the hospitalised group compared to the control group. Our results are in accordance with a study from Noble et al in which the index children had significantly reduced lung function at nine to ten years of age [32]. According to a study by Sigurs et al severe RSV bronchiolitis in infancy was significantly associated with reduced lung function even at 13 years of age [6].
Studies have shown that pulmonary function at birth of newborns suffering from acute bronchiolitis later on during infancy are reduced compared to pulmonary function at birth of newborns not suffering from acute bronchiolitis later on during infancy [33,34]. This might indicate a congenital pulmonal predisposition for bronchiolitis susceptibility and later obstructive lung disease. A significant lower birth weight in the hospitalised infants in our study might indicate that they had smaller lungs or delayed pulmonal development compared to the controls.
Conclusion
Acute RSV bronchiolitis is a major health problem in our region with a mean hospitalisation incidence for the period 1993–2000 in infants of 23.6 per 1.000 [35]. Many infants with acute viral bronchiolitis, not only those due to RSV, will have episodes of recurrent wheezing, resubmissions to hospital and reduced lung function, at least up to age seven years. The results of the present study indicate that host factors may be as important as the nature of the infecting agent for the development of bronchiolitis and later childhood wheeze and need further studies.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
HOF had primary responsibility for protocol development, outcome assessment, data acquisition and analysis and writing of the manuscript. TF participated in the development of the protocol, outcome assessment, data acquisition and analysis and writing of the manuscript. GR participated in protocol development, data acquisition and analysis and writing of the manuscript. GKU participated in data acquisition and writing of the manuscript. PG participated in data analysis and interpretation and critical revision of the manuscript for important intellectual content. BN participated in outcome assessment, data analysis and writing of the manuscript. All authors have read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
==== Refs
Law BJ Carbonell-Estrany X Simoes EA An update on respiratory syncytial virus epidemiology: a developed country perspective Respir Med 2002 96 Suppl B S1 S7 11996399
Glezen WP Greenberg SB Atmar RL Piedra PA Couch RB Impact of respiratory virus infections on persons with chronic underlying conditions JAMA 2000 283 499 505 10659876 10.1001/jama.283.4.499
Welliver RC Review of epidemiology and clinical risk factors for severe respiratory syncytial virus (RSV) infection J Pediatr 2003 143 S112 S117 14615709 10.1067/S0022-3476(03)00508-0
Carlsen KH Larsen S Orstavik I Acute bronchiolitis in infancy. The relationship to later recurrent obstructive airways disease Eur J Respir Dis 1987 70 86 92 3675726
Kotaniemi-Syrjanen A Reijonen TM Korhonen K Korppi M Wheezing requiring hospitalization in early childhood: predictive factors for asthma in a six-year follow-up Pediatr Allergy Immunol 2002 13 418 425 12485317 10.1034/j.1399-3038.2002.02091.x
Sigurs N Gustafsson PM Bjarnason R Lundberg F Schmidt S Sigurbergsson F Kjellman B Severe respiratory syncytial virus bronchiolitis in infancy and asthma and allergy at age 13 Am J Respir Crit Care Med 2005 171 137 141 15516534 10.1164/rccm.200406-730OC
Martinez FD Respiratory syncytial virus bronchiolitis and the pathogenesis of childhood asthma Pediatr Infect Dis J 2003 22 S76 S82 12671456 10.1097/00006454-200302001-00011
Openshaw PJ Dean GS Culley FJ Links between respiratory syncytial virus bronchiolitis and childhood asthma: clinical and research approaches Pediatr Infect Dis J 2003 22 S58 S64 12671454 10.1097/00006454-200302001-00009
Piedimonte G Neural mechanisms of respiratory syncytial virus-induced inflammation and prevention of respiratory syncytial virus sequelae Am J Respir Crit Care Med 2001 163 S18 S21 11254547
Freymuth F Vabret A Gouarin S Petitjean J Campet M [Epidemiology of respiratory virus infections] Allerg Immunol (Paris) 2001 33 66 69 11339056
Dollner H Risnes K Radtke A Nordbo SA Outbreak of human metapneumovirus infection in norwegian children Pediatr Infect Dis J 2004 23 436 440 15131467
Williams JV Harris PA Tollefson SJ Halburnt-Rush LL Pingsterhaus JM Edwards KM Wright PF Crowe JEJ Human metapneumovirus and lower respiratory tract disease in otherwise healthy infants and children N Engl J Med 2004 350 443 450 14749452 10.1056/NEJMoa025472
Holter E Abrahamsen TG Rod G Holten E Discrepancy between results of a commercial enzyme immunoassay kit and immunofluorescence staining for detection of respiratory syncytial virus antigen Eur J Clin Microbiol Infect Dis 1998 17 595 596 9796664 10.1007/s100960050140
Wren CG Bate BJ Masters HB Lauer BA Detection of respiratory syncytial virus antigen in nasal washings by Abbott TestPack enzyme immunoassay J Clin Microbiol 1990 28 1395 1397 2199500
Behrendt CE Decker MD Burch DJ Watson PH International variation in the management of infants hospitalized with respiratory syncytial virus. International RSV Study Group Eur J Pediatr 1998 157 215 220 9537488 10.1007/s004310050798
Eriksson M Bennet R Rotzen-Ostlund M von SM Wirgart BZ Population-based rates of severe respiratory syncytial virus infection in children with and without risk factors, and outcome in a tertiary care setting Acta Paediatr 2002 91 593 598 12113331 10.1080/080352502753711740
Sigurs N Bjarnason R Sigurbergsson F Kjellman B Bjorksten B Asthma and immunoglobulin E antibodies after respiratory syncytial virus bronchiolitis: a prospective cohort study with matched controls Pediatrics 1995 95 500 505 7700748
Ruuskanen O Ogra PL Respiratory syncytial virus Curr Probl Pediatr 1993 23 50 79 7681743 10.1016/0045-9380(93)90003-U
SD C The definition of acute respiratory illnesses in children Postgrad Med J 1973 49 771 776 4806395
Abels S Nadal D Stroehle A Bossart W Reliable detection of respiratory syncytial virus infection in children for adequate hospital infection control management J Clin Microbiol 2001 39 3135 3139 11526141 10.1128/JCM.39.9.3135-3139.2001
Fan J Henrickson KJ Savatski LL Rapid simultaneous diagnosis of infections with respiratory syncytial viruses A and B, influenza viruses A and B, and human parainfluenza virus types 1, 2, and 3 by multiplex quantitative reverse transcription-polymerase chain reaction-enzyme hybridization assay (Hexaplex) Clin Infect Dis 1998 26 1397 1402 9636869
Hindiyeh M Hillyard DR Carroll KC Evaluation of the Prodesse Hexaplex multiplex PCR assay for direct detection of seven respiratory viruses in clinical specimens Am J Clin Pathol 2001 116 218 224 11488068 10.1309/F1R7-XD6T-RN09-1U6L
Kehl SC Henrickson KJ Hua W Fan J Evaluation of the Hexaplex assay for detection of respiratory viruses in children J Clin Microbiol 2001 39 1696 1701 11325976 10.1128/JCM.39.5.1696-1701.2001
Kneyber MCJ Steyerberg EW de GR Moll HA Long-term effects of respiratory syncytial virus (RSV) bronchiolitis in infants and young children: a quantitative review Acta Paediatr 2000 89 654 660 10914957 10.1080/080352500750043945
Wennergren G Kristjansson S Relationship between respiratory syncytial virus bronchiolitis and future obstructive airway diseases Eur Respir J 2001 18 1044 1058 11829086 10.1183/09031936.01.00254101
Garofalo RP Hintz KH Hill V Patti J Ogra PL Welliver RCS A comparison of epidemiologic and immunologic features of bronchiolitis caused by influenza virus and respiratory syncytial virus J Med Virol 2005 75 282 289 15602730 10.1002/jmv.20268
Semple MG Cowell A Dove W Greensill J McNamara PS Halfhide C Shears P Smyth RL Hart CA Dual infection of infants by human metapneumovirus and human respiratory syncytial virus is strongly associated with severe bronchiolitis J Infect Dis 2005 191 382 386 15633097 10.1086/426457
Montalbano MM Lemanske RFJ Infections and asthma in children Curr Opin Pediatr 2002 14 334 337 12011675 10.1097/00008480-200206000-00010
Welliver RC Respiratory syncytial virus and other respiratory viruses Pediatr Infect Dis J 2003 22 S6 10 12671447 10.1097/00006454-200302001-00002
Stein RT Holberg CJ Sherrill D Wright AL Morgan WJ Taussig L Martinez FD Influence of parental smoking on respiratory symptoms during the first decade of life: the Tucson Children's Respiratory Study Am J Epidemiol 1999 149 1030 1037 10355379
Burke W Fesinmeyer M Reed K Hampson L Carlsten C Family history as a predictor of asthma risk Am J Prev Med 2003 24 160 169 12568822 10.1016/S0749-3797(02)00589-5
Noble V Murray M Webb MS Alexander J Swarbrick AS Milner AD Respiratory status and allergy nine to 10 years after acute bronchiolitis Arch Dis Child 1997 76 315 319 9166022
Lodrup Carlsen KC Carlsen KH Nafstad P Bakketeig L Perinatal risk factors for recurrent wheeze in early life Pediatr Allergy Immunol 1999 10 89 95 10478609 10.1034/j.1399-3038.1999.00028.x
Murray CS Pipis SD McArdle EC Lowe LA Custovic A Woodcock A Lung function at one month of age as a risk factor for infant respiratory symptoms in a high risk population Thorax 2002 57 388 392 11978912 10.1136/thorax.57.5.388
Fjaerli HO Farstad T Bratlid D Hospitalisations for respiratory syncytial virus bronchiolitis in Akershus, Norway, 1993-2000: a population-based retrospective study BMC Pediatr 2004 4 25 15606912 10.1186/1471-2431-4-25
|
16109158
|
PMC1199604
|
CC BY
|
2021-01-04 16:31:06
|
no
|
BMC Pediatr. 2005 Aug 18; 5:31
|
utf-8
|
BMC Pediatr
| 2,005 |
10.1186/1471-2431-5-31
|
oa_comm
|
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-841610217710.1186/1471-2458-5-84Research ArticleThe impact of changes to heroin supply on blood-borne virus notifications and injecting related harms in New South Wales, Australia Day Carolyn [email protected] Louisa [email protected] Stuart [email protected] Wayne [email protected] National Centre in HIV Epidemiology and Clinical Research, University of New South Wales Level 2, 376 Victoria Street, Darlinghurst, NSW 2010, Australia2 National Drug and Alcohol Research Centre University of New South Wales, Sydney, NSW 2052, Australia3 Office of Public Policy and Ethics Institute for Molecular Bioscience, University of Queensland, St Lucia, Brisbane, 4072, Australia2005 16 8 2005 5 84 84 2 3 2005 16 8 2005 Copyright © 2005 Day et al; licensee BioMed Central Ltd.2005Day et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
In early 2001 Australia experienced a sudden and unexpected disruption to heroin availability, know as the 'heroin shortage'. This 'shortage has been linked to a decrease in needle and syringe output and therefore possibly a reduction in injecting drug use. We aimed to examine changes, if any, in blood-borne viral infections and presentations for injecting related problems related to injecting drug use following the reduction heroin availability in Australia, in the context of widespread harm reduction measures.
Methods
Time series analysis of State level databases on HIV, hepatitis B, hepatitis C notifications and hospital and emergency department data. Examination of changes in HIV, hepatitis B, hepatitis C notifications and hospital and emergency department admissions for injection-related problems following the onset of the heroin shortage; non-parametric curve-fitting of number of hepatitis C notifications among those aged 15–19 years.
Results
There were no changes observed in hospital visits for injection-related problems. There was no change related to the onset heroin shortage in the number of hepatitis C notifications among persons aged 15–19 years, but HCV notifications have subsequently decreased in this group. No change occurred in HIV and hepatitis B notifications.
Conclusion
A marked reduction in heroin supply resulted in no increase in injection-related harm at the community level. However, a delayed decrease in HCV notifications among young people may be related. These changes occurred in a setting with widespread, publicly funded harm reduction initiatives.
==== Body
Background
Injecting drug use is an important risk factor for the transmission of blood-borne viral infections (BBVI) such as the human immunodeficiency virus (HIV) and the hepatitis C virus (HCV) [1]. Harm reduction strategies such as needle and syringe programs have been instrumental in maintaining low HIV prevalence among injecting drug users (IDU) in settings where they were implemented early [2], such as Australia [3,4]. Reducing the prevalence and incidence of HCV, however, has proved more challenging with continuing high prevalence among injecting drug users in Australia [5,6] and elsewhere [7,8] and high incidence among young injectors [9]. IDU are also subject to increased risk of other injection-related problems such as abscesses and thromboses [10,11].
Caulkins has suggested that illicit drug-related harms are influenced by drug price [12]. A significant relationship was found between drug prices, and mentions of cocaine and heroin in United States Emergency Departments, whereby admissions increased as drug prices decreased [12]. These findings raise the question: Would injecting-related harms including BBVI decrease if drug prices increased and availability decreased [13]? The answer to this question, until now, has been based upon theory rather than evidence, because sharp reductions in drug supply have rarely occurred. In this paper, we take advantage of recent changes in the Australian heroin market to answer this question empirically.
In early 2001, there were consistent reports of a dramatic decline in the availability of heroin in New South Wales (NSW) [14,15], where previously heroin had been readily available at historically low price and high purity [16]. Australia's Illicit Drug Reporting System revealed a reduction in heroin supply across Australia, reflected by reduced purity and availability, and a marked increase in price [17]. This 'heroin shortage' was most severe from January to April 2001 [18]. The market began to stabilise after this date but heroin availability has not returned to pre-shortage levels, and the heroin market in Australia appears to have been altered in structure [18].
The reduced heroin supply resulted in at least some former primary heroin users substituting a range of other drug classes for heroin during the shortage, in particular cocaine [15,19,20]. Changes in the heroin market was also associated with a significant decrease in the distribution of needles and syringes, a proxy measure for injecting drug use [21].
This change is important because increased prevalence of cocaine injection has been associated with a number of adverse consequences. The HIV outbreak during the mid 1990s in Vancouver has been attributed in part to increased cocaine injection [22]. It has since been found to be an independent predictor of HIV infection [23] and related to high levels of Emergency Department use for drug related problems, most notably soft tissue infection [24]. Given that heroin use has driven the hepatitis C epidemic in Australia, it is unclear what impact changes in patterns of drug use will have on the epidemic [25].
An increase in benzodiazepine injection was also reported to occur at the time of the shortage [19]. Benzodiazepine injection is associated with a range of harms, including soft tissue damage and increased BBVI risk [26,27]. Harms such as these were reported in qualitative accounts of the shortage by both IDU and key informants.
These changes in drug availability and patterns of injecting drug use occurred in a setting in which harm reduction strategies were widely used. Publicly funded NSPs were introduced to Australia in 1987 [28], and methadone maintenance programs which were established in the 1970s were significantly expanded in 1985 and again in 1999 [29]. More recently, a Medically Supervised Injecting Centre was trialled in one of the key NSW drug markets [30]. Although needle and syringe sharing continues to be reported in Australia, the available evidence suggests that the prevalence of this behaviour has remained stable at 15–20% sharing in the past month [31].
Given these changes in drug use in an established harm reduction setting, this study aimed to examine population level changes in harms associated with injecting drug use, such as HCV, HIV and hepatitis B (HBV) that occurred after the change in heroin supply in NSW, the State containing Australia's largest heroin markets [32]. Specifically, the study examined:
1. Changes, if any, in the number of notifications of BBVI cases including the hepatitis B virus (HBV), HCV and HIV. In particular trends in HCV notifications among young persons, who are at greatest risk of infection [9,33] and among whom prevalence provides a reasonable approximation of incidence were examined; and
2. Changes, if any, in ED and hospital data on presentations for injection-related problems (possibly as a result of the increased use of other drugs such as cocaine and benzodiazepines).
Methods
Data
New cases of HCV, HBV and HIV must be notified by either doctor or laboratory to the State Health Department where they are recorded in the Notifiable Diseases Database and the HIV/AIDS database. Notification of these infections is mandatory in Australia, thus all laboratories are legally obligated to notify all positive results. Notification includes date of birth and detailed name codes, thereby reducing duplicate notifications. De-identified data were retrieved from these databases on the number of individuals diagnosed by age and gender. Monthly data by date of diagnosis on diagnoses were available from 1997.
Although not all HBV and HIV notifications are due to injecting drug use (indeed in Australia, very few HIV infections are likely to be [3]), this is true across all time points examined, and there is no extraneous reason to believe that this pattern would have changed over the period examined. In the case of HCV, more than 90% of such cases occur through injecting drug use [34], thus any changes in these data are very probably directly related to injecting drug use.
Due to the often asymptomatic nature of HCV, not all new diagnoses represent acute or newly acquired infections. Given this, notifications among young IDU are considered those most likely to reflect new hepatitis C infections, or incident cases. Large proportions of IDU (>60%) report recent HCV testing [31]. Thus data on HCV notification among young people was analysed separately.
Diagnostic information on persons presenting to the NSW hospital EDs is recorded at the time of presentation and is coded using the International Classification of Diseases, 9th Revision (ICD_9). The NSW Emergency Department Collection is a database of information collected from approximately a third of NSW EDs. Because these are mostly the larger EDs in NSW, the Collection represents approximately two thirds of all NSW emergency patients. The following codes were used to examine injection-related problems: unspecified septicaemia (038.9); acute and subacute endocarditis (421.0); other peripheral vascular disease (443.9); phlebitis and thrombophlebitis (451.0); other venous embolism and thrombosis (458.8); cellulitis and abscess (681.00, 681.10, 682.0–9); other local infections of skin and subcutaneous tissue (686.9) and chronic ulcer of skin (707.1, 707.9).
Data analysis
Ideally time series of the sort presented here would be analysed using intervention time series models, which allow estimation of the effect of posited interventions in the series after adjusting for the relationships which exist between different values over time in many time series data (serial dependence). For HCV notifications amongst 15–19 year olds, the intervention model time series was not appropriate because the series showed different behaviour in different time periods, so a loess smoother [35] was fitted to this data series using S-Plus 6.1. Such a smoother fits a line to the data using a series of linear regressions fitted in a small neighbourhood of points. For this data the size of this neighbourhood (the span) was estimated using Generalised Cross Validation [36]. Such methods enable the shape of the modelled data to be estimated based only on the data, and avoid subjective decisions about details of the plot, such as the point at which periods of different behaviour occur (changes in slope) or the precise point of onset of sudden changes in level (steps) [36]. This enabled the researchers to avoid making, for example, subjective judgements about exactly when HCV notifications began to decline. Other time series had small counts and no formal time series analysis was conducted – these series were inspected for signs of large-scale changes at the point of the heroin shortage.
Results
Blood borne viral infections (BBVIs)
The possibility of changes in BBVIs was examined using data on NSW notifications of HBV, HCV and HIV infection. There were no apparent differences in the overall number of total notifications following the onset of the shortage for HIV, HCV or HBV (Figure 1). However, the number of HCV and HBV notifications for persons aged 15–19 years, the groups most likely to be effected, were examined (Figure 2). Cross-correlation functions indicated that there was no significant relationship between those functions which describe the heroin shortage and the HCV notifications data at plausible positive lags. A loess smoother (span = 0.75) was fitted to the HCV notification data in order to describe the general structure of the series (Figure 3), and indicated that the series peaked several months before the onset of the heroin shortage, with a downward trend in the series after this point.
Figure 1 Number of HIV, Hepatitis B and C notifications, NSW 1997 – 2003. NDD and HIV/AIDS databases, Communicable Diseases Branch, NSW Health Department.
Figure 2 Hepatitis C notifications among 15–19 and 20–24 year olds, NSW 1997–2003. NDD and HIV/AIDS databases, Communicable Diseases Branch, NSW Health Department.
Figure 3 observed and predicted numbers of Hepatitis C Virus notifications in 15–19 year olds, 1997 – 2003. NDD and HIV/AIDS databases, Communicable Diseases Branch, NSW Health Department.
There is no evidence that this downward trend changed at the time of the heroin shortage or that there were transient non-random increases in the series after the shortage. Unfortunately the series was not amenable to ARIMA methods and more informative analysis could not be conducted in the absence of plausible evidence from cross-correlation functions of a shortage effect on the series (as described above).
Injection-related problems
Figure 4 shows the number of ED admissions and hospital separations where injection-related problems were noted, from 1997–2003. No formal time series was conducted due to the small numbers of such admissions noted each month. No noticeable increase occurred following the onset of the shortage.
Figure 4 Number of drug related ED admissions and hospital separations for injection-related problems, NSW 1997 – 2003. Emergency Department Information System, NSW Department of Health; and NSW Department of Health.
Discussion
This study found no increase in notifications for HIV, HCV or HBV following the reduction in heroin supply. Given that only a proportion of HIV and HBV cases are attributable to injecting drug use, this is perhaps not surprising HIV and HBV. However, the clear reduction in HCV cases among those aged 15–19 years, commencing slightly before the onset of the reduction in heroin supply, was consistent with declines in other indicators of heroin-related harm in the community, including a State-wide 28% reduction in needle and syringe distribution [21]; a 63% decrease in ambulance callouts, a 40% decrease in emergency department admissions for non-fatal heroin overdose and a 43% decrease in heroin related deaths [37]; and a 45% decrease in heroin possession/use offences [38]. This suggests that the decrease in HCV notifications in this group is probably related to the extent of heroin injection in the community. There are no alternative explanations for the decrease in notifications, which was not predicted by mathematical models of the hepatitis C epidemic in Australia [39].
The most plausible explanation of the reduction in HCV notifications among young people is that fewer young persons were initiating injecting drug use. This suggests that there has been a reduction in the extent of injecting drug use and hence probably fewer persons, particularly younger persons, at risk of contracting BBVIs [9,40]. If this is the case, any reductions in HCV notification related to the heroin shortage may be subject to a lag, given the period between initiation to injecting and HCV seroconversion. The extent of this lag is unclear, but USA data suggest that many new HCV infections occur within months of initiating injecting [41]. The proportion of HCV positive IDUs injecting for less than three years has varied between 13% and 22% [5,25]. However, given the uncertainly of such a lag, it could not be validly modelled as any intervention term included, other than the date of the heroin shortage, would have been chosen arbitrarily.
Moreover, evidence from the loess smoother showed that the HCV notifications data series among young people had levelled off before the shortage. Evidence from this same smoother showed that it began to decline some time after the shortage. This behaviour of the data series is possibly explained by the shortage.
Despite reports of increased injecting risk, no increase in other BBVI was reported. Any increase in BBVI risk attributable to the shortage would most likely have occurred during the peak period of the shortage, a relatively brief period of approximately four months [18]. The rate of HIV infection among IDU in Australia is probably too low to produce dramatic increases in the number of HIV infections in that time, even if the risk of infection was increased [42]. Moreover both HIV and HBV are more easily and, in Australia, more commonly sexually transmitted and would therefore be less responsive to any changes in injecting drug use.
It is important to note that these changes occurred in a setting in which harm reduction measures were readily available to IDU. It is unclear whether similar trends would be observed in countries where the prevalence of HIV among IDU is higher, and where access to harm reduction measures is more limited. Estimates suggest that IDU in Australia would need to share needles/syringes with 31 others over a year to increase the prevalence of HIV among IDU in Australia [42]. In contexts where new injecting equipment is not as readily available as in Australia, or where the use of cocaine is much more widespread, changes in heroin availability may have a different effect.
There was also no evidence of increased injection-related harm, as measured by changes to the number of presentations at hospitals for injecting-related problems. This was despite concern about injection-related problems related to frenetic or risky injecting of benzodiazepines and cocaine by some IDU [43,44]. It seems most likely that either these changes occurred among a relatively small group, and/or that the resulting problems were relatively minor and treated by general practitioners or other primary health care services. The number of these infections attributable to injecting drug use is also likely to be small. However, a Canadian study found IDU have a high level of emergency department utilisation, predominantly for injection related problems, especially soft-tissue infection, and this is most common among cocaine users and frequent injectors [24].
Limitations
This study has relied on secondary data sources as indirect measures of the complications of injecting drug use. As changes at the population level were examined, notifications were the most appropriate data. As notification of HIV, HBV and HCV is mandatory in Australia and cases are typically reported by the testing laboratory, these data are unlikely to be subject to the reporting bias reported by others [e.g. [45]]. In Australia, more than 90% of HCV infections are related to injecting drug use [31], and infections among 15–19 year olds are most likely to be newly acquired. Nonetheless, the clear decrease in HCV infection among young people is not easily explained by any other hypotheses, and is consistent with other research on the consequences of the heroin shortage [46].
Several of the conclusions presented here are based on visual inspection of data series with small numbers, rather than on statistical analysis. The authors do not believe that observable effects in these series would be considered of public health importance even if statistical significance could be shown, and consider visual inspection suffices in these instances.
Conclusion
This research has found that following a reduction in heroin supply there was no change in HIV or HBV at the population level, but there was an apparent decrease in HCV infections among those aged 15–19 years. These changes occurred in a setting with widespread, publicly funded harm reduction initiatives.
Abbreviations
ARIMA Auto Regressive Integrated Moving Average
BBVI Blood-borne viral infections
ED Emergency department
HCV Hepatitis C virus
HBV Hepatitis B virus
HIV Human immunodeficiency virus
IDU Injecting drug users
NSP Needle and syringe program
NSW New South Wales
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
All authors contributed to the paper. C.Day, L.Degenhardt and W.Hall conceived the study. L.Degehnardt supervised the research. S.Gilmour led the analysis and C.Day led the writing. All authors helped to conceptualise ideas, interpret the findings and reviewed drafts of the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This work was completed as part of a larger project funded by the Australian National Drug Law Enforcement Research Fund (NDLERF). The National Drug and Alcohol Research Centre and the National Centre in HIV Epidemiology and Clinical Research are funded by the Australian Department of Health and Ageing. We thank the agencies and individuals who provided advice and data for inclusion in the study: Libby Topp, Linette Collins, Amy Gibson and Elizabeth Conroy for assistance with the project; Owen Westcott and Jenny Iverson from the AIDS and Infectious Diseases Branch of NSW Health; and Greg Dore and Margaret MacDonald for comments on an earlier draft.
==== Refs
MacDonald M Crofts N Kaldor J Transmission of hepatitis C virus Epidemiologic Reviews 1996 18 137 148 9021308
Des Jarlais D Hagan H Friedman SR Friedman P Goldberg D Frischer M Green S Tunving K Ljungberg B Wodak A Ross M Purchase D Millson ME Myers T Maintaining low HIV seroprevalence in populations of injecting drug users Journal of the American Medical Association 1995 274 1226 1231 7563513 10.1001/jama.274.15.1226
MacDonald M Wodak AD Ali R Crofts N Cunningham PH Dolan KA Kelaher M Loxley WM van Beek I Kaldor JM HIV prevalence and risk behaviour in needle exchange attenders: a national study Medical Journal of Australia 1997 166 237 240 9076266
MacDonald M Law M Kaldor J Hales J Dore GJ Effectiveness of needle and syringe programmes for preventing HIV transmission International Journal of Drug Policy 2003 14 353 357 10.1016/S0955-3959(03)00133-6
MacDonald MA Wodak AD Dolan KA van Beek I Cunningham PH Kaldor JM Hepatitis C virus antibody prevalence among injecting drug users at selected needle and syringe programs in Australia, 1995-1997 Medical Journal of Australia 2000 172 57 61 10738473
Crofts N Aitken C Incidence of bloodborne virus infection and risk behaviours in a cohort of injecting drug users in Victoria, 1990-1995 Medical Journal of Australia 1997 167 17 20 9236754
Mathei C Buntinx F van Damme P Seroprevalence of hepatitis C markers among intravenous drug users in western European countries: a systematic review Journal of Viral Hepatitis 2002 9 157 173 12010503 10.1046/j.1365-2893.2002.00339.x
Ray Kim W Global epidemiology and burden of hepatitis C Microbes Infect 2002 4 1219 1225 12467763 10.1016/S1286-4579(02)01649-0
van Beek I Dwyer R Dore GJ Luo K Kaldor JM Infection with HIV and hepatitis C virus among injecting drug users in a prevention setting: retrospective cohort study British Medical Journal 1998 317 433 437 9703523
Haverkos HW Langer R Serious infections other than human immunodeficiency virus among intravenous drug users Journal of Infectious Diseases 1990 161 894 902 2182730
Binswanger IA Kral AH Bluthenthal RN Rybold DJ Edlin ER High prevalence of abscesses and cellulitis among community recruited injection drug users in San Francisco Clinical Infectious Diseases 2000 30 579 581 10722447 10.1086/313703
Caulkins JP Drug prices and emergency department mentions for cocaine and heroin American Journal of Public Health 2001 91 1446 1448 11527779
Hagan H Public health and changes in illicit drug prices American Journal of Public Health 2001 91 1350 11527752
Day C Topp L Rouen D Darke S Hall W Dolan K Decreased heroin availability in Sydney in early 2001 Addiction 2003 98 93 95 12492759 10.1046/j.1360-0443.2003.00266.x
Weatherburn D Jones C Freeman K Makkai T Supply control and harm reduction: Lessons from the Australian heroin 'drought' Addiction 2003 98 83 91 12492758 10.1046/j.1360-0443.2003.00248.x
Darke S Topp L Kaye S Hall W Heroin use in New South Wales, Australia, 1996-2000: 5 year monitoring of trends in price, purity, availability and use from the Illicit Drug Reporting System (IDRS) Addiction 2002 97 179 186 11860389 10.1046/j.1360-0443.2002.00032.x
Topp L Kaye S Bruno R Longo M Williams P O'Reilly B Fry C Rose G Darke S Australian Drug Trends 2001: Findings from the Illicit Drug Reporting System (IDRS) 2002 Sydney, National Drug and Alcohol Research Centre, University of New South Wales
Day C Degenhardt L and Day C Documenting the heroin shortage in New South Wales The Course and Consequence of the Heroin Shortage in New South Wales 2004 Adelaide, Australasian Centre for Policing Research 11 17
Topp L Day C Degenhardt L Changes in patterns of drug injection concurrent with a sustained reduction in the availability of heroin in Australia Drug and Alcohol Dependence 2003 70 275 286 12757965 10.1016/S0376-8716(03)00013-9
Roxburgh A Degenhardt L Breen C Changes in patterns of drug use among injecting drug users following a reduction in the availability of heroin in New South Wales, Australia Drug and Alcohol Review 2004 23 287 294 15370008 10.1080/09595230412331289446
Day C Degenhardt L Gilmour S Hall W Effects of supply reduction upon injecting drug use British Medical Journal 2004 329 428 429 15292056
Strathdee S Patrick DM Currie S Cornelisse PGA Rekart ML Montaner JSG Schechter MT O'Shaughnessy MV Needle exchange is not enough: Lessons from the Vancouver injection drug use study AIDS 1997 11 F59 65 9223727 10.1097/00002030-199708000-00001
Tyndall MW Currie S Spittal P Li K Wood E O'Shaughnessy MV Schechter MT Intensive injection cocaine use as the primary risk factor in the Vancouver HIV-1 epidemic AIDS 2003 17 887 893 12660536 10.1097/00002030-200304110-00014
Palepu A Tyndall M Leon H Muller J O'Shaughnessy MV Schechter MT Anis AH Hospital utilization and costs in a cohort of injection drug users Canadian Medical Association Journal 2001 165 415 420 11531049
Dore G Law M MacDonald M Kaldor JM Epidemiology of hepatitis C virus infection in Australia Journal of Clinical Virology 2003 26 171 184 12600649 10.1016/S1386-6532(02)00116-6
Aitken CK Higgs P Severe vein damage caused by Tamezepam injecting (letter) Australian and New Zealand Journal of Public Health 2002 26 79 11895033
Ross J Darke S The nature of benzodiazepine dependence among heroin users in Sydney, Australia Addiction 2000 95 1785 1793 11177494 10.1046/j.1360-0443.2000.951217858.x
Feacham RGA Valuing the past …Investing in the future: Evaluation of the National HIV/AIDS Strategy 1993-94 to 1995-96. 1995 Canberra, Australian Government Publishing Service
Mattick RP Hall W A treatment outline for approaches to opioid dependence 1993 Canberra, Australian Government Publishing Service
MSIC Evaluation Committee Final Report of the Evaluation of the Sydney Medically Supervised Injecting Centre 2003 Sydney,
National Centre in HIV Epidemiology and Clinical Research HIV/AIDS, viral hepatitis and sexually transmissible infections in Australia: Annual Surveillance Report 2003 2003 Sydney, National Centre in HIV Epidemiology and Clinical Research, University of New South Wales
Hall WD Ross JE Lynskey MT Law MG Degenhardt LJ How many dependent heroin users are there in Australia? Medical Journal of Australia 2000 173 528 531 11194736
Diaz T Des Jarlais DC Vlahov D Perlis TE Edwards V Friedman SR Rockwell R Hoover D Williams IT Monterroso ER Factors associated with prevalent hepatitis C: differences among young adult injection drug users in lower and upper Manhattan, New York City American Journal of Public Health 2001 91 23 30 11189819
Robotin MC Copland J Tallis G Coleman D Giele C Carter L Spencer J Kaldor JM Dore GJ Surveillance for newly acquired hepatitis C in Australia J Gastroenterol Hepatol 2004 19 283 288 14748875 10.1111/j.1440-1746.2003.03270.x
Cleveland WS Devlin SJ Locally-weighted fitting: An approach to fitting analysis by local fitting Journal of the American Statistical Association 1988 83 596 610
Craven P Wahba G Smoothing Noisy Data with Spline Functions Numerical Mathematics 1979 31 377 3403 10.1007/BF01404567
Degenhardt L Conroy E Gilmour S The effect of a reduction in heroin supply upon population trends in fatal and non-fatal drug overdoses Medical Journal of Australia 2005 182 20 23 15651943
Degenhardt L Conroy E Gilmour S Collins L The effect of a reduction in heroin supply in Australia upon drug distribution and acquisitive crime British Journal of Criminology 2005 45 2 24 10.1093/bjc/azh096
Law M Dore G Bath N Thompson S Crofts N Dolan K Giles W Gow P Kaldor J Loveday S Powell E Spencer J Wodak A Modelling hepatitis C virus incidence, prevalence and long-term sequelae in Australia, 2001 International Journal of Epidemiology 2003 32 717 724 14559738 10.1093/ije/dyg101
Zhou J MacDonald MA Wodak A Kaldor JM Collaboration of Australian NSPs Hepatitis C antibody prevalence and risk behaviours among young injecting drug users at sentinel needle and syringe programs in Australia, 1995-2001: October; Sydney, Australia. 2002 40
Garfein RS Vlahov D Galai N Doherty MC Nelson KE Viral infection in short-term injection drug users: The prevalence of the Hepatitis C, Hepatitis B, immunodeficiency and Human T-Lymphotropic Viruses American Journal of Public Health 1996 86 655 661 8629715
Murray JM Law MG Gao Z Kaldor JM The impact of behavioural changes on the prevalence of human immunodeficiency virus and hepatitis C among injecting drug uses International Journal of Epidemiology 2003 32 708 714 14559736 10.1093/ije/dyg102
Maher L Don't leave us this way: ethnography and injecting drug use in the age of AIDS International Journal of Drug Policy 2002 13 311 325 10.1016/S0955-3959(02)00118-4
Degenhardt L Day C Collins L Gibson A Degenhardt L and Day C Changes in injecting drug use The Course and Consequence of the Heroin Shortage in New South Wales 2004 Adelaide, Australasian Centre for Policing Research 31 43
Hagan H Snyder N Hough E Yu T McKeirnan S Boase J Duchin J Case-reporting of acute hepatitis B and C among injection drug users Journal of Urban Health 2002 79 579 585 12468677
Degenhardt L Day C Dietze P Pointer S Conroy E Collins L Hall W The consequences of the heroin shortage in three Australian States Addiction 2005 100 908 920 15954999 10.1111/j.1360-0443.2005.01094.x
|
16102177
|
PMC1199605
|
CC BY
|
2021-01-04 16:28:58
|
no
|
BMC Public Health. 2005 Aug 16; 5:84
|
utf-8
|
BMC Public Health
| 2,005 |
10.1186/1471-2458-5-84
|
oa_comm
|
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-881611150210.1186/1471-2458-5-88Research ArticleEffectiveness of different methods of health education: A comparative assessment in a scientific conference Saha Asim [email protected] Era [email protected] Minal [email protected] Occupational Medicine Division, National Institute of Occupational Health, Ahmedabad, India2 Ergonomics and Work Physiology Division, National Institute of Occupational Health, Ahmedabad, India3 Reproductive Toxicology Division, National Institute of Occupational Health, Ahmedabad, India2005 22 8 2005 5 88 88 25 3 2005 22 8 2005 Copyright © 2005 Saha et al; licensee BioMed Central Ltd.2005Saha et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Abstract
Background
Every individual mode of health education has its own merits, drawbacks as well as their own sphere of effectiveness. A specific mode of communication is more useful in a specific setting on a specific group than others. To search for optimum mode of communication for a specific audience is a major area of research in health education. The issue of imparting health education to a gathering of educated people, representing different fields of knowledge has remained a relatively less lighted aspect of health education research. In this backdrop this study was initiated for making a comparative assessment of different methods of dissemination of health education among educated people.
Methods
A cross-sectional interviewer administered questionnaire survey was conducted involving 142 randomly selected subjects during the last session of a five-day conference having health as main theme when the opinion of the delegates regarding different communication methods was asked for. Collected data was analyzed not only to find out the optimum mode of education dissemination in such a setting but also to find the contribution of different factors in the preferences of the study subjects.
Results
The participants opted more (60%) for focused programs of smaller audience (sectional program). In both broad area (main program) and focused area programs (sectional), the participants preferred lectures (62% and 65.7% respectively). Specific topics were preferred both in lectures (67.6%) and symposia (57.7%). In the exhibition, exhibits seemed to be more attractive (62%) than the posters. Qualification has emerged to be a contributing factor in peoples' choice towards sectional programme and also in their affinity to symposia. Increased age was a significant contributor in participants' preference towards specific topics. Physical barriers of communication appeared to be a problem in the main program as well as in the exhibition. Lack of coherence among the speakers was reported (69%) to be a major reason for which symposia was not preferred.
Conclusion
This study concluded that while planning for health education dissemination in an educated group a focused programme should be formulated in small groups preferably in the form of lectures on specific topics, more so while dealing with participants of higher age group having higher educational qualification.
==== Body
Background
Health education is a process by which individuals and groups of people learn to behave in a manner conducive to the promotion, maintenance or restoration of health [1]. Communication in relation to health education involves different modes like lectures, group or panel discussions, symposia, poster or exhibit presentation etc. Every individual mode of health education has its own merits, drawbacks as well as their own sphere of effectiveness. In addition it has to overcome the barriers of communication (e.g. physiological, psychological, environmental and cultural). Research on the effectiveness of different modes of health education dissemination is already in progress to examine the utility of a specific mode of communication in a specific setting [2,3] on a specific group [4]. It has been observed that different educational methods may be specially suitable for different groups of people depending upon their age, sex, educational qualification, background and nature of job [5]. Comparative assessment of effectiveness of different educational methods has also been done on some target groups in different communicational settings [6].
Imparting health education to an educated group is a special arena of interest because of the fact that this educated group may have a major role in the propagation of the achieved knowledge in future. This why communication of health education in a gathering of educated people (e.g. conferences) should have separate specifications in relation to its content and mode of communication. Naturally, this becomes an arena and area of special interest and not much of research is undertaken in this aspect till date. In this backdrop this cross-sectional study was initiated during a scientific conference for making a comparative assessment of different methods of dissemination of health education among educated people.
Methods
This study was conducted in a scientific conference where 2250 scientists from different branches of science gathered. It was a mega event having health as the theme and experts of various fields attended this conference from different parts of the globe. This conference was organized by one of the scientific bodies of the country and this conference aimed at disseminating health related issues among the scientists, science managers, policy makers, students and general public. This conference is an annual event (largest scientific gathering of the country), which undertakes an issue every year as the theme and communicates the messages on the theme. This activity being the oldest of the country also is well known for its impact on building awareness and opinion among the scientific community as well as general public. In this way it has not only generated scientific movement in the past involving common mass also but also many times it has substantially influenced policy making. Various aspects of health promotion, health technology, implication of health in nation's development etc. were discussed and three modes of education dissemination were used; lecture, symposium and exhibition. The conference activity had two divisions; main program and sectional program (fourteen sections were there). Main programme consisted of deliberation containing discussion on different aspects of science and health, addressing the conference participants at large whereas sectional programmes dealt with a section of the attendees and focused only the issues related to the specific section. Main program consisted of lectures, symposia and an exhibition whereas the sectional program consisted of lectures and symposia. In the main program, lectures and symposia were of two types; some were based on specific topics and some were based on relatively broader topics. So far as the lectures are concerned, specific topic lectures included presentation like "cholera-epidemiology, genetics and vaccine development", "role of a tool box of diagnosis for tuberculosis in endemic country", "disease elimination: the kala-azar experience" etc. and broader topic lectures were like "health science and our future", "role of public health in national economics" etc. In case of symposia, the specific topic symposia dealt with the issues like "challenges in combating malaria", "high altitude dysfunction" etc. and the broader topic symposia were on the topics like "environment & health", "bridging the gap between health science and society". Each of these symposia consisted of three or more deliberations from different speakers talking on different aspects of the topic. For example the symposium on "combating malaria" contained topics like "control of malaria in mosquito vector", "current status and strategies for old and new drugs for treatment, prevention and control", "prospects of vaccine", "cost and benefit of malaria control", and "malaria research, development and control strategies". Topics of lectures and symposia of sectional programs were specific to the concerned section. For example, lectures of medical science section consisted of lectures like "factors other than iodine deficiency in endemic goiter", "mechanism of action of enterotoxin of vibrio cholerae", etc. and lectures of environmental science section contained lectures like "biomonitoring of health effects of urban air pollution", "arsenic exposure and effects on liver", etc. Similarly, symposia of medical science section had topic like "development of ergonomics in India" and symposia of environmental science included topics like "environmental endocrine disruptors and reproductive health". These symposia again consisted of different speakers' deliberations on various aspects of the topic of the symposia. The exhibition contained two types of materials: posters and exhibits. Posters were prepared on the topics like "prevention of dust related diseases", "how to combat diarrhoeal diseases?" etc. Exhibits were models/instruments, which were displayed and demonstrated for easy conveyance of the related messages. Exhibits contained "spirometer – an instrument early diagnosis of morbidity related to dust related diseases", "model showing transmission of malaria from mosquito vector to human host" etc.
This cross-sectional interviewer administered questionnaire survey was conducted during the last session of this five-day conference when the opinion of the delegates regarding different programs was asked for. Necessary ethical clearance was obtained from the institutional ethics committee of National Institute of Occupational Health, India for the purpose of this study. While calculating the sample size for this study we presumed the lowest choice prevalence to be 10% (as there was no available literature of this nature) and accordingly we calculated the sample size for prevalence study using acceptable range 5–15%. Thus the minimum sample size for 5% level of significance was calculated as 130. We set our target as 150 subjects. Selection of subjects was done by using random numbers generated by Microsoft Excel Software. Initially 3 sections (out of 14 sections) were selected randomly and 50 participants from each section were approached for the study. Of the 150 persons approached for study, 142 agreed to participate.
All the participants were enquired about their choices in relation to all the different aspects of the conference. Analysis of the collected information was undertaken using SPSS release 6.1.4 software. Along with descriptive analysis of the data, univariate analysis was done initially. Afterwards logistic regression technique was applied to obtain contribution of different factors in the choices of the participants. As we intended to identify the most suitable mode of communication for each division/section of this conference (e.g. lectures, symposia), it was essential to ensure that the findings should be on the basis of merits/demerits of the mode of communication only. For this reason, while going for multivariate analysis, our intention was to observe whether the decision of choices made by the study participants was independent of the factors that might affect the choices (e.g. age, qualification, background, presence of physical barriers of communication, coherence among speakers, etc.). Variables like higher qualification (Ph.D/MD or higher), higher designation (Associate Professor or equivalent and above), attending alone or with friends, education background (medical/non-medical) problem in understanding English, origin (urban/rural), noise-congestion-invisibility (absence/presence), coherence among speakers in case of symposia (absent/present) were taken as categorical variables & age was introduced as continuous variable in the logistic regression model. These variables were introduced as covariates in the logistic regression model and the choices of the study participants (e.g. section programme better, lecture, better, specific topic better etc.) were introduced one by one as the outcome variable. In this way the role of the possible interfering factors on each of the choices of the participants could be evaluated. In our analysis we accommodated all variables together in the logistic regression model to obtain the contribution of every individual variable adjusting for the effects of other variables.
Results
Mean age of the study subjects was 33.2 (11.1) years. 67.6% of the subjects were males and 32.4% of the participants were females. 25.4% subjects were more than 40 years of age. 52 (36.6) subjects had higher qualification whereas 44 (31) subjects had higher designation. 16.9% subjects were attending alone whereas rest were along with their friends. Only 16 (11.3) persons had some difficulty in understanding communication in English language. Medical background was found in 8 (5.6) subjects and 20 (14.1) subjects had their origin in rural areas. 54.9% participants reported presence of noise-congestion-invisibility and 69% talked about lack of coherence among the speakers of the symposia.
So far as choice of the participants is concerned, 86 (60.0) subjects opined that sectional programme was better than the main programme. When assessment of main programme was asked for 62% subjects remarked that lectures were best, whereas 29.2% and 13% participants were of the opinion that exhibition and symposium was best. Regarding the sectional programme, it was observed that 65.7% subjects liked lectures rather than symposia. In case of lectures and symposia of main programme, 96 (67.6) and 82 (57.7) subjects respectively liked specific topics better. In the exhibition, exhibits seemed to be more attractive (62%) than the posters (Table-1).
Table 1 Opinion of the study subjects regarding effectiveness of communication.
Assessment area Criteria Number (Percentage) Significance
Main/Sectional Programme Sectional Programme was better 86(60.6) χ2 = 12.68; p < 0.001
Main Programme was better 56(39.4)
Main Programme Lecture was best 85(59.9) χ2 = 70.46; p < 0.001
Symposium was best 17(11.9) -
Exhibition was best 40(28.2) χ2 = 10.78; p < 0.01
Sectional Programme Lecture was better 92(65.7) χ2 = 27.66; p < 0.001
Symposium was better 48(34.3)
Lectures of Main Programme Specific topic was better 96(67.6) χ2 = 35.21; p < 0.001
Broad topic was better 46(32.4)
Symposia of Main Programme Specific topic was better 82(57.7) χ2 = 6.82; p < 0.01
Broad topic was better 60(42.3)
Exhibition Exhibits were better 88(62.0) χ2 = 16.28; p < 0.001
Posters were better 54(38.0)
Table-2 and Table-3, shows the contribution of different factors in determining the choices of the participants. Age of the participants had significant effect in their choices in relation to assessment of lectures and symposia of main programme (multivariate analysis). In case of both lectures and symposia of main programme, significantly positive regression co-efficient showed that specific topic was better for advanced age people. Higher qualification was a significant contributor in preferring sectional programme as such and also in preferring symposia of the sectional programme rather than the lecture (univariate analysis). On multivariate analysis, it was found that higher qualification was a stronger (odds ratio raised from 3.2 to 9.1) contributor for preference of sectional programme. But in case of preference of symposia of sectional programme it became a weaker contributor (though odds ratio increased from 2.9 to 3.7, it became non-significant). On this analysis, higher qualification was also observed to be a significant contributor in case of preference of symposia of main programme. Medical background could not show any significant effect in case of any of the choices except for preference of exhibits (odds ratio was 6.4 in univariate analysis and 19.7 in multivariate analysis) even though the content of all the communications were health related issues. Absence of barriers like noise-congestion-invisibility was a significant contributor (multivariate analysis) while preferring sectional programme as such and also for preference of exhibits. Coherence among the speakers appeared to be the most important factor while assessing symposia of both main and sectional programme (univariate analysis). The significance of this factor increased many folds when the data was subjected to multivariate analysis.
Table 2 Distribution of odds ratio in relation to different contributing factors (univariate analysis)
Covariates Choice of Main/ Sectional Programme (Sectional better) Assessment of Main Programme (Symposium Better) Assessment of Sectional Prog. (Symposium better) Assessment of Exhibition (Exhibit better)
Higher Qualification 3.19 (1.38 – 7.45) 2.5 NS 2.92 (1.27 – 6.82) 1.13 NS
Higher Designation 2.17 NS 1.17 NS 2.77 (1.15 – 6.82) 1.27 NS
Medical Background 0.2 NS _____ 1.9 NS 6.43 (1.09 – 48.98)
Coherence of the Speakers in Symposium NA 32.0 (6.28 – 220.10) 43.0 (13.45 – 146.12) NA
NS = Non Significant, NA = Not Applicable, Figures within parenthesis indicate 95% Confidence Interval.
Covariates like age>40, attending alone, problem of understanding English, urban background, absence of physical barriers like noise-congestion-invisibility did not show any significant impact on the choices of the study participants. Choices like lecture in main programme better, exhibition in main programme better, lecture in sectional programme better, specific topic in main programme symposium better and specific topic in main programme lecture better were independent of all the covariates.
Table 3 Distribution of odds ratio in relation to different contributing factors (multivariate analysis)
Covariates Choice of Main/ Sectional Programme (Sectional better) Assessment of Main Programme Symposium better Assessment of Main Prog. Symposium (Specific topic better) Assessment of Main Prog. Lecture (Specific topic better) Assessment of Sectional Prog. (Symposium better) Assessment of Exhibition (Exhibit better)
Age NS NS p < 0.001, Reg. Co-eff. = 0.1842 p < 0.001, Reg. Co-eff. = 0.1641 NS NS
Higher Qualification 9.1 (1.20–17.01) 40.62 (14.12 – 67.12) 1.57
NS 0.67
NS 3.71
NS 0.65
NS
Medical Background 0.1
NS 0.06
NS 0.32
NS ___ 2.82
NS 19.75 (10.34 – 29.16)
Absence of Noise, Congestion, Invisibility 3.46 (1.09–5.82) 0.02
NS 0.51
NS 0.47
NS 0.57
NS 7.78 (2.36 – 13.21)
Coherence of the Speakers in Symposium NA 105.79 (95.91–115.68) NA NA 308.77 (301.39 – 316.14) NA
NS = Non Significant, NA = Not Applicable, Figures within parenthesis indicate 95% Confidence Interval.
Covariates like higher designation, attending alone, problem of understanding English and urban background did not show any significant impact on the choices of the study subjects. Choices like lecture in main programme better, exhibition in main programme better and lecture in sectional programme better were independent of all the covariates.
Discussion
Sectional programmes were being attended by concerned audience in the form of a relatively smaller group and the topics were specific to the concerned section. This may have been the reason of participants' preference towards sectional programme over main programme (main programme was addressing a broader audience of non-specific nature). In main as well as sectional programmes, lectures were preferred over symposia. This may be due to the fact that educated mass may have liked a comprehensive communication by a single deliverer more than a non-coherent message from multiple communicators (69% of subjects reported that there was poor coherence among the speakers of the symposia). For example a comprehensive lecture on "cholera – epidemiology, genetics and vaccine development" by a single deliverer has been more acceptable and useful than a symposium on "challenges in combating malaria" where different aspects of the topics were dealt with by different experts. This may have been due to the fact that the audience have liked a focused discussion a limited topic rather than a composite message on different aspects of a relatively larger area at a time. Lack of linkage between the speakers may also have been a matter of concern because it hinders the process of comprehensive learning on a larger topic. In case of main programme lecture had more impact than exhibition even. This has probably been a special feature of the educated audience. In spite of the lucidity of the message delivery inherent in exhibition, the study participants have opted more for lectures possibly because of the reason that the lectures contained optimum volume of messages delivered in a more elaborate and systematic manner. The completeness of a topic achieved through a lecture may have been the more attraction than the discreteness of message passed though individual posters or exhibits. Though people from a varied discipline of science were the audience in the main programme, specific topics were better in the lectures as well as in the symposia. This observation has been a salient finding of this study. Specific topics have been preferred everywhere by educated audience over relatively broader topics. People may have found specific in depth knowledge on topics like "cholera" or "malaria" more useful rather than general discussion on relatively broader topics like "environment and health" or "role of public health in national economics". In the exhibition, exhibits have carried more impression than the posters. Participants may have liked hands on experience of operating different exhibits (instruments) more than the message disseminated by the posters.
So far as different possible determining factors of participants' decision are concerned, higher qualification has been a contributing factor in participants' preference towards sectional programme (OR 9.1, 95% CI 1.2–17.0) and also in choosing main programme symposia better (OR 40.6, 95% CI 14.1–67.1). In the preference of symposia of sectional programme also higher qualification played a role. In this case, higher designation also showed some impact (though significant in univariate analysis it was not significant in multivariate analysis). Thus, qualification (in some cases designation also) has emerged to be a decisive factor in peoples' choice towards more specific subject oriented programme (sectional programme) and also in their affinity to symposia. Increased age was a significant contributor in participants' preference towards specific topics. Medical background has helped people only in understanding exhibits. The scientific details may have been easily understandable to such people due to their medical background. Physical barriers of communication (noise-congestion-invisibility) have contributed significantly in subjects' preference towards sectional programme as such and also in the choice of exhibits rather than the posters. This finding points towards the fact that physical barriers of communication play an important role in the success of a health education dissemination programme.
Some of the earlier studies have already stressed the need of exploding the background and character of the recipient group while imparting health education [7]. Some studies have shown the success of different modes of communication in different situations [8-12]. View of different recipient groups are different towards various modes of communication and the success of a health education programme depends on the planning of the structure of such a programme taking care of all the relevant factors [13]. The implication of a well-planned health education programme is far spreading and such a programme has a great potential in changing public attitudes [14,15]. This study also has strengthened the idea of planning the health education programme according to the background and character of participating groups. While addressing the issue of imparting health education to an educated mass, this study has come out with very specific observations. It has showed that health education programme in the form of lectures on specific topics dealing with a small section is more likely to succeed in case of educated audience. This study has pointed out that if an exhibition is planned for such audience, it should contain more and more exhibits rather than posters. Moreover, it has also been observed in this study that higher age has a positive role in participants' choice towards specific topics. Higher qualification has some positive impact in choice towards focused programme involving smaller groups. Importance of basic criteria for the success of a heath education endeavor like comprehensiveness of the content, role of physical barriers of communication and coherence among multiple speakers covering various aspects of a topic has also been highlighted by virtue of this study.
Conclusion
This study has come out with an important but relatively less lighted aspect of health education dissemination. It has addressed some of the important issues in imparting health education to a gathering of educated people, representing different fields of knowledge. On one hand this study has spoken for preference of a well-designed comprehensive lecture rather than a non-coherent symposia while on the other hand it has stressed the need of adoption of specific topics (more so with increasing age of the receptor population). At the end, this study has concluded that while planning for health education dissemination in an educated group a focused programme should be formulated in small groups preferably in the form of lectures on specific topics, more so while dealing with participants of higher age group having higher educational qualification.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
AS: Planned and designed the study, executed the study, analyzed the data and prepared the final write up.
EP: Executed the study, associated with the planning of the study, contributed in write up and also in literature search.
MM: Executed the study, helped in data analysis, contributed in literature search and write up.
Pre-publication history
The pre-publication history for this paper can be accessed here:
==== Refs
Park K Park's Textbook of Preventive and Social Medicine 2000 16 Banarsidas Bhanot Jabalpur 603
Nishtar S Zoka N Nishtar SS Khan SY Jehan S Mirza YA Posters as a tool for disseminating health related information in a developing country: a pilot experience J Pak Med Assoc 2004 54 456 60 15518367
Werner RT Sr Wilson JM Are health education conferences effective? An evaluation of knowledge gain in a three-day institute Health Educ 1981 12 22 4 6115832
Abolfotouh MA The impact of a lecture on AIDS on knowledge, attitudes and beliefs of male school-age adolescents in the Asir Region of southwestern Saudi Arabia J Community Health 1995 20 271 81 7657860
Mahajan BK Gupta MC Textbook of Preventive and Social Medicine 1995 2 Jaypee. New Delhi 562
Rezaei MB Seydi S Alizadeh SM Effects of 2 educational methods on the knowledge, attitude, and practice of women high school teachers in prevention of cervical cancer Cancer Nurs 2004 27 364 9 15525863
Kannapiran C Ganguly I Shiva M Sehgal M Khanna P Bhatia R Health education Health Millions 1992 18 30 4 12343651
Christie TG Educational effectiveness of health exhibits J Sch Health 1970 40 206 9 5198702
Mol ME van Boxtel MP Jolles J Education about dementia. Effectiveness of a public lecture Tijdschr Gerontol Geriatr 2004 35 72 5 15230056
Randolph RF Hudak RL Vaught C Communicating hearing loss information to young children effectiveness of lecture and printed materials AAOHN J 2003 51 433 8 14596383
Greer KA Symposium offers education, empowerment, and entertainment Adv Skin Wound Care 2004 17 495 7
Yaser Y Outputs far exceeding inputs: the first multimedia communication campaign in Turkey Integration 1990 23 32 5 12316325
Abdullahi Fido Raghad Al-Kazemi Effective method of teaching psychiatry to undergraduate medical students: The student perspective Medical Principles and Practice 2000 9 255 259
Tanaka G Ogawa T Inadomi H Kikuchi Y Ohta Y Effects of an educational program on public attitudes towards mental illness Psychiatry Clin Neurosci 2003 57 595 602 14629708 10.1046/j.1440-1819.2003.01173.x
Smith JP The potential of health education J Adv Nurs 1980 5 1 2 6101604
|
16111502
|
PMC1199606
|
CC BY
|
2021-01-04 16:28:58
|
no
|
BMC Public Health. 2005 Aug 22; 5:88
|
utf-8
|
BMC Public Health
| 2,005 |
10.1186/1471-2458-5-88
|
oa_comm
|
==== Front
BMC Plant BiolBMC Plant Biology1471-2229BioMed Central London 1471-2229-5-141609115110.1186/1471-2229-5-14Research ArticleArabidopsis ETO1 specifically interacts with and negatively regulates type 2 1-aminocyclopropane-1-carboxylate synthases Yoshida Hitoshi [email protected] Masayasu [email protected] Koji [email protected] Kevin LC [email protected] Joseph R [email protected] Department of Rice Research, National Agricultural Research Center, Jo-etsu, Niigata 943–0193, Japan2 Department of Low-Temperature Sciences, National Agricultural Research Center for Hokkaido Region, Sapporo, Hokkaido 062–8555, Japan3 Department of Physiology and Quality Science, National Institute of Vegetable and Tea Science, Ano, Mie 514–2392, Japan4 Plant Biology Laboratory, The Salk Institute for Biological Studies, La Jolla, California 92037, U.S.A5 Present address: Institute of Plant and Microbial Biology, Academia Sinica, Nankang, Taipei 115, Taiwan2005 10 8 2005 5 14 14 26 4 2005 10 8 2005 Copyright © 2005 Yoshida et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
In Arabidopsis, ETO1 (ETHYLENE-OVERPRODUCER1) is a negative regulator of ethylene evolution by interacting with AtACS5, an isoform of the rate-limiting enzyme, 1-aminocyclopropane-1-carboxylate synthases (ACC synthase or ACS), in ethylene biosynthetic pathway. ETO1 directly inhibits the enzymatic activity of AtACS5. In addition, a specific interaction between ETO1 and AtCUL3, a constituent of a new type of E3 ubiquitin ligase complex, suggests the molecular mechanism in promoting AtACS5 degradation by the proteasome-dependent pathway. Because orthologous sequences to ETO1 are found in many plant species including tomato, we transformed tomato with Arabidopsis ETO1 to evaluate its ability to suppress ethylene production in tomato fruits.
Results
Transgenic tomato lines that overexpress Arabidopsis ETO1 (ETO1-OE) did not show a significant delay of fruit ripening. So, we performed yeast two-hybrid assays to investigate potential heterologous interaction between ETO1 and three isozymes of ACC synthases from tomato. In the yeast two-hybrid system, ETO1 interacts with LE-ACS3 as well as AtACS5 but not with LE-ACS2 or LE-ACS4, two major isozymes whose gene expression is induced markedly in ripening fruits. According to the classification of ACC synthases, which is based on the C-terminal amino acid sequences, both LE-ACS3 and AtACS5 are categorized as type 2 isozymes and possess a consensus C-terminal sequence. In contrast, LE-ACS2 and LE-ACS4 are type 1 and type 3 isozymes, respectively, both of which do not possess this specific C-terminal sequence. Yeast two-hybrid analysis using chimeric constructs between LE-ACS2 and LE-ACS3 revealed that the type-2-ACS-specific C-terminal tail is required for interaction with ETO1. When treated with auxin to induce LE-ACS3, seedlings of ETO1-OE produced less ethylene than the wild type, despite comparable expression of the LE-ACS3 gene in the wild type.
Conclusion
These results suggest that ETO1 family proteins specifically interact with and negatively regulate type 2 ACC synthases. Our data also show that Arabidopsis ETO1 can regulate type 2 ACS in a heterologous plant, tomato.
==== Body
Background
Ethylene is a simple gas that acts as a plant hormone; it controls various processes in the plant life cycle, including seed germination, root hair development, root nodulation, flower senescence, abscission, and fruit ripening [1]. Ethylene also is synthesized in response to stresses such as pathogen attack, wounding, hypoxia, ozone, chilling, and freezing [2]. These responses are controlled through integration of the pathways for ethylene biosynthesis, perception, and signal transduction. The ethylene biosynthesis and signalling pathways have been well characterized with regard to physiology, biochemistry, molecular biology, and genetics. Studies of Arabidopsis thaliana revealed a universally conserved set of components in the ethylene signaling pathway for all plants. Ethylene is perceived by ethylene receptor family ETR1, ETR2, EIN4, ERS1, ERS2 [3-6] and their assembly requires the RAN1 copper transporter [7]. The signal is further transduced by the signaling pathway components CTR1, a Raf-like MAPKKK [8,9], EIN2, an Nramp-related integral membrane protein [10], and a cascade of transcriptional regulators EIN3/EILs [11], ERF1 [12], EDF1-4 [13]. The proteolysis of EIN3/EIL proteins by ethylene-regulated EBF1/2 F-box proteins plays a key role in regulation of all known responses to the hormone [14,15]. A MAPK (including AtMPK6) cascade was proposed to be involved in ethylene signaling [16], however, a recent study showed that AtMPK6 is not involved in ethylene signaling but instead involved in the regulation of ethylene biosynthesis through AtACS6 [17,18]. Other yet unidentified components are thought to be involved as well [19,20].
The biosynthetic pathway of ethylene has been studied in detail, and genes encoding the two key enzymes have been cloned and characterized [21-25]. 1-aminocyclopropane-1-carboxylate (ACC) synthase (ACS) converts S-adenosyl-L-methionine (SAM) to ACC and then, ACC oxidase (ACO) produces ethylene through oxidization of ACC. Generally, the reaction catalyzed by ACS is a rate-limiting step [26].
Both ACS and ACO are encoded by gene families in many plant species. In the case of tomato (Lycopersicon esculentum Mill.), there are at least 10 ACS [27] and four ACO genes [28]. The members of ACS and ACO gene families are differentially expressed in development or in response to stimuli such as germination, leaf senescence and flower abscission, fruit ripening, wounding, flooding, exposure to ozone, touch, hormone treatment, and pathogen attack [28-33]. Recent findings suggest that posttranscriptional regulation is an important aspect of the control of ACS expression [22,34-38]. Pharmacological and molecular biological studies have suggested that phosphorylation is involved in the regulation of ACS activity [34,39]. A tomato ACS, LE-ACS2, is phosphorylated in the C-terminal region and this modification appears to be involved in the posttranslational regulation of the enzyme [40]. However, the molecular mechanisms by which plants regulate ethylene biosynthesis at the posttranscriptional or posttranslational levels remain unclear.
Recent studies of the Arabidopsis ethylene-overproducer mutants (eto1,eto2-1,eto3) [36-38,41] revealed mechanisms underlying the posttranslational regulation of ethylene biosynthesis. The eto mutants constitutively display the ethylene-evoked triple response phenotypes in the absence of exogenously applied hormone; they can be distinguished from ctr1 mutants, which also displays the triple response in the absence of ethylene, because the phenotypes of eto mutants are suppressed by inhibitors of ethylene biosynthesis and action [8,42]. Therefore, the eto mutants are likely impaired in either the regulators or the structural enzymes of ethylene biosynthesis. eto1 is a recessive mutation that results in an approximately 10-fold ethylene overproduction in etiolated seedlings compared to wild-type plants [42], whereas eto2-1 and eto3 are dominant mutations that cause 20- and 100-fold increases of ethylene biosynthesis, respectively, in etiolated seedlings [8]. The eto2-1 and eto3 were identified as mutations within the closely related AtACS5 and AtACS9 genes, respectively. These mutations cause alterations of the C-terminal amino acid sequences of each protein [36,38]. Furthermore, both eto1 and eto2-1 mutations increase the stability of the AtACS5 protein [38], suggesting that the C-termini of some ACC synthases are involved in posttranslational regulation/processing or stability of the ACS proteins mediated by ETO1. Also another study showed that AtMPK6 phosphorylates the C-terminus of AtACS6 to stabilize the isozyme by a yet unknown mechanism [17].
The molecular mechanism(s) regulating the activity and stability of AtACS5 was revealed by identification of the ETO1 protein [41]. ETO1 is a member of a novel plant-specific protein family with three distinct protein-protein interaction motifs, namely the BTB domain in its N-terminus and the TPR motifs together with a coiled-coil motif in its C-terminus. The C-terminal TPR domain interacts with AtACS5 and the N-terminal BTB domain interacts with AtCUL3, a constituent of E3 ubiquitin ligase complexes in which ETO1 is proposed to serve as a substrate-specific adaptor protein. ETO1 inhibits the enzyme activity of AtACS5 and targets this protein for degradation in a proteasome-dependent manner. Other studies also suggest that proteolysis is involved in regulation of ethylene biosynthesis [43,44].
ETO1 has two paralogs in Arabidopsis, EOL1 and EOL2 (ETO1-LIKE), both of which also interact with and inhibit the activity of AtACS5. Although ETO1-related sequences are found in many other plant species including tomato, no studies have examined whether members of the ETO1 protein family can interact with and regulate ACC synthases other than AtACS5. In this study, using a heterologous system in tomato, we demonstrate that ETO1 specifically interacts with a subfamily of ACC synthases, namely type 2 ACC synthases including AtACS5, AtACS9 and LE-ACS3, but not with other types of ACC synthases including AtACS6, LE-ACS2 and LE-ACS4. We also show that constitutive expression of ETO1 results in posttranscriptional suppression of a type 2 ACC synthase, LE-ACS3, in transgenic tomato. These results suggest that members of the ETO1 protein family are components in the negative regulation of type 2 ACC synthases in the plant kingdom.
Results
Fruit ripening is not altered in transgenic tomato plants that overexpress ETO1
Orthologous sequences to ETO1 are found in many plant species including tomato [HY and JRE, unpublished data]. These orthologs share high similarity with ETO1 in their TPR domains that are thought to be involved in specific interaction with ACC synthases [41]. Actually, two ETO1 paralogs of Arabidopsis, EOL1 and EOL2, interact with AtACS5 in yeast cells and suppress its activity in Escherichia coli. In tomato, ethylene is a critical regulator of fruit ripening [45], and LE-ACS2 and LE-ACS4 have been shown to be involved in this process [46]. So, there is a possibility that any of the ETO1 ortholog(s) in tomato interact with and regulate these ACS isozymes. Because we expected heterologous interaction between Arabidopsis ETO1 and LE-ACS2 or LE-ACS4, leading to suppression of ethylene biosynthesis in tomato fruits and retardation of its development, we introduced the Arabidopsis ETO1 cDNA controlled by the CaMV 35S promoter into tomato (L. esculentum cv. Shu-gyoku) via Agrobacterium-mediated transformation. Transformants were selected on kanamycin-containing medium and verified using the polymerase chain reaction (PCR) (see methods). Expression of the ETO1 transgene in the T1 segregating individuals was detected by reverse transcription (RT)-PCR assays. All of the T1 lines harbouring the transgene showed expression of the exogenous ETO1 gene (Fig. 1A). Twenty-two T1 individual ETO1-overexpressing (ETO1-OE) lines were derived from four independent T0 lines (Fig. 1A). However, all of the fruits of ETO1-OE lines developed a full red color over the same time course as wild-type plants (Fig. 1B). These results suggest that ETO1 may not suppress the two major ACS isozymes in ripening tomato fruit, LE-ACS2 and LE-ACS4, because there is no noteworthy interaction of ETO1 with either LE-ACS2 or LE-ACS4.
ETO1 specifically interacts with LE-ACS3, but not with LE-ACS2 or LE-ACS4, in yeast cells
To investigate the potential interaction between the Arabidopsis ETO1 and ACC synthases of tomato, we performed yeast two-hybrid assays between ETO1 and three ACS isozymes of tomato, LE-ACS2, LE-ACS3, and LE-ACS4. As described above, LE-ACS2 and LE-ACS4 are major ACS isozymes involved in ripening tomato fruits. LE-ACS3 is induced by auxin and flooding [46,47]. In the yeast two-hybrid assay, whereas LE-ACS3 showed a strong interaction with ETO1, at a level comparable to that of AtACS5, neither LE-ACS2 nor LE-ACS4 interacted with ETO1 (Fig. 2). These results suggest that ETO1 may have a preference for a type of ACS isozymes that include LE-ACS3 and AtACS5.
To investigate this possibility, we focused on the C-termini of various ACC synthases from Arabidopsis and tomato as potential targets of ETO1. C-terminal amino acid sequences of ACC synthases are the least-conserved portion of these proteins [26,48]. However, there is one small conserved motif, RLSF (arginine [R] – leucine [L] – serine [S] – phenylalanine [F]), in the C-termini of several ACC synthases [27,38,40]. A classification of ACC synthases based on the similarity of DNA sequences was reported by Oetiker et al. [35], and phylogenetic classifications based on the similarity of amino acid sequences of Arabidopsis ACC synthases were also devised [17,49]. Apart from these classifications, based on the C-terminal consensus motif mentioned above, we further classified the ACS isozymes of tomato and Arabidopsis into three types as shown in Fig. 3: (1) isozymes with long tails (23–27 amino acids) after the RLSF consensus sequence (like LE-ACS2); (2) isozymes that possess the 'WVF (tryptophan [W] – valine [V] – phenylalanine [F])' consensus sequence just before the RLSF, have short tails (5–8 amino acids) rich in arginine [R] and acidic amino acids (aspartic acid [D] or glutamic acid [E]) after the RLSF, and end with the 'ER (glutamic acid [E] – arginine [R])' consensus (like AtACS5 and LE-ACS3); and (3) isozymes lacking the RLSF consensus sequence (like AtACS7 and LE-ACS4). AtACS3, AtACS10, and AtACS12 are not included in this alignment because AtACS3 is a pseudogene, and AtACS10 and 12 are not ACC synthases but aminotransferases [27]. Hereafter, we use the word "type" instead of "class" to avoid confusion between former sequence-based classifications and our functional classification based on C-terminal motifs and the capacity to bind ETO1/EOL proteins.
Although the lengths of the C-terminal tails of AtACS11, LE-ACS4, and LE-ACS5 are comparable to those of type 2 isozymes, they lack the RLSF motif and the R/D/E-rich region. Therefore, we classified these three isozymes as type 3 isozymes along with AtACS7, which has a much shorter C-terminal tail. Both AtACS5 and LE-ACS3, which interact strongly with ETO1, belong to type 2, whereas LE-ACS2 belongs to type 1 and LE-ACS4 to type 3. Because we previously showed that the C-terminus of AtACS5 is a target of ETO1, this raises the possibility that the C-termini of type 2 ACC synthases contain some specific features necessary for the interaction with ETO1 that may also be conserved in other plants.
ETO1 does not interact with a series of C-terminal deletion mutants of LE-ACS2 in yeast
As described above, one notable feature of type 2 ACC synthases is the length of the C-terminal tail after the RLSF motif. Type 1 ACC synthases possess longer C-terminal tails after the RLSF than type 2, whereas type 3 isozymes lack the RLSF. Furthermore, the C-terminus of ACS appears to be proteolyzed in vivo [50]. Although another study showed that C-terminal truncation of a type 1 enzyme, LE-ACS2, does not occur in vivo [40], other type 1 ACC synthases may have their C-termini truncated to an "optimal length" (i.e., length comparable to type 2) and may then become capable of interacting with ETO1 family proteins. Therefore, we examined the effect of altering the length of ACS protein C-terminal tails on their interaction with ETO1.
Two C-terminally truncated mutants of LE-ACS2 were constructed and the cDNA was cloned into pACT2 (see methods). LE-ACS2Δ16 contains a deletion of 16 amino acids from the C-terminal end; and as a result its C-terminus has a length comparable to native type 2 ACS proteins (Fig. 4). The other mutant, LE-ACS2Δ28, was trimmed to one amino acid upstream of the RLSF (Fig. 4). If a specific length of the C-terminal tail after the RLSF is necessary for the interaction with ETO1, then LE-ACS2Δ16 should interact with ETO1 whereas LE-ACS2Δ28 should not. However, neither of these truncated mutants was able to interact with ETO1 in the yeast two-hybrid assay (Fig. 4). These results suggest that the length of C-termini of type 1 ACS is not preventing ETO1 binding. Simply truncating the C-terminal tail of type 1 ACS to the length of type 2 isozymes is not sufficient to allow for interaction with ETO1 and implies that other factors are necessary for mediating this interaction.
The C-terminal sequence specific to type 2 ACC synthases is required for the interaction with ETO1
Next we examined the effect of type-2-specific amino acid sequence around the RLSF on the interaction with ETO1. As mentioned above, type 2 ACC synthases have a specific C-terminal consensus sequence (i.e., the WVFRLSF motif followed by the R/D/E-rich region). The RLSF is conserved in both type 1 and type 2 ACC synthases, whereas the WVF motif and R/D/E-rich region are conserved only in type 2 ACC synthases in tomato and Arabidopsis (Fig. 3). In the eto3 mutant, the valine residue in this WVF motif is mutated to aspartic acid in AtACS9 [38], suggesting an important role of this small motif (Fig. 3). We replaced the C-terminal 31 amino acids of LE-ACS2 with amino acids 456–469 of LE-ACS3 (LE-ACS2ΔC31::LE-ACS3456–469). This chimeric ACS protein was sufficient to recover the strong interaction with ETO1 to a level comparable to LE-ACS3 (Fig. 4). These results strongly suggest that the C-terminal tail specific to type 2 ACS, comprising the WVFRLSF motif and the R/D/E-rich region, is necessary and sufficient for the interaction with ETO1.
Overexpression of ETO1 suppresses auxin-induced ethylene biosynthesis in tomato seedlings
To investigate the interaction between ETO1 and LE-ACS3 in planta, we examined the induction of ethylene biosynthesis in etiolated tomato seedlings by a synthetic auxin, 2,4-D. Using T2 seedlings of a homozygous line of ETO1-OE (14-1-H2; not included in Fig. 1A), we measured ethylene evolution from auxin-treated wild-type and ETO1 transgenic tomato seedlings.
When 6-day-old etiolated wild-type seedlings were treated with 100 μM 2,4-D for 24 h, they produced 2.1-fold more ethylene than untreated seedlings (Fig. 5A). LE-ACS3 mRNA was induced by the 2,4-D treatment in the wild type (Fig. 5B). However, when the ETO1-OE transgenic plants were treated in this manner, they produced only 1.6-fold more ethylene when compared to untreated plants, despite LE-ACS3 mRNA being strongly induced to a level comparable to wild type (Fig. 5A and 5B). These results, together with those of the yeast two-hybrid assays, indicate that overexpressed ETO1 protein may suppress the ACS activity of LE-ACS3 by a direction, which results in a reduction in ethylene biosynthesis, same as the case of AtACS5 and cytokinin treatment in Arabidopsis [41].
Discussion
ETO1 interacts with and inhibits type 2 ACC synthases
In a previous study, we demonstrated that ETO1, a novel plant-specific BTB/TPR protein, negatively regulates an ethylene biosynthetic enzyme of Arabidopsis seedlings, AtACS5, via direct interaction [41]. However, whether the interaction and the regulatory effect are limited to AtACS5 and its paralogs/orthologs – or are common to the entire plant ACC synthase family – has not been clarified. In this study, we showed the interaction between ETO1 and ACS is restricted to type 2 ACS isozymes. We classified ACC synthases into three types based on their C-terminal amino acid sequences. All members of 'class I' ACC synthases of tomato (LE-ACS1A, LE-ACS1B, LE-ACS6) in the classification of Oetiker et al. [35] correspond to our type 1 ACC synthases, while both of their 'class II' (LE-ACS2 and LE-ACS4) and 'class III' (LE-ACS3 and LE-ACS5) contains members of our type 1 and type 3. This discrepancy may result from the different methods of the phylogenetic anlysis (i.e. partial DNA sequences [35] and C-terminal amino acid sequences [this paper]). On the other hand, our type 2 isozymes of Arabidopsis are grouped into two closely related phylogenetic branches in the group B ACC synthases [49], and the type 3 isozymes belong to the branches in the group B other than the 'type 2' branches. Also our type 1 isozymes correspond to the group A isozymes [49]. These coincidences may reflect the functional relevance of each type of the isozymes, and, indeed, functional heterodimerizations between closely related ACS isozymes have been demonstrated [49].
Little is known about the endogenous regulators of ethylene biosynthesis in the control of plant development, response to stress or fruit ripening. Because tomato has at least one ETO1 ortholog [LeEOL1; HY and JRE, unpublished], a family of ACS isozymes and many ethylene-related phenotypes, we tested whether ETO1 could interact with and inhibit members of the ACC synthase family from tomato. We showed that ETO1 specifically interacts with two type 2 ACC synthases, AtACS5 and LE-ACS3, but not with LE-ACS2 (type 1) or LE-ACS4 (type 3) in the yeast two-hybrid system. Furthermore, we demonstrated that overexpression of the ETO1 transgene posttranscriptionally suppresses the activity of LE-ACS3 in a heterologous plant. These results suggest that type 2 ACC synthases represent a group that specifically interact with and are inhibited by ETO1 family proteins (Fig. 6). Whether the inhibitory effect of the ETO1 family in planta is limited to type 2 is still unknown; we have not yet checked the interaction of ETO1 with LE-ACS2 and LE-ACS4 in planta. In the yeast two-hybrid system, the interaction of ETO1 was obviously restricted to type 2 isozymes. However, in planta, AtMPK6-dependent phosphorylation of AtACS6 (a type 1 ACS) can stabilize this isozyme and increase ethylene biosynthesis [17]. Phosphorylation of such kind of modification of type 1 ACC synthases may confer an ability to interact with ETO1 [52]. Also, a calcium-dependent phosphorylation of the serine residue within the RLSF motif of LE-ACS2 [40], or any proteolytic cleavage [27,40] may change the C-terminal conformation and enable interaction with the ETO1 family. These possibilities may be elucidated by coimmunoprecipitation experiments using antibodies that can distinguish phosphorylated from nonphosphorylated forms of type 1 ACS, or pull-down assay using type-1-specific C-terminal peptide. In contrast, type 3 ACC synthases does not contain this potential phosphorylation site, suggesting that they may not be similarly regulated. The Ctr- phenotype of plants constitutively expressing the C-terminally truncated version of AtACS5 [41] supports this hypothesis.
Another question is whether any other members of the ETO1 family interact with other types of ACC synthases. In the yeast two-hybrid system, both of the two paralogs of ETO1, EOL1 and EOL2, showed an interaction with ACC synthases from tomato similar to that of ETO1 (i.e., they interacted only with type 2 ACS but not with type 1 or type 3; data not shown). This suggests that type-2-specific regulation may be common to other members of the ETO1 family at least in yeast cells. However, a possibility that any EOL proteins may interact with and inhibit ACC synthases other than type 2 isozymes modified in planta is still not excluded.
C-Terminal sequence specific to type 2 ACS is a target of ETO1
The C-terminal amino acid sequence specific to type 2 ACS (i.e., the WVFRLSF followed by the R/D/E-rich region), is a target of ETO1. In addition, mutations found within the RLSF to the R/D/E-rich region of AtACS5 of eto2-1 and the WVF motif of AtACS9 of the eto3 mutant [36,38] strongly implicate the importance of this amino acid sequence. Among molecular lesions of Arabidopsis eto1 mutations identified so far, eto1-1, which lacks only the last TPR motif, has a phenotype similar to the other alleles, and a C-terminal deletion mutant of ETO1 (tETO1) impaired in its last TPR motif failed to interact with AtACS5 in the yeast two-hybrid assay [41]. These results indicate that TPR motifs of ETO1 are essential for interaction with AtACS5.
TPR motifs are involved in various protein-protein interactions. Generally, the interaction between TPR domains and their target peptide are strictly limited by both electrostatic and hydrophobic interactions [53]. In this regard, type-2-specific C-termini of ACC synthases have some features required for the strict interaction with the TPR of ETO1. The crystal structure of LE-ACS2 has been determined, and the C-terminal tail of LE-ACS2 seems to protrude from the surface of the ACS monomer and dimer [54]. In the dimeric form (head-to-tail orientation) of LE-ACS2, N-terminal residues 11–19 make contact with the C-terminal helix H14 just before the C-terminal tail. The significance of the interaction between the N- and the C-termini is not clear but may be important for conformation stabilization and catalysis, as suggested by biochemical studies [50]. The WVF motif is likely located outside this N- and C-terminal interaction region and may be important for formation of the interacting surface with ETO1. Crystal structure analysis of the ETO1-ACS complex would reveal the significance of the WVF motif. Also, an isozyme of type 2 ACS, LE-ACS8, does not possess the precise WVF sequence. Instead, it contains WGF, which is very closely related to WVF (Fig. 3). In contrast to the eto3 version of AtACS9, in which the valine residue of the WVF motif was altered to the charged aspartic acid residue, the glycine residue in the C-terminus of LE-ACS8 is a neutral amino acid. Therefore, it is likely that LE-ACS8 may also interact with ETO1 as do other type 2 isozymes.
Posttranslational regulation of ethylene biosynthesis
We have identified many orthologous sequences of ETO1 from other plant species including tomato, suggesting that the ETO1-based regulatory system is common among the plant kingdom. In the present study, we showed that Arabidopsis ETO1 interacts with and posttranscriptionally regulates LE-ACS3, a type 2 ACS, in a heterologous plant, tomato. This indicates that the ETO1 protein family is a common component in the negative regulation of ethylene biosynthesis in plants. Given that ETO1-homologous sequences are found in all plants but not in animals or prokaryotes, this system of regulation of ethylene may have developed early in plant evolution. ACC synthases are similar to the subgroup I family of pyridoxal 5-phosphate (PLP)-dependent aminotransferases [55]. Most of the similarity between these two families is found around the active sites, while the target of ETO1 lies in the C-terminus of ACS, which is outside the conserved region. It is an intriguing question as to how early plants acquired this unique mechanism to regulate ethylene biosynthesis.
Finally, one may ask why only type 2 ACC synthases are regulated by ETO1. It is imperative for plants to have ethylene synthesized in a timely manner. For instance, a remarkable phenotype of all eto mutants was observed in the etiolated seedling stage, suggesting that production of ethylene is regulated by ETO1 during germination. Although ethylene evolution is important for germination, seedlings with excessive level of ethylene for an extended time would continue to exhibit phenotype similar to that of the triple response, which is normally disappeared after germination. Also ethylene is important to respond to various stresses. To avoid unnecessary overproduction of ethylene in physiological and developmental processes, a tightly regulated system consisted of the type 2 ACC synthases and ETO1 protein family together with yet unknown system (for instance, AtMPK6-AtACS6 system) must have been evolved in plants to timely control ethylene biosynthesis until this important growth regulator and stress phytohormone is needed.
Conclusion
In this study, we elucidated the substrate specificity of ETO1 protein. We showed the interaction between ETO1 and ACS protein family is restricted to type 2 ACS isozymes which possess specific C-terminal amino acid sequences. Our data that ETO1 suppress auxin-induced ethylene evolution through induction of LE-ACS3 also show that Arabidopsis ETO1 can regulate type 2 ACS in a heterologous plant.
Methods
Plant material
Plants of wild type and transformed tomato (L. esculentum cv. Shu-gyoku) were grown in greenhouses under standard conditions. For transformation, seedlings were grown on medium supplemented with 1/2 × Murashige and Skoog (MS) minimal salts.
Transformation of tomato
Tomato hypocotyls or cotyledons precultured on MS medium containing 1 mg/l NAA, 0.5 mg/l BA, and 3% sucrose were transformed by an Agrobacterium-mediated procedure [56]. Full-length ETO1 cDNA was introduced in the sense orientation between the CaMV 35S promoter and Nos terminator of the pROK2 vector harbouring NPT2 gene as a selectable marker [41]. Transformed plants were confirmed by PCR amplification using specific primers for the 35S promoter and the ETO1 cDNA.
Nucleic acid analysis
Genomic DNA was isolated using Isoplant II (Nippon Gene). Total RNA was extracted using RNeasy Plant Mini Kit (Qiagen). Transcription levels of ETO1 were analyzed by reverse RT-PCR analysis of total RNA. RNA (1 μg) was used in each RT-PCR following the manufacturer's protocol (Titan One Tube RT-PCR System, Roche Diagnostics). Gene-specific primers for ETO1 were used, and GAPDH was used as an internal control. Both of the primer sets were added to the same tube. PCR amplification was performed for a first round of 10 cycles as follows: denaturation at 94°C for 30 sec, annealing at 55°C for 30 sec, and extension at 68°C for 45 sec. Then the second round of cycles was performed as follows: denaturation at 94°C for 30 sec, annealing at 55°C for 30 sec, and extension at 68°C for 45 sec + 5/cycle sec. The final extension was carried out at 68°C for 7 min, and the reaction was then stopped at 4°C. For Northern blot analysis, total RNA (10 μg) was used for each lanes and blotted onto positively charged nylon membranes (Roche Diagnostics). DIG-labeled RNA probe for LE-ACS3 was synthesized following the manufacturer's protocol (Roche Diagnostics) and detected by CDP-Star (Roche Diagnostics).
Yeast two-hybrid assay
Yeast two-hybrid assays were performed as described previously [41]. ETO1 cDNA was cloned into the pAS2 vector. cDNAs for LE-ACS2, LE-ACS3, and LE-ACS4 were synthesized by RT-PCR using SuperScriptII (Invitrogen) reverse transcriptase and PfuTurbo DNA polymerase (Stratagene) and cloned into pACT2. Deletion or chimeric mutants of LE-ACS2 and LE-ACS3 were constructed by PCR using PfuTurbo DNA polymerase. A quantitative liquid assay for β-galactosidase activity was performed according to the manufacturer's instructions (BD Biosciences). Each experiment was repeated at least three times using independent clones.
Measurement of ethylene biosynthesis
Etiolated seedlings (1 per vial) of T2 homozygous lines 14-1-H2 (ETO1/ETO1) and 14-1-H4 (wt/wt) were grown in vials (10-mm diameter, 75-mm length with a rubber septum) containing 1 g of sea sand (20–35 mesh, Wako Chemicals) and 0.5 ml of water at 28°C in the dark. After 6 d, 2,4-D (in 1% ethanol) was added to the final concentration of 100 μM and seedlings were incubated under the same conditions. On the next day, the accumulated ethylene was measured using a gas chromatograph (Model GC-7A, Shimadzu) equipped with an active alumina column (60/80 mesh, 3 mm × 1.5 m) and FID. A 0.5-ml volume of each sample from the headspace was injected onto the column. Eight individuals were used for each treatment. Ethylene production was calculated as pL· seedling-1· hr-1.
Authors' contributions
HY conceived of the study, arranged the funding for the project, made transgenic tomato, analyze their phenotypes and expression of the ETO1 transgene, made the two-hybrid vectors with mutant versions of ACS, and carried out all the yeast two-hybrid assays. MN measured ethylene evolution of tomatoes and carried out the Northern hybridization of LE-ACS3. KS participated in genotyping and maintenance of the transgenic plants. KLCW made the two-hybrid vectors with ETO1, participated in discussion and helped to draft the manuscript. JRE participate in discussion, and helped to draft the manuscript. All the authors read and approved the final manuscript.
Acknowledgements
The authors thank E. Kadowaki-Fujii and T. Watase for their skilful assistance. We also acknowledge support from members of the Plant Biotechnology Laboratory, National Agricultural Research Center for Hokkaido Region, and members of the Laboratory of Rice Applied Genetics, National Agricultural Research Center. This work was partly supported by the Ministry of Agriculture, Forestry and Fishery of Japan (to HY).
Figures and Tables
Figure 1 Transgenic tomato plants that overexpress the ETO1 transgene did not show altered fruit ripening. A. Expression of the ETO1 transgene in leaves of T1 individuals. Expression of ETO1 was analyzed by RT-PCR. GAPDH was used as an internal control. One microgram of total RNA was used for each reaction. WT: wild type (Lycopersicon esculentum cv. Shu-gyoku); OE 9, OE 14-0, OE 14-1, OE 14-2 represent independent T0 transformants. Numbers under the horizontal lines represent T1 segregating individuals derived from the corresponding T0 parents. Genotyping for the ETO1 transgene was also performed by PCR and shown under the photograph. B. Representative phenotype of two independent T1 progenies of ETO1 transgenic tomato (line #14-1-5 and #9-2) and wild type (WT). Fruits were harvested at breaker stage and allowed to ripen for further days as indicated.
Figure 2 Arabidopsis ETO1 interacts with LE-ACS3 but not with LE-ACS2 or LE-ACS4 in yeast. Interaction between Arabidopsis ETO1 (in pAS2) and ACC synthases (in pACT2) from tomato (LE-ACS2, LE-ACS3, and LE-ACS4) were analyzed in the yeast two-hybrid system. AtACS5 from Arabidopsis was used as a positive control representing a strong interaction partner with ETO1. pACT2 vector was used as a negative control. Three independent original transformants were analyzed for each combination. Means ± SE (n = 3) are indicated.
Figure 3 Classification based on C-termini of ACC synthases of tomato and Arabidopsis. ACC synthases can be grouped into three distinct types based on the similarity of their C-terminal amino acid sequences. The C-terminal amino acid sequences of ACC synthases were aligned using the CLASTALW program, drawn using MacBoxShade, and grouped into three distinct types. Eight ACC synthases from Arabidopsis (AtACS1, AtACS2, AtACS4, AtACS5, AtACS6, AtACS7, AtACS8, AtACS9, and AtACS11; AtACS3 is thought to be a pseudogene, and AtACS10 and AtACS12 are aminotransferases) and nine from tomato (LE-ACS1A, LE-ACS1B, LE-ACS2, LE-ACS3, LE-ACS4, LE-ACS5, LE-ACS6, LE-ACS7, and LE-ACS8) were aligned. Identical amino acids and conservative changes are indicated by reversed and shaded characters, respectively. The WVF and RLSF motifs and mutation sites for eto2-1 and eto3 are indicated.
Figure 4 Specific C-terminal amino acid sequence of type 2 ACC synthases is necessary for the interaction with ETO1 in the yeast two-hybrid system. Interaction of ETO1 with C-terminal mutants of LE-ACS2 deleted or swapped with LE-ACS3 were analyzed by quantitative yeast two-hybrid assay. ETO1 was cloned into pAS2 and ACS mutants were cloned into pACT2 vectors, respectively. C-terminal amino acid sequences of each mutant are shown on the left. The WVF motif and the R/D/E-rich region derived from LE-ACS3 are shown in red. The RLSF motif common to both LE-ACS3 and LE-ACS2 is shown in blue. The pACT2 vector was used as a negative control. Three independent original transformants were analyzed for each combination. Means ± SE (n = 3) are indicated.
Figure 5 ETO1 suppresses auxin-induced ethylene production in tomato. A. Six-day-old etiolated seedlings of T2 homozygous lines, 14-1-H2 (ETO1 +/+) and 14-1-H4 (ETO1 -/-) were treated with 2,4-D (in 1% ethanol) at the final concentration of 100 μM for 1 d. Ethylene concentration in the headspace was measured by a gas chromatograph. Means ± SE (n = 8) are indicated. B, Induction of LE-ACS3 by 2,4-D. Total RNA (10 μg) from etiolated seedlings (ETO1 +/+ or -/-) with or without 2,4-D treatment was probed with LE-ACS3. Four individuals were used for each treatment. Arrowhead indicates the functional unspliced transcript size [51].
Figure 6 Model for type-2-specific regulation of ACC synthases by ETO1. ACC synthases and ETO1 proteins are shown in light blue and red, respectively. C-terminus of ACS is drawn as a small light blue circle or oval. Note that each pair of ACC synthases forms a dimer with shared active site (two pockets per dimer). Upper panel:, type 2 ACS dimer interacts with ETO1. As a result, the ACS is inhibited for enzymatic activity and targeted for proteasome-dependent degradation. Yet unknown modification (possibly phosphorylation (yellow circles)) of the C-termini of type 2 ACS may inhibit interaction between ACS and ETO1, resulting in ACC production. Lower panel (left) type 1: Although ETO1 does not interact with type 1 ACS dimer because type 1 ACS has a C-terminal tail (blue oval) not suitable for the interaction with ETO1, it is also degraded with yet unidentified mechanism. Triple phosphorylation of the C-termini by MAPK stabilizes type 1 ACS dimer and the dimer produces ACC from SAM. Lower panel (right), type 3: ETO1 does not interact with type 3 ACS dimer because it lacks the specific C-terminal tail required for the interaction with ETO1.
==== Refs
Johnson PR Ecker JR The ethylene gas signal transduction pathway: a molecular perspective Annu Rev Genet 1998 32 227 254 9928480 10.1146/annurev.genet.32.1.227
Wang KLC Li H Ecker JR Ethylene biosynthesis and signaling networks Plant Cell 2002 14 S131 S151 12045274
Chang C Kwok SF Bleecker AB Meyerowitz EM Arabidopsis ethylene-response gene ETR1: similarity of product to two-component regulators Science 1993 262 539 544 8211181
Hua J Chang C Sun Q Meyerowitz EM Ethylene insensitivity conferred by Arabidopsis ERS gene Science 1995 269 1712 1714 7569898
Hua J Sakai H Nourizadeh S Chen QG Bleecker AB Ecker JR Meyerowitz EM EIN4 and ERS2 are members of the putative ethylene receptor gene family in Arabidopsis Plant Cell 1998 10 1321 1332 9707532 10.1105/tpc.10.8.1321
Hua J Meyerowitz EM Ethylene responses are negatively regulated by a receptor gene family in Arabidopsis thaliana Cell 1998 94 261 271 9695954 10.1016/S0092-8674(00)81425-7
Hirayama T Kieber JJ Hirayama N Kogan M Guzman P Nourizadeh S Alonso JM Dailey WP Dancis A Ecker JR RESPONSIVE-TO-ANTAGONIST1, a Menkes/Wilson disease-related copper transporter, is required for ethylene signaling in Arabidopsis Cell 1999 97 383 393 10319818 10.1016/S0092-8674(00)80747-3
Kieber JJ Rothenberg M Roman G Feldmann KA Ecker JR CTR1, a negative regulator of the ethylene response pathway in Arabidopsis, encodes a member of the Raf family of protein kinases Cell 1993 72 427 441 8431946 10.1016/0092-8674(93)90119-B
Clark KL Larsen PB Wang X Chang C Association of the Arabidopsis CTR1 Raf-like kinase with the ETR1 and ERS ethylene receptors Proc Natl Acad Sci USA 1998 95 5401 5406 9560288 10.1073/pnas.95.9.5401
Alonso JM Hirayama T Roman G Nourizadeh S Ecker JR EIN2, a bifunctional transducer of ethylene and stress responses in Arabidopsis Science 1999 284 2148 2152 10381874 10.1126/science.284.5423.2148
Chao Q Rothenberg M Solano R Roman G Terzaghi W Ecker JR Activation of the ethylene gas response pathway in Arabidopsis by the nuclear protein ETHYLENE-INSENSITIVE3 and related proteins Cell 1997 89 1133 1144 9215635 10.1016/S0092-8674(00)80300-1
Solano R Stepanova A Chao Q Ecker JR Nuclear events in ethylene signaling: a transcriptional cascade mediated by ETHYLENE-INSENSITIVE3 and ETHYLENE-RESPONSE-FACTOR1 Genes Dev 1998 12 3703 3714 9851977
Alonso JM Stepanova AN Leisse TJ Kim CJ Chen H Shinn P Stevenson DK Zimmerman J Barajas P Cheuk R Gadrinab C Heller C Jeske A Koesema E Meyers CC Parker H Prednis L Ansari Y Choy N Deen H Geralt M Hazari N Hom E Karnes M Mulholland C Ndubaku R Schmidt I Guzman P Aguilar-Henonin L Schmid M Weigel D Carter DE Marchand T Risseeuw E Brogden D Zeko A Crosby WL Berry CC Ecker JR Genome-wide insertional mutagenesis of Arabidopsis thaliana Science 2003 301 653 657 12893945 10.1126/science.1086391
Guo H Ecker JR Plant responses to ethylene gas are mediated by SCFEBF1/EBF2-dependent proteolysis of EIN3 transcription factor Cell 2003 115 667 677 14675532 10.1016/S0092-8674(03)00969-3
Potuschak T Lechner E Parmentier Y Yanagisawa S Grava S Koncz C Genschik P EIN3-dependent regulation of plant ethylene hormone signaling by two Arabidopsis F box proteins: EBF1 and EBF2 Cell 2003 115 679 689 14675533 10.1016/S0092-8674(03)00968-1
Ouaked F Rozhon W Lecourieux D Hirt H A MAPK pathway mediates ethylene signaling in plants EMBO J 2003 22 1282 1288 12628921 10.1093/emboj/cdg131
Liu Y Zhang S Phosphorylation of 1-aminocyclopropane-1-carboxylic acid synthase by MPK6, a stress-responsive mitogen-activated protein kinase, induces ethylene biosynthesis in Arabidopsis Plant Cell 2004 16 3386 3399 15539472 10.1105/tpc.104.026609
Ecker JR Reentry of the Ethylene MPK6 Module Plant Cell 2004 16 3169 3173 10.1105/tpc.104.161210
Roman G Lubarsky B Kieber JJ Rothenberg M Ecker JR Genetic analysis of ethylene signal transduction in Arabidopsis thaliana : five novel mutant loci integrated into a stress response pathway Genetics 1995 139 1393 1409 7768447
Larsen PB Chang C The Arabidopsis eer1 mutant has enhanced ethylene responses in the hypocotyl and stem Plant Physiol 2001 125 1061 1073 11161061 10.1104/pp.125.2.1061
Sato T Theologis A Cloning the mRNA encoding 1-aminocyclopropane-1-carboxylate synthase, the key enzyme for ethylene biosynthesis in plants Proc Natl Acad Sci USA 1989 86 6621 6625 2671999
Nakajima N Nakagawa N Imaseki H Molecular size of wound-induced 1-aminocyclopropane-1-carboxylate synthase from Cucurbita maxima Dutch. and change of translatable mRNA of the enzyme after wounding Plant Cell Physiol 1990 29 989 998
Hamilton AJ Bouzayen M Grierson D Identification of a tomato gene for the ethylene-forming enzyme by expression in yeast Proc Natl Acad Sci USA 1991 88 7434 7437 1714605
Spanu P Reinhardt D Boller T Analysis and cloning of the ethylene-forming enzyme from tomato by functional expression of its mRNA in Xenopus laevis oocytes EMBO J 1991 10 2007 2013 2065651
Van der Straeten D Rodrigues-Pousada RA Villarroel R Hanley S Goodman HM Van Montagu M Cloning, genetic mapping, and expression analysis of an Arabidopsis thaliana gene that encodes 1-aminocyclopropane-1-carboxylate synthase Proc Natl Acad Sci USA 1992 89 9969 9973 1357670
Kende H Ethylene biosynthesis Annu Rev Plant Physiol Plant Mol Biol 1993 44 283 307 10.1146/annurev.pp.44.060193.001435
Yamagami T Tsuchisaka A Yamada K Haddon WF Harden LA Theologis A Biochemical diversity among the 1-amino-cyclopropane-1-carboxylate synthase isozymes encoded by the Arabidopsis gene family J Biol Chem 2003 278 49102 49112 12968022 10.1074/jbc.M308297200
Moeder W Barry CS Tauriainen AA Betz C Tuomainen J Utriainen M Grierson D Sandermann H Langebartels C Kangasjarvi J Ethylene synthesis regulated by biphasic induction of 1-aminocyclopropane-1-carboxylic acid synthase and 1-aminocyclopropane-1-carboxylic acid oxidase genes is required for hydrogen peroxide accumulation and cell death in ozone-exposed tomato Plant Physiol 2002 130 1918 1926 12481074 10.1104/pp.009712
Yang SF Hoffman NE Ethylene biosynthesis and its regulation in higher plants Annu Rev Plant Physiol 1984 35 155 189 10.1146/annurev.pp.35.060184.001103
Mattoo AK Suttle JC The Plant Hormone Ethylene 1991 CRC Press
Abeles FB Morgan PW Salveit ME Jr Ethylene in Plant Biology 1992 2 Academic Press
Nakatsuka A Murachi S Okunishi H Shiomi S Nakano R Kubo Y Inaba A Differential expression and internal feedback regulation of 1-aminocyclopropane-1-carboxylate synthase, 1-aminocyclopropane-1-carboxylate oxidase, and ethylene receptor genes in tomato fruit during development and ripening Plant Physiol 1998 118 1295 1305 9847103 10.1104/pp.118.4.1295
Peck SC Kende H Differential regulation of genes encoding 1-aminocyclopropane-1-carboxylate (ACC) synthase in etiolated pea seedlings: effects of indole-3-acetic acid, wounding, and ethylene Plant Mol Biol 1998 38 977 982 9869404 10.1023/A:1006033030081
Spanu P Grosskopf DG Felix G Boller T The apparent turnover of 1-aminocyclopropane-1-carboxylate synthase in tomato cells regulated by protein phosphorylation and dephosphorylation Plant Physiol 1994 106 529 535 12232347
Oetiker JH Olson DC Shiu OY Yang SF Differential induction of seven 1-aminocyclopropane-1-carboxylate synthase genes by elicitor in suspension cultures of tomato (Lycopersicon esculentum) Plant Mol Biol 1997 34 275 286 9207843 10.1023/A:1005800511372
Vogel JP Woeste KE Theologis A Kieber JJ Recessive and dominant mutations in the ethylene biosynthetic gene ACS5 of Arabidopsisconfer cytokinin insensitivity and ethylene overproduction, respectively Proc Natl Acad Sci USA 1998 95 4766 4771 9539813 10.1073/pnas.95.8.4766
Woeste KE Ye C Kieber JJ Two Arabidopsis mutants that overproduce ethylene are affected in the posttranscriptional regulation of 1-aminocyclopropane-1-carboxylic acid synthase Plant Physiol 1999 119 521 530 9952448 10.1104/pp.119.2.521
Chae HS Faure F Kieber JJ The eto1, eto2, and eto3 mutations and cytokinin treatment increase ethylene biosynthesis in Arabidopsis by increasing the stability of ACS protein Plant Cell 2003 15 545 559 12566591 10.1105/tpc.006882
Liang X Shen NF Theologis A Li+-regulated 1-aminocyclopropane-1-carboxylate synthase gene expression in Arabidopsis thaliana Plant J 1996 10 1027 1036 9011084 10.1046/j.1365-313X.1996.10061027.x
Tatsuki M Mori H Phosphorylation of tomato 1-aminocyclopropane-1-carboxylic acid synthase, LE-ACS2, at the C-terminal region J Biol Chem 2001 276 28051 28057 11375393 10.1074/jbc.M101543200
Wang KLC Yoshida H Lurin C Ecker JR Regulation of ethylene gas biosynthesis by the Arabidopsis ETO1 protein Nature 2004 428 945 950 15118728 10.1038/nature02516
Guzman P Ecker JR Exploiting the triple response of Arabidopsis to identify ethylene-related mutants Plant Cell 1990 2 513 523 2152173 10.1105/tpc.2.6.513
Larsen PB Cancel JD A recessive mutation in the RUB1-conjugating enzyme, RCE1, reveals a requirement for RUB modification for control of ethylene biosynthesis and proper induction of basic chitinase and PDF1.2 in Arabidopsis Plant J 2004 38 626 638 15125769 10.1111/j.1365-313X.2004.02068.x
Bostick M Lochhead SR Honda A Palmer S Callis J Related to ubiquitin 1 and 2 are redundant and essential and regulate vegetative growth, auxin signaling, and ethylene production in Arabidopsis Plant Cell 2004 16 2418 2432 15319484 10.1105/tpc.104.024943
Oeller PW Lu MW Taylor LP Pike DA Theologis A Reversible inhibition of tomato fruit senescence by antisense RNA Science 1991 254 437 439 1925603
Lincoln JE Campbell AD Oetiker J Rottmann WH Oeller PW Shen NF Theologis A LE-ACS4, a fruit ripening and wound-induced 1-aminocyclopropane-1- carboxylate synthase gene of tomato (Lycopersicon esculentum). Expression in Escherichia coli, structural characterization, expression characteristics, and phylogenetic analysis J Biol Chem 1993 268 19422 19430 8366090
Yip WK Moore T Yang SF Differential accumulation of transcripts for four tomato 1-aminocyclopropane-1-carboxylate synthase homologs under various conditions Proc Natl Acad Sci USA 1992 89 2475 2479 1549612
Zarembinski TI Theologis A Ethylene biosynthesis and action: a case of conservation Plant Mol Biol 1994 26 1579 1597 7858205 10.1007/BF00016491
Tsuchisaka A Theologis A Heterodimeric interactions among the 1-amino-cyclopropane-1-carboxylate synthase polypeptides encoded by the Arabidopsis gene family Proc Natl Acad Sci USA 2004 101 2275 2280 14983000 10.1073/pnas.0308515101
Li N Mattoo AK Deletion of the carboxyl-terminal region of 1-aminocyclopropane-1-carboxylic acid synthase, a key protein in the biosynthesis of ethylene, results in catalytically hyperactive, monomeric enzyme J Biol Chem 1994 269 6908 6917 8120053
Olson DC Oetiker JH Yang SF Analysis of LE-ACS3, a 1-aminocyclopropane-1-carboxylic acid synthase gene expressed during flooding in the roots of tomato plants J Biol Chem 1995 270 14056 14061 7775465 10.1074/jbc.270.23.14056
Kim CY Liu Y Thorne ET Yang H Fukushige H Gassmann W Hildebrand D Sharp RE Zhang S Activation of a stress-responsive mitogen-activated protein kinase cascade induces the biosynthesis of ethylene in plants Plant Cell 2003 15 2707 2718 14555690 10.1105/tpc.011411
Scheufler C Brinker A Bourenkov G Pegoraro S Moroder L Bartunik H Hartl FU Moarefi I Structure of TPR domain-peptide complexes: critical elements in the assembly of the Hsp70-Hsp90 multichaperone machine Cell 2000 101 199 210 10786835 10.1016/S0092-8674(00)80830-2
Huai Q Xia Y Chen Y Callahan B Li N Ke H Crystal structures of 1-aminocyclopropane-1-carboxylate (ACC) synthase in complex with aminoethoxyvinylglycine and pyridoxal-5'-phosphate provide new insight into catalytic mechanisms J Biol Chem 2001 276 38210 38216 11431475 10.1074/jbc.M008127200
Mehta PK Hale TI Christen P Aminotransferases: demonstration of homology and division into evolutionary subgroups Eur J Biochem 1993 214 549 561 8513804 10.1111/j.1432-1033.1993.tb17953.x
Nagata M Mori H Tabei Y Sato T Hirai M Imaseki H Modification of tomato fruit ripening by transformation with sense or antisense chimeric 1-aminocyclopropane-1-carboxylate synthase genes Acta Horticulture 1995 394 213 218
|
16091151
|
PMC1199607
|
CC BY
|
2021-01-04 16:03:52
|
no
|
BMC Plant Biol. 2005 Aug 10; 5:14
|
utf-8
|
BMC Plant Biol
| 2,005 |
10.1186/1471-2229-5-14
|
oa_comm
|
==== Front
Cancer Cell IntCancer Cell International1475-2867BioMed Central London 1475-2867-5-211600461810.1186/1475-2867-5-21Primary ResearchFunctional and structural characteristics of anticancer peptide Pep27 analogues Lee Dong Gun [email protected] Kyung-Soo [email protected] Yoonkyung [email protected] Hai-Young [email protected] Weontae [email protected] Sung-Chul [email protected] Youn-Kyung [email protected] Cheol-Hee [email protected] School of Life Science and Biotechnology, College of Natural Sciences, Kyungpook National University, 1370 Sankyuk-dong, Puk-ku, Taegu 702-701, Korea2 Research Center for Proteinous Materials, Chosun University, Gwangju 501–759, Korea3 Departement of Biochemistry and HTSD-NMR National Research Laboratory, Yonsei University, Seoul, South Korea4 Department of Pathology, College of Medicine, Chosun University, Gwangju 501–759, Korea5 Research Center for Resistant Cells and Department of Pharmacology, College of Medicine, Chosun University, Gwangju 501–759, Korea2005 11 7 2005 5 21 21 3 11 2004 11 7 2005 Copyright © 2005 Lee et al; licensee BioMed Central Ltd.2005Lee et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
A secreted peptide Pep27 initiates the cell death program in S. pneumoniae through signal transduction. This study was undertaken to evaluate the relation between the structure and cytotoxic activity of Pep27 and its analogues on cancer cells.
Results
Pep27anal2 characterized substituting (2R→W), (4E→W), (11S→W) and (13Q→W) in native Pep27, exhibited greater hydrophobicity and anticancer activity than Pep27 and other analogues. The IC50 values of Pep27anal2 were approximately 10 – 30 μM in a number of cell lines (AML-2, HL-60, Jurkat, MCF-7 and SNU-601). Confocal microscopy showed that Pep27anal2-FITC was localized in the plasma membrane, and then moving from the membrane to subcellular compartments with the initiation of membrane blebbing. Flow cytometric analysis using propidium iodide and Annexin V also revealed that Pep27anal2 induced apoptosis with minor membrane damage. Electron microscopy revealed that Pep27 induced apoptosis in Jurkat cells. The anticancer activity of Pep27anal2 was neither abrogated by pan-caspase inhibitor (Z-VAD-fmk) nor related to cytochrome c release from mitochondria. The 3D solution structures of these two Pep27 peptides revealed that both form a random coil conformation in water; however, they adopted stable α-helical conformations in solutions.
Conclusion
The results indicate that Pep27anal2 can penetrate the plasma membrane, and then induce apoptosis in both caspase-and cytochrome c-independent manner. The hydrophobicity of Pep27anal2 appears to play an important role in membrane permeabilization and/or anticancer properties. The structure-functional relationships of these peptides are also discussed. It is proposed that Pep27anal2 is a potential candidate for anticancer therapeutic agents.
Peptide Pep27Pep27 analoguesS. pneumoniaeapoptosisanticancer activity3D structure
==== Body
Introduction
Newly discovered peptides are becoming increasingly important, not only as molecular tools for the understanding of protein-protein interactions, but also as lead compounds. Cationic amphipathic peptides, such as cecropins, defensins, and mellitin, induce cell death in prokaryotic and eukaryotic cells by increasing membrane permeability [1]. This increased permeability may lead to cell lysis or, alternatively, may produce subtle changes in the membrane barrier function and promote cell death [2]. It has also been shown that the anticancer activities of natural and synthetic cationic peptides exhibit therapeutic activity in murine models of human tumors [3-5].
It has been proposed that processes occurring within bacteria are necessary to trigger endogenous suicidal enzymes that dissolve the cell wall during antibiotic-induced autolysis. Recently, Pep27, was found to initiate the cell death program in S. pneumoniae through signal transduction triggered via the two-component system (VncR/S) consisting of a membrane-bound histidine protein kinase (e.g., VncS) and a cytoplasmic effector termed a response regulator (e.g., VncR) [6]. Pep27 is secreted through the Vex ABC transporter and accumulates in the medium [6,7]. Accumulated Pep27 is then sensed by VncS, converting this histidine protein kinase into a phosphatase that dephosphorylates phosphorylated VncR, allowing derepression of autolytic pathways and resulting in the loss of viability and lysis of pneumococcal cells.
The present study was undertaken to identify analogues of bacterial death signal peptide Pep27 capable of inhibiting the growth of cancer cells, and to investigate the nature of the relation between their biologic behaviors and their molecular structures.
Materials and methods
Culture
OCI-AML-2 (AML-2) cell line was obtained from the Ontario Cancer Institute (Toronto, Canada), and Jurkart and HL-60 cell lines from the American Type Culture Collection (ATCC, USA). These three cell lines were cultured at 37°C in a 5% CO2 atmosphere using α-MEM medium (GibcoBRL, Gland Island, NY, USA) containing 10% heat-inactivated fetal bovine serum (FBS, Sigma, ST. Louis, MO, USA). MCF-7 and NIH/3T3 cell lines from ATCC and SNU-601 cell line from the Korean Cancer Cell Bank (Seoul, Korea) were cultured under the same conditions but using RPMI-1640 medium (GibcoBRL, Gland Island, NY, USA) containing 10% heat-inactivated FBS. Cells were maintained in suspension or as monolayer cultures, and subcultured.
Peptide synthesis
Peptides were synthesized by using a solid phase method using (based on) Fmoc (9-fluorenyl-methoxycarbonyl)-chemistry [8]. Rink amide 4-methyl benzhydrylamine (MBHA) resin (0.55 mmol/g) was used as the support. Coupling of Fmoc-amino acids was performed by DCC (dicyclohexyl-carbodiimide/HOBt (1-hydroxybenzotriazole). After completion of peptide chain elongation, the protected final peptide resins were treated with the reagent K. The crude peptide obtained was then washed repeatedly with diethylether, dried in vacuum, and purified by reverse-phase (RP) HPLC (LC10A, Shimadzu, Tokyo, Japan) on a Waters 15-m Deltapak C18 column. The purity of the peptide was checked by analytical RP-HPLC using an Ultrasphere C18 column (4.625 cm, Beckman, Miami, FL, USA). Purified peptides were hydrolyzed with 6 N HCl at 110°C for 22 h, and then dried in vacuum. The residues were dissolved in 0.02 N HCl and subjected to amino acid analysis (8500 A, Hitachi, Tokyo, Japan) to determine peptide concentration. The molecular weights of the synthetic peptides were determined by MALDI (matrix-assisted laser desorption ionization) mass spectra, respectively (data not shown).
Cytotoxicity test
The in vitro cytotoxicity of the Pep27 analogues produced was measured by MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide, Sigma, ST. Louis, USA] assay, as described by [9]. IC50 values were determined directly from semilogarithmic dose-response curves. Experiments were carried out at least in duplicate.
Hemolytic activity
Peptide hemolytic activities were measured as described by [10], by determining hemoglobin release by 4% suspensions of fresh human erythrocytes at 414 nm. Briefly, human erythrocyte cells were centrifuged and washed three times with phosphate-buffered saline (PBS: 35 mM phosphate buffer/0.15 M NaCl, pH 7.0). 100 ml of human red blood cells suspended 8% (v/v) in PBS were the inoculated into 96-well plates. 100 ml of the peptide solution, Pep27 or its analogues, was then added to each well. The plates were incubated for 1 h at 37°C, and centrifuged at 150 g for 5 min. 100 ml aliquots of the supernatant obtained were transferred to flat-bottom 96-microtiter plates. Hemolysis was determined by measuring absorbance at 414 nm with an ELISA plate reader (Molecular Devices Emax, Sunnyvale, CA, USA). Hemolysis rates of 0% or 100% were determined as the same way with PBS and 0.1% Triton-100, respectively. Hemolysis percent was calculated using the following equation: % hemolysis = [(Abs414nm in the peptide solution - Abs414nm in PBS)/(Abs414nm in 0.1% Triton-100 - Abs414nm in PBS)] × 100.
Preparation of fluorescein isothiocyanate (FITC)-labeled peptide
FITC was freshly dissolved in DMSO to 10 mg/ml, and added to 1 mg/ml of peptide in 100 mM sodium bicarbonate, pH 9.3 to a final concentration of 1 mg/ml FITC. After incubation for 4 hr in the dark at room temperature, 1 M ethanolamine was added to inactivate the residual FITC. The solution was then left in the dark for an additional 2 hr, and chromatographed sequentially on a gel-filtration column followed by a 120 cm Sephadex G-50 column to remove unconjugated dye. The FITC-labeled peptide (FITC-Pep27anal2) was obtained by reverse phase HPLC using a C18 column. Conjugation between FITC and the peptide of interest was confirmed by SDS-PAGE gel as intense peptide fluorescence under UV light.
Confocal microscopy
Jurkat cells grown on glass coverslips overnight were treated with FITC-labeled Pep27anal2 at a concentration of 5 μg/ml. The cells were washed with PBS at 30, 60 and 120 min being treated of Pep27anal2, and then fixed using 3.7% formaldehyde for 15 min. Coverslips were mounted on glass slides, FITC was excited using a 488-nm laser, and images were manipulated using the manufacture's software.
Determination of apoptosis and membrane integrity
Annexin V binding was assessed by flow cytometry, and cell staining was evaluated using FITC-labeled annexin V (green fluorescence), and propidium iodide (PI) (negative for red fluorescence) simultaneously. This test was able to discriminate between intact cells (FITC-/PI-), apoptotic cells (FITC+/PI-) and necrotic cells (FITC+/PI+) (Vermes et al., 1995). Jurkat cells (105/ml) were exposed to etoposide (100 μM), mellitin (12.5 μM) and Pep27anal2 (12.5 μM and 62.5 μM) for 4 hr. Cells were washed with PBS, resuspended in 400 μl of binding buffer (10 mM Hepes/NaOH, pH 7.4, 140 mM NaCl, 2.5.mM CaCl2), and incubated with 10 μl of annexin V-FITC (MedSystems Diagnostics, GmbH, Australia) for 10 min at room temperature. After incubation, cells were washed with binding buffer and exposed to 1 μg of PI in a final volume of 400 μl. FITC and PI fluorescences were determined using a flow cytometer (Becton Dickinson, Franklin Lakes, FL, USA)
Electron microscopy
Jurkat cells were harvested and immediately fixed in phosphate-buffered glutaraldehyde solution (pH 7.2) for 2 hrs and then fixed in 1% buffered osmium tetroxide for 2 hrs at 4°C. After processing through a graded series of ethanol and propylene oxide, the cells were embedded in Epon. A semi-thin (1 μm) section was cut and stained with toluidine blue for the ultrastructural examination. Ultra-thin (60–90 nm) sections were also prepared from the Epon block above using an ultramicrotome (LKB-V, Sweden), stained with uranyl acetate and lead citrate, and examined under a Hitachi H-7600 electron microscope (Hitachi, Kyoto, Japan). Apoptosis detection was done by taking photographs (×1,000) of the entire field of each ultra-thin sectioned sample. The degree of apoptosis (apoptosis index) was defined as the number of apoptotic cells expressed as a percentage of the total cell count.
Preparation of cytosolic extracts and the immunoblotting of cytochrome c
Cytochrome c immunoblotting was performed using a slight modification of a previously described method [11]. Jurkat cells were collected by centrifugation at 200 g for 5 min at 4°C. The cells were then washed twice with ice-cold PBS, pH 7.4, and centrifuged at 200 g for 5 min. The cell pellet obtained was then resuspended in 600 μl of extraction buffer, containing 20 mM Hepes-KOH (pH 7.5), 10 mM KCl, 250 mM sucrose, 1 mM EGTA, 1 mM EDTA, 1.5 mM MgCl2, 1 mM dithiothreitol and 0.1 mM phenylmethylsulfonyl fluoride.
After incubation for 30 min on ice, the cells were homogenized (20 strokes), spun at 14,000 g for 15 min, supernatants were separated from the mitochondrial fraction, and both were stored at -70°C, until required for gel electrophoresis. The mitosol fraction was obtained by adding 0.1% SDS. Fifty μg of cytosolic protein extracts and the mitosol extracts were loaded into the lanes of an 15% SDS-polyacrylamide gel, separated and blotted onto a nitrocellulose membrane (Amersham, Piscataway, NJ, USA) at 150 mA overnight in transfer buffer (20 mM Tris-base, 150 mM glycine, 20% methanol). Non-specific binding was blocked by incubating the membrane in 5% non-fat milk in TBS-T (Tris-buffered saline and Tween-20; 10 mM Tris-HCl pH 8.0, 50 mM NaCl, and 0.05% Tween-20) for 3 h at RT. The membrane was incubated in anti-cytochrome c monoclonal antibody (7H8.2Cl2) solution (1:2000 dilution, BD PharMingen, San Diego, CA, USA) containing 0.5% non-fat milk in TBS-T. After overnight incubation with agitation at 4°C, the membrane was washed three times with 0.5% non-fat milk in TBS-T. The secondary antibody, a horseradish peroxidase-coupled anti-mouse antibody (Sigma, ST. Louis, MO, USA), was incubated at a 1:1000 dilution for 2 h at RT. The membrane was then washed four times (5 min) with 5% non-fat milk in TBS-T.
NMR (nuclear magnetic resonance) spectroscopy
The peptide samples used for NMR experiments contained 1.5 mM peptide in water /2, 2, 2-trifluoro-(d3)-ethanol (TFE) mixture (50:50; v/v) at pH 7.2, which was prepared after lyophilizing samples in an aqueous solution. NMR spectra were recorded at 15°C on a Bruker DRX-500 spectrometer equipped with a triple-resonance probe and an x, y, z-shielded pulsed-field gradient coil. Two-dimensional (2D) NMR spectra were recorded in the phase-sensitive mode using time-proportional phase incrementation [12] for quadrature detection in the t1 domain. The 2D experiments such as TOCSY was performed using a MLEV-17 spin-lock pulse sequence with a mixing time of 69.7 ms and 2D-NOESY with mixing time of 300 ms. Solvent suppression during TOCSY and NOESY experiments was achieved using a WATERGATE pulse sequence [13] in combination with a pulsed-field gradient pulse. All NMR spectra were acquired with 2048 complex the t2 data points and 256 increments in the t1 dimension, with 32 scans per increment. Slowly exchanging amide protons were identified by lyophilization of a fully protonated sample in 50% H2O/ 50% TFE, redissolution in 50% D2O/ 50% TFE solution, and the immediate acquisition of a series of one-dimensional NOESY and 2D-NOESY spectra. 3JHNα coupling constants were measured from the 1D spectrum. NMR data were processed on a Silicon Graphics Indigo II workstation using XWINNMR (Bruker Instruments, Karlsruhe, Germany) software. Proton chemical shifts were expressed relative to the methyl resonance of sodium 2, 2-dimethyl-2-silapentane-5-sulfonic acid (DSS), used as an internal standard.
Statistical analysis
Data are expressed as means ± SEM. One-way ANOVA and the Student's t-test were performed using the Statistical Package for the Social Sciences (SPSS) for Windows, version 7.5 (SPSS, Korea). Results were considered significant when p was <0.05.
Results
Design and synthesis of peptides
The amphipathic features of α-helical antimicrobial peptides plays an important role against target cells. Furthermore, a number of parameters, including a net positive charge, α-helicity, and overall hydrophobicity have been shown to modulate the antibiotic activity of the α-helical amphipathic antimicrobial peptides. Recent investigations using amphipathic α-helical model peptides have revealed that the modulation of the hydrophilic-hydrophobic balance of α-helical antimicrobial peptides is a crucial factor in the design of peptides with potent antibiotic activity [14]. In a preliminary experiment, we could not find that Pep27 has anticancer activity. Therefore, we designed Pep27 analogue peptides and evaluated their anticancer and chemosensitizing effects. In order to elucidate the relationship between the hydrophobic and the hydrophilic regions of antibiotic peptides with respect to antibiotic activity and cytotoxicity, we designed and synthesized certain Pep27 analogues, on the basis of the α-helical wheel diagram of Pep27. In particular, Pep27 hydrophobicity was increased by substituting tryptophan. The amino acid sequences used in the present study are summarized in Table 1. Synthesized peptides were purified by RP-HPLC and their molecular weights were confirmed by MALDI-MS.
Table 1 Amino acid sequences of Pep27 derived from S. pneumoniae and its analogue peptides, and their retention times
Peptide Amino acid sequences Remarks (substitution) RT (min)
Pep27 MRKEFHNVLSSGQLLADKRPARDYNRK (2R→W), (4E→W) 14.54
Pep27anal1 MWKWFHNVLSSWQLLADKRPARDYNRK (2R→W), (4E→W) 18.67
Pep27anal2 MWKWFHNVLSWWWLLADKRPARDYNRK (2R→W), (4E→W), (11S→W), (13Q→W) 22.50
Pep27anal3 MRKWFHNVLSSGQLLADKWPAWDYNRK (4E→W), (19R→W), (22R→W), (26R→W) 20.08
Pep27anal4 MWKEFHNVLSSGQLLADKRWARWYNRW (2R→W), (20P→W), (23D→W), (27K→W) 19.15
Pep27anal5 MWKWFHNVLSSGQLLADKWWAWWYNWW (2R→W), (4E→W), (19R→W), (20P→W), (22R→W), (23D→W), (26R→W), (27K→W) 19.17
RT, retention time obtained by RP-HPLC
RP-HPLC has often been used to experimentally determine the quantitative hydrophobic-hydrophilic balance of amphipathic peptides. Peptides are separated by RP-HPLC according to their hydrophobic interactions with C-18 of the stationary phase. Such hydrophobic interactions can be considered comparable to the interaction between amphipathic antimicrobial peptides and the lipid bilayer of plasma membranes. In order to investigate the possibility of a correlation between peptide hydrophobicity and the biological activities on cancer cells, the hydrophobic characteristics of the peptides were investigated by comparing their retention times on RP-HPLC (Table 1). As expected, all Pep27 analogues eluted later than Pep27, and the analogue Pep27anal2 had the longest retention time.
Anticancer activities of Pep27 and its analogues
To evaluate the anticancer activities of the Pep27 analogues, different concentrations of these peptides were administered to various cancer cell lines, and cytotoxicities were then determined by MTT assay. In this study, we used both anchorage-dependent and independent cancer cell lines. These included AML-2 an acute myelogenous leukemia cell line, HL-60 an acute promyelocytic leukemia cell line, Jurkat a T cell leukemia cell line, SNU-601 a gastric cancer cell line and MCF-7 a breast cancer cell line. Pep27 analogue peptides showed similar anticancer activities in five cancer cell lines tested. Of the analogues tested, Pep27anal2 showed most anticancer activity (Fig. 1). The IC50 and IC90 values of Pep27anal2 in five cancer cell lines tested were 10 – 28 μM and 35 – 55 μM, respectively (Fig. 1 and Table 2). Since the substitution of (2R→W) and (4E→W) to form Pep27anal1 showed less activity than Pep27anal2, the additional substitution of (11S→W) and (13Q→W) in Pep27anal2 was viewed as being important for anticancer activity. On the other hand, other Pep27 analogues substituted with more than two tryptophans at different sites did not show more anticancer activity than Pep27anal2 in 5 cell lines tested. Moreover, of the Pep27 analogues produced, Pep27anal2 had the greatest hydrophobicity, based on its RP-HPLC retention time (Table 1).
Table 2 Anticancer activity of Pep27 and of its analogues in various cancer cell lines by MTT assay
Peptide IC50 (μM)
AML-2 HL-60 Jurkat SNU-601 MCF-7
Pep27 >70 >70 >70 >70 >70
Pep27anal1 50 53 47 37 55
Pep27anal2 29 20 23 25 <10
Pep27anal3 67 52 50 50 38
Pep27anal4 50 51 46 37 29
Pep27anal5 >70 >70 >70 >70 >70
AML-2, acute myelogenous leukemia cell line; HL-60, acute promyelocytic leukemia cell line; Jurkat, T cell leukemia cell line; SNU-601, gastric cancer cell line; MCF-7, breast cancer cell line
Figure 1 Cytotoxicity of Pep27 and of its analogues in various cancer cell lines by MTT assay. AML-2, human acute myelogenous leukemia cell line; HL-60, human promyelocytic leukemia cell line; Jurkat, T-cell leukemia cell line; SNU-601, human gastric adenocarcinoma cell line; MCF-7, human breast cancer cell line
How does Pep27anal2 show its cytotoxic effect in cancer cells?
In order to determine whether or not Pep27anal2 can cause membrane damage, an RBC hemolysis test was performed (Table 3). Of the Pep27 analogues, only Pep27anal 2 at 12.5 μM for 1 h at 37°C, hemolysed 18% of RBCs whereas mellitin at the same concentration induced 100% hemolysis, suggesting that Pep27anal2 may cause membrane damage like mellitin. Since the mechanism of Pep27anal 2 peptides is believed to involve disruption of the cell membrane, confocal microscopy used to determine whether Pep27anal2 can enter cells and act as a membrane toxin. Jurkat cells were treated with Pep27anal2-FITC at a concentration of 12.5 μM for 30, 60 and 120 min. After incubation cells were fixed with 3.7% formalin for 30 min and then observed under a confocal microscope. Confocal microscopy showed ring-shaped cells, suggesting that Pep27anal2-FITC is localized to the cell membrane without disturbing its integrity, and then moves from the membrane intracellularly with time (Fig. 2). However, Jurkat cells treated with 12.5 μM Pep27anal2 for more than 60 min showed cell shrinkage and membrane blebbing, suggesting apoptosis, and indicating that Pep27anal2 can enter cells by an unknown mechanism and induce apoptosis whilst maintaining membrane integrity, within 2 hr of treatment.
Table 3 Hemolytic activity of Pep27 and its analogues
Peptide % Hemolysis
0.19 μM (μM) 0.39 μM 0.78 μM 1.56 μM 3.12 μM 6.25 μM 12.5 μM
Pep27 0 0 0 0 0 0 0
Pep27anal1 0 0 0 0 0 0 0
Pep27anal2 0 0 0 0 0 0 18
Pep27anal3 0 0 0 0 0 0 0
Pep27anal4 0 0 0 0 0 0 0
Pep27anal5 0 0 0 0 0 0 0
Mellitin* 0 0 12 30 60 86 100
*, honey bee venom, which translocates across the membrane by forming a pore.
Figure 2 Confocal microscopy of Jurkat cells after treatment with FITC-labeled Pep27anal2 as a function of time. After incubating FITC-labeled Pep27anal2 for 30, 60, and 120 min, cells were fixed with 3.7% formaldehyde and analyzed using a laser scanning confocal microscope. Pep27anal2 peptide expression is shown as green on the left confocal microphotograph. Cells morphology is shown by the phase contrast microphotograph on the right.
Determination of apoptosis after treating with Pep27anal2
One of the earliest events to occur during apoptosis is the externalization of phosphatidylserine, a phospholipid that is normally restricted the inner leaflet of the plasma membrane. Importantly, this precedes membrane damage. Moreover, the externalization of phosphatidylserine can be monitored using annexin V, a phosphatidylserine-specific binding protein [15]. Another important secondary event during apoptosis is the loss of plasma membrane integrity. Intact plasma membranes exclude certain dyes, like PI, the most commonly used marker of membrane integrity [16].
In the present study, Jurkat cells were treated with 100 μM of etoposide, 12.5 μM of mellitin or 12.5 μM or 63 μM of Pep27anal2 for 4 hr. Apoptosis and membrane integrity were then determined using annexin V and propidium iodide, respectively, as described in "Materials and methods". In Fig. 4. living cells are confined to the lower left quadrant. Dead cells appear first in the lower right quadrant (apoptosis), where death was due to the externalization of PS (detected with annexin V-FITC). The upper right quadrant shows cell death resulting from the loss of cell membrane integrity (PI incorporation). Fig. 3 shows that Pep27anal2 caused a higher level of annexin V-FITC staining with a little increase of PI staining than etoposide, whereas mellitin increased the distribution of PI staining without an increase of annexin V-FITC staining.
Figure 3 Determination of apoptosis and membrane integrity by FACS. Cells were treated with etoposide, mellitin and Pep27 anal2. Apoptosis and membrane integrity were determined using annexin V and propidium iodide (PI), respectively, as described in Materials and Methods. A, control; B, 100 μM etoposide; C, 12.5 μM mellitin; D, 12.5 μM Pep27anal2; E, 62.5 μM Pep27anal2. Left, dual staining for annexin V-FITC (FL1-H) and PI (FL2-H); Middle, distribution of annexin V-FITC staining in cell populations; Right, distribution of PI staining. Similar results were obtained in independent duplicated experiments.
Figure 4 Time-dependency on Pep27anal2 – induced apoptosis in Jukart cells. Apoptotic cells were counted among 200–400 cells in each group. The representative electron microscopic findings show apoptosis induced by etoposide or by Pep27anal2 after 4-hr incubation. (a) Control group. Some apoptotic cells were found. (b) Etoposide treated group. Apoptotic cells were more frequent than in the control group. (c) Pep27anal2 treated group. Apoptotic cells were most frequent in this group. Lead citrate and uranyl acetate staining. Scale bar measures 9.9 μm. Arrows indicate apoptotic cells.
An electron microscopic examination revealed that Pep27anal2 induced apoptosis in Jurkat cells, as characterized by condensation of the cytoplasm, cell shrinkage, loss of plasma membrane microvilli, condensed or fragmented nuclei, and the formation of membrane vesicles, as in etoposide (Fig. 4). Cells treated with Pep27anal2 increased apoptosis significantly (P < 0.05) in a time-dependent manner. Moreover, 62.5 μM Pep27anal2 induced more apoptosis than 100 μM of etoposide (Fig. 4).
Pep27anal2 does not induce cytochrome c release from mitochondria to the cytosol in Jurkat cells
Cytochrome c is a well-characterized mitochondrial protein and is involved in cellular energy metabolism. The precursor of cytochrome c, apocytochrome c, is synthesized in the cytoplasm. Upon translocation to mitochoindria, cytochrome c is refolded and acquires a heme moiety, which is required for functionality in the mitochondrial respiration chain. The heme-bound form of cytochrome c is called holocytochrome c, which can activate the caspases responsible for apoptosis by interacting with protease activating factors when released into the cytosol [17]. Jurkat cells (5 × 105/ml) were treated with Pep27anal2 at a concentration of 12.5 μM for 1 h, 2 h, and 4 h and at 62.5 μM for 4 h. After these incubations, cytochrome c in the cytosol fraction, obtained by 14,000 g centrifugation for 15 min, and in the mitosol, obtained by adding 0.1% SDS to the mitochondrial fraction, was determined by Western blotting. Pep27anal2 at 62.5 μM did not induce cytochrome c release from mitochondria into the cytosol within 4 hrs (Fig. 5).
Figure 5 Western blotting of cytochrome c released from mitochondria. Jurkat cells (5 × 105/ml) were treated with Pep27anal2 at 12.5 μM for 1 h, 2 h, and 4 h, and at 62.5 μM for 4 h. Mitochondria were pelleted by centrifugation, and the supernatants retained. As a control, mitochondria were treated with 0.1% (vol/vol) SDS, which induces cytochrome c release from mitochondria. Protein from each fraction was subjected to SDS/PAGE for Western blot analysis using an anti-cytochrome c antibody. 1, the result of analyzing 50 μg of cytosol after adding no Pep27anal2; 2, 50 μg cytosol after 12.5 μM for 1 hr; 3, 50 μg cytosol after 12.5 μM Pep27anal2 for 2 hr; 4, 50 μg cytosol after 12.5 μM Pep27anal2 for 4 hr; 5, 50 μg cytosol after 62.5 μM Pep27anal2 for 4 hr; 6, 50 μg mitosol after 0.1% SDS; 7, 25 μg mitosol after 0.1% SDS.
Solution structures of both Pep27 and Pep27anal2
Complete proton resonance assignments for Pep27 and Pep27anal2 were obtained using the standard sequential resonance assignment procedure. Once the individual spin systems had been classified, backbone sequential resonance assignments were completed by dαN (i,i+1) NOE connectivities in the 2D-NOESY spectra. Figure 6 summarizes the sequential and short-range NOE connectivities observed for Pep27 and Pep27anal2 in TFE/H2O (1v/1v). The observations of continuous dNN(i,i+1) contacts, together with the characteristic dαN(i,i+3) and dαβ (i,i+3) NOEs strongly support the existence of α-helices (Fig. 6). The amide proton temperature coefficient also supports the hypothesis that intramolecular hydrogen bonds stabilize the structures of the Pep27 and Pep27anal2 helices. This result is further supported by the backbone amide proton exchange data of both peptides [18], and the small 3JHNα coupling constants of Pep27anal2 (that of Pep27 could not be measured). The NMR structures of Pep27 and Pep27anal2 were calculated using the experimental restraints derived from the 2D-NOESY spectra. A total of 50 distance geometry geometric structures served as starting structures for the dynamic simulated-annealing calculations for the peptides in TFE/H2O solution. The 20 lowest-energy structures (<SA>k) of 50 simulated annealing structures were selected for detailed structural analysis. A best-fit superposition of 20 <SA>k structures and an energy-minimized average structure () are shown in Fig. 7. The main secondary-structural feature of Pep27 is one α-helix spanning residues, His6 -Leu15 in TFE/H2O (1v/1v), whereas Pep27anal2 has a more stable α-helix spanning Trp 4-Leu15 (Fig. 7).
Figure 6 Summary of the sequential and short-range NOEs for Pep27 and for Pep27anal 2 in TFE/water (1:1 v/v) at pH 7.2 and 15°C, showing sequential and short-range NOE contacts. Amide proton exchange rates (■), backbone vicinal coupling constants (●; 3JHNα < 6 Hz) are also indicated. represents a proline residue The helical region is represented as a long cylinder.
Figure 7 A comparison of the 3D solution structures of Pep27 and Pep27anal2. Final simulated-annealing structures of Pep27 and Pep27anal2. Superposition of the final 20 <SA>k structures of Pep27 and Pep27anal2 over the energy-minimized average structure () is shown. Backbone atoms in Heavy backbone atoms are superimposed for Pep27 His6-Leu15 and Pep27anal 2 residues Trp4-Leu15. The helical ribbon was generated using the MOLMOL program.
Discussion
In this study, we investigated the anticancer activities of Pep27 analogues. Although Pep27 did not show anticancer activity at concentrations greater than 70 μM, its analogues were found to have various cytotoxicities in a number of cancer cell lines. Of the five Pep27 analogue peptides synthesized, Pep27anal2 showed both greatest hydrophobicity and anticancer activity. Our results suggest that substitutions of (2R→W), (4E→W), (11S→W), and (13Q→W) in Pep27 anal2 plays an important role in its hydrophobicity and/or anticancer activity. In particular, the latter two substitutions contribute hydrophobicity or anticancer activity to Pep27anal2, by virtue of the lower anticancer activity of Pep27anal1, containing (2R→W) and (4E→W) substitutions. In addition, substitutions with the more hydrophobic amino acid tryptophan were not always reflected by a higher hydrophobicity or a greater anticancer activity. Pep27anal2 had greater hydrophobicity than other analogues as assessed by RP-HPLC. A positive linear correlation was observed between apoprotein hydrophobicity and column retention time obtained by RP-HPLC [19]. Thus, it could be implicated that the hydrophobicity of peptides, rather than the number of hydrophobic amino acids, may play an important role in anticancer activity or membrane transport. Since peptides cannot readily enter cells, substitution of amino acids not interfering with activity or conjugation with other peptides to facilitate peptide cell entry has been tried. Certain short peptides, which are able to traverse the cell membranes, and which have low lytic activity, can be useful as carriers (vectors) for hydrophilic molecules. Cell penetrating peptides include pAntp ('penetratin'), transportan, MAP (KLAL), Tat (48–60) and VP22. pAntp is a 60-amino acid polypeptide with a sequence that corresponds to the Drosophila antennapedia homeobox gene [20]. Transportan is a 27 amino acid peptide containing 12 functional amino acids from the amino terminus of the neuropeptide galanin and mastoparan at the carboxyl terminus, which are connected by a lysine [21]. MAP (KLAL peptide) is derived from an amphipathic model peptide, KLALKLALKALKAAKLA-NH2 [22]. TAT protein from human immunodeficiency virus (HIV-1), is able to deliver biologically active proteins in vivo, and is of considerable interest for protein therapeutics [23]. Recently the HIV-1 transcription (Tat) protein transduction domain (RKKRRQRRR) has been conjugated into various types of amino acids or proteins [24-26]. Of the cell penetrating peptides, MAP (KLAL) was found to have the fastest uptake, followed by transportan, Tat(48–60), and finally, penetratin [27]. The herpes simplex virus type 1 tegument protein VP22 has recently been shown to mediate the intercellular transport of proteins, raising the possibility that it may be helpful in a setting where the global delivery of therapeutic proteins is desired [28,29].
In the present study, confocal microscopy showed that Pep27anal2 does not act as a membrane toxin like mellitin, but rather penetrates cells by an unknown mechanism, and induces apoptosis as characterized by cell shrinkage and membrane blebbing. Flow cytometry was performed to determine if Pep27anal2 induces apoptosis and membrane damage. Annexin V binding with the cell surface was used as being indicative of apoptosis, in conjunction with a dye exclusion test to establish integrity of the cell membrane. Annexin V is a Ca2+ dependent phospholipid-binding protein with high affinity for phosphatidylserine. However, the translocation of phosphatidylserine to the external cell surface is not unique to apoptosis, and also occurs during necrosis. The difference between these two forms of cell death is that during the initial stages of apoptosis the cell membrane remains intact, while at the very moment that necrosis occurs the cell membrane looses its integrity, becoming leaky [16]. Pep27anal2 increased the annexin V-FITC staining in Jurkat cells, whereas mellitin at the same concentration highly increased the distribution of propidium iodide staining. However, much higher concentrations (62.5 μM) of Pep27anal2 further distribution of annexin V-FITC staining, whereas propidium iodide staining increased only slightly. Cell-penetrating peptides generally have the ability to gain cell entry in an energy-dependent manner [26]. In the present study, Pep27anal2-induced apoptosis was not influenced by pretreating with the ATP depleter sodium azide (data not shown), suggesting an energy-independent transport and/or apoptosis mediated by Pep27anal2. In addition, electron microscopy revealed that Pep27anal2 induced the morphological features of apoptosis in Jurkat cells, and showed cytoplasmic condensation, cell shrinkage, loss of plasma membrane microvilli, condensed or fragmented nuclei, and the formation of membrane vesicles. The highest apoptotic indices were observed for Pep27anal2, and this was statistically significant. Cells undergoing apoptosis in vivo showed increased cytochrome c release to the cytosol, suggesting that mitochondria may be involved by releasing cytochrome c [30]. When cytochrome c is released into the cytosol, it activates the caspases responsible for apoptosis by interacting with protease activating factors [15]. In the cytosol, cytochrome c binds to Apaf-1, which triggers the activation of caspase-9. Caspase-9 is believed to propagate the death signal by triggering six additional caspases (caspases-2, -3, -6, -7, -8, and -10) but not caspases-1, -4, and -5 [31]. However, Pep27anal2 did not appear to induce cytochrome c release from mitochondria within 4 h, suggesting that it induces apoptosis by a cytochrome c-independent mechanism. In addition, the anticancer activity of Pep27anal2 was not abrogated by the pan-caspase inhibitor Z-VAD-fmk (data not shown), indicating that Pep27anal2-induced apoptosis proceeds via a caspase-independent pathway. Like Pep27anal2, there have been reported about apoptosis induced in a caspase-independent fashion without subsequent release of cytochrome c [32,33].
Based on the elucidated solution structures, Pep27 has a short coil-helix-coil composed of residues 6–16. However, the Pep27anal2 analogue characterized by substitutions at 2R→W, 4E→W, 11S→W and 13Q→W was found to have a slightly longer more stable helical conformation than Pep27. Our biological activity data also showed that Pep27anal2 is more potent than Pep27. This may have been due to an increased hydrophobic interaction by Pep27anal2. Therefore, we suggest that helical stability has an important role in the biological activity of Pep27. Taken together, our results indicate that the increased hydrophobicity of Pep27anal2 aids its penetration of the membrane. This is followed by apoptotic induction, but which does not involve cytochrome c release from mitochondria. The hydrophobicity of Pep27anal2 appears to play an important role in its ability to penetrate the plasma membrane and/or in its anticancer activity. Thus, we believe that Pep27anal2 is a potential candidate for the design of anticancer peptides.
Abbreviations
RP, reverse phase ; MALDI, matrix-assisted laser desorption ionization; PI, propidium iodide; DSS, sodium 2, 2-dimethyl-2-silapentane-5-sulfonic acid; MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide; TFE, 2, 2, 2-trifluoro-(d3)-ethanol; PBS, phosphate-buffered saline; TBS, Tris-buffered saline; SDS, sodium dodesyl sulfate
Acknowledgements
This work was supported, in part, by grants from the Ministry of Science and Technology, Korea and the Korea Science and Engineering Foundation through the Research Center for Proteineous Materials from the Life Phenomena & Function Research Group program (01-J-LF-01-B-09) as well as the Research Center for Resistant Cells (R13-2003-009), and from the NRL program of MOST NRDP (M1-0203-00-0020) (W.L). The authors wish to thank Mr. Su-Man Jung for his expert technical assistance during the electron microscopic examination.
==== Refs
Fernandez RC Weiss AA Susceptibilities of Bordetella pertussis strains to antimicrobial peptides Antimicrob Agents Chemother 1996 40 1041 1043 8849226
Sharom FJ Lu P Liu R Yu X Linear and cyclic peptides as substrates and modulators of P-glycoprotein: peptide binding and effects on drug transport and accumulation Biochem J 1998 333 ( Pt 3) 621 630 9677321
Baker MA Maloy WL Zasloff M Jacob LS Anticancer efficacy of Magainin2 and analogue peptides Cancer Res 1993 53 3052 3057 8319212
Moore AJ Devine DA Bibby MC Preliminary experimental anticancer activity of cecropins Pept Res 1994 7 265 269 7849420
Soballe PW Maloy WL Myrga ML Jacob LS Herlyn M Experimental local therapy of human melanoma with lytic magainin peptides Int J Cancer 1995 60 280 284 7829229
Novak R Charpentier E Braun JS Tuomanen E Signal transduction by a death signal peptide: uncovering the mechanism of bacterial killing by penicillin Mol Cell 2000 5 49 57 10678168 10.1016/S1097-2765(00)80402-5
Linton KJ Higgins CF The Escherichia coli ATP-binding cassette (ABC) proteins Mol Microbiol 1998 28 5 13 9593292 10.1046/j.1365-2958.1998.00764.x
Merrifield B Solid phase synthesis Science 1986 232 341 347 3961484
Pieters R Huismans DR Leyva A Veerman AJ Adaptation of the rapid automated tetrazolium dye based (MTT) assay for chemosensitivity testing in childhood leukemia Cancer Lett 1988 41 323 332 3165705 10.1016/0304-3835(88)90294-7
Subbalakshmi C Krishnakumari V Nagaraj R Sitaram N Requirements for antibacterial and hemolytic activities in the bovine neutrophil derived 13-residue peptide indolicidin FEBS Lett 1996 395 48 52 8849687 10.1016/0014-5793(96)00996-9
Bossy-Wetzel E Newmeyer DD Green DR Mitochondrial cytochrome c release in apoptosis occurs upstream of DEVD-specific caspase activation and independently of mitochondrial transmembrane depolarization Embo J 1998 17 37 49 9427739 10.1093/emboj/17.1.37
Marion D Wuthrich K Application of phase sensitive two-dimensional correlated spectroscopy (COSY) for measurements of 1H-1H spin-spin coupling constants in proteins Biochem Biophys Res Commun 1983 113 967 974 6307308 10.1016/0006-291X(83)91093-8
Piotto M Saudek V Sklenar V Gradient-tailored excitation for single-quantum NMR spectroscopy of aqueous solutions J Biomol NMR 1992 2 661 665 1490109
Kiyota T Lee S Sugihara G Design and synthesis of amphiphilic alpha-helical model peptides with systematically varied hydrophobic-hydrophilic balance and their interaction with lipid- and bio-membranes Biochemistry 1996 35 13196 13204 8855958 10.1021/bi961289t
Koopman G Reutelingsperger CP Kuijten GA Keehnen RM Pals ST van Oers MH Annexin V for flow cytometric detection of phosphatidylserine expression on B cells undergoing apoptosis Blood 1994 84 1415 1420 8068938
Vermes I Haanen C Steffens-Nakken H Reutelingsperger C A novel assay for apoptosis. Flow cytometric detection of phosphatidylserine expression on early apoptotic cells using fluorescein labelled Annexin V J Immunol Methods 1995 184 39 51 7622868 10.1016/0022-1759(95)00072-I
Hampton MB Zhivotovsky B Slater AF Burgess DH Orrenius S Importance of the redox state of cytochrome c during caspase activation in cytosolic extracts Biochem J 1998 329 ( Pt 1) 95 99 9405280
Wishart DS Sykes BD Richards FM The chemical shift index: a fast and simple method for the assignment of protein secondary structure through NMR spectroscopy Biochemistry 1992 31 1647 1651 1737021 10.1021/bi00121a010
Weinberg R Patton C DaGue B Analytic and preparative separation of apolipoproteins A-I, A-II, and A-IV by reverse phase high pressure liquid chromatography J Lipid Res 1988 29 819 824 3139814
Joliot A Pernelle C Deagostini-Bazin H Prochiantz A Antennapedia homeobox peptide regulates neural morphogenesis Proc Natl Acad Sci U S A 1991 88 1864 1868 1672046
Pooga M Hallbrink M Zorko M Langel U Cell penetration by transportan Faseb J 1998 12 67 77 9438412
Dathe M Schumann M Wieprecht T Winkler A Beyermann M Krause E Matsuzaki K Murase O Bienert M Peptide helicity and membrane surface charge modulate the balance of electrostatic and hydrophobic interactions with lipid bilayers and biological membranes Biochemistry 1996 35 12612 12622 8823199 10.1021/bi960835f
Green M Loewenstein PM Autonomous functional domains of chemically synthesized human immunodeficiency virus tat trans-activator protein Cell 1988 55 1179 1188 2849509 10.1016/0092-8674(88)90262-0
Datta K Sundberg C Karumanchi SA Mukhopadhyay D The 104-123 amino acid sequence of the beta-domain of von Hippel-Lindau gene product is sufficient to inhibit renal tumor growth and invasion Cancer Res 2001 61 1768 1775 11280720
Jin LH Bahn JH Eum WS Kwon HY Jang SH Han KH Kang TC Won MH Kang JH Cho SW Park J Choi SY Transduction of human catalase mediated by an HIV-1 TAT protein basic domain and arginine-rich peptides into mammalian cells Free Radic Biol Med 2001 31 1509 1519 11728823 10.1016/S0891-5849(01)00734-1
Kwon HY Eum WS Jang HW Kang JH Ryu J Ryong Lee B Jin LH Park J Choi SY Transduction of Cu,Zn-superoxide dismutase mediated by an HIV-1 Tat protein basic domain into mammalian cells FEBS Lett 2000 485 163 167 11094160 10.1016/S0014-5793(00)02215-8
Hallbrink M Floren A Elmquist A Pooga M Bartfai T Langel U Cargo delivery kinetics of cell-penetrating peptides Biochim Biophys Acta 2001 1515 101 109 11718666
Phelan A Elliott G O'Hare P Intercellular delivery of functional p53 by the herpesvirus protein VP22 Nat Biotechnol 1998 16 440 443 9592391 10.1038/nbt0598-440
Lai Z Han I Zirzow G Brady RO Reiser J Intercellular delivery of a herpes simplex virus VP22 fusion protein from cells infected with lentiviral vectors Proc Natl Acad Sci U S A 2000 97 11297 11302 11027330 10.1073/pnas.97.21.11297
Liu X Kim CN Yang J Jemmerson R Wang X Induction of apoptotic program in cell-free extracts: requirement for dATP and cytochrome c Cell 1996 86 147 157 8689682 10.1016/S0092-8674(00)80085-9
Slee EA Harte MT Kluck RM Wolf BB Casiano CA Newmeyer DD Wang HG Reed JC Nicholson DW Alnemri ES Green DR Martin SJ Ordering the cytochrome c-initiated caspase cascade: hierarchical activation of caspases-2, -3, -6, -7, -8, and -10 in a caspase-9-dependent manner J Cell Biol 1999 144 281 292 9922454 10.1083/jcb.144.2.281
Hahn HP Pang M He J Hernandez JD Yang RY Li LY Wang X Liu FT Baum LG Galectin-1 induces nuclear translocation of endonuclease G in caspase- and cytochrome c-independent T cell death Cell Death Differ 2004
Ohtsuka T Zhou T Bisindolylmaleimide VIII enhances DR5-mediated apoptosis through the MKK4/JNK/p38 kinase and the mitochondrial pathways J Biol Chem 2002 277 29294 29303 12034736 10.1074/jbc.M203342200
|
16004618
|
PMC1199608
|
CC BY
|
2021-01-04 16:24:16
|
no
|
Cancer Cell Int. 2005 Jul 11; 5:21
|
utf-8
|
Cancer Cell Int
| 2,005 |
10.1186/1475-2867-5-21
|
oa_comm
|
==== Front
J CarcinogJournal of Carcinogenesis1477-3163BioMed Central London 1477-3163-4-121609295610.1186/1477-3163-4-12ResearchCharacterization of insulin-like-growth factor II (IGF II) mRNA positive hepatic altered foci and IGF II expression in hepatocellular carcinoma during diethylnitrosamine-induced hepatocarcinogenesis in rats Mukherjee Biswajit [email protected] Shampa [email protected] Tanushree [email protected] Manika [email protected] Department of Pharmaceutical Technology, Jadavpur University, Kolkata 700 032, India2 Department of Biochemistry and Nutrition, All India Institute of Hygiene and Public Health, C.R. Avenue, Kolkata 700 073, India2005 10 8 2005 4 12 12 6 7 2005 10 8 2005 Copyright © 2005 Mukherjee et al; licensee BioMed Central Ltd.2005Mukherjee et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Insulin-like-growth factor II (IGF II) has been implicated in the pathogenesis of neoplasm of different tissues, including liver of rats and men. This growth factor is believed to exert its effect during cellular proliferation. During the process of development of hepatocellular carcinoma (HCC), different hepatic altered foci appear. They are believed to be the putative precursors of HCC in rats and in men. Thus, to study the role of the gene in a defined model of hepatocarcinogenesis was the target to elucidate its role in various cancer phenotypes during the entire development stage of cancer, right from earlier preneoplastic lesions to HCC
Methods
Antisense in situ hybridization technique was used here to characterize the type(s) of foci in which IGF II mRNA had expressed during the development of hepatocarcinogenesis-induced by diethylnitrosamine and promoted by phenobarbital in rats. Various focal lesions have been categorized depending on the stages and sizes along with IGF II expression patterns in them. Immunohistochemical detection for proliferating cell nuclear antigen (PCNA) was made to detect the role of the gene in preneoplastic and neoplastic cellular proliferation.
Results
IGF II expression was located in the glycogen-storage acidophilic cell foci maximally followed by mixed cell lesions and the least in basophilic lesions. The expression of IGF II was found to be predominant in the HCC. The expression of gene was also located at the peripheral cells of spongiosis hepatis which are believed to be the precursor of ito cell carcinoma. It was noted that there is a direct correlation between IGF II expression and Immunohistochemical detection for PCNA.
Conclusion
It may be concluded that IGF II gene expression plays an important role during the development of neoplasia and the gene expresses in the sequence of events leading from glycogen-rich-acidophilic lesions to glycogen poor basophilic lesions to HCC with an expression pattern of “high-low-high” in terms of degree of expression. Moreover, the essential role of the gene at the immediate initiation stage of carcinogenesis (first few weeks) and during HCC development cannot be ignored. Thus this expression can be used as a suitable marker for very early detection of the cancerous process and can save numbers of future cancer victims by very early detection of this disease.
IGF IIhepatocarcinogenesisratmRNAhepatic altered foci
==== Body
Background
Insulin-like-growth factor II (IGF II)b [somatomedin A ] is a mitogenic polypeptide having structural similarity to proinsulin and insulin-like-growth factor I (IGF I) [somatomedin C] [1,2]. IGF II is known to be widely distributed in various fetal and neonatal human and rat tissues, including liver [3-5], but restricted in the adult rat tissues [6], except brain [7]. It has also been implicated in the pathogenesis of neoplasm of different tissues, including liver of rats and men [8-11]. The growth factor is believed to exert their effect during cellular proliferation through endocrine, autocrine and paracrine mechanisms [12]. In breast cancer epithelial cells, IGFII was reported to serve as autocrine growth stimulus and IGFII overexpression was suggested to be capable of mediating malignant progression [13]. Expression of IGFII in rodents remains high around birth and becomes undetectable within three weeks [14]. But the reactivation of this gene has been reported in variety of neoplasms including HCC in rats. IGF II expression has been shown to be reactivated during progression of neoplastic nodules to HCC [15]. IGFII reactivation was reported to be a common phenomenon in hepatocarcinogenesis irrespective of species and the process of hepatocarcinogenesis [11]. The reactivation of endogenous IGF II was suggested to exert a growth stimulatory effect via IGFI and IGFII/mannose 6 phosphate receptors [16]. It was stated that IGFII might trigger a proliferating signal through hybrid insulin/IGFI receptors stimulated by exogenous IGF [17]. Carcinogen treatment may generate IGFII microenvironment in preneoplastic foci that lead to selective growth advantage for this cell.
Chemical hepatocarcinogenesis is a multistep and complex process and is a favorite model in rat to study the mechanism of transformation of a normal cell to malignant population(s) [18]. During the process of development of HCC from the normal hapatocytes, hepatic altered foci appear. These foci are the putative precursors of HCC in rat, and probably in man [19,20]. Depending on the morphological and biochemical hepatocellular changes during the process of development from normal hepatocytes to HCC induced by various chemical carcinogens, radiation, hormones, chronic viral hepatitis and transgenic manipulation in rodents, three main altered hepatocellular lesions-the glycogenotic, mixed cell and basophilic cell lesions have been distinguished [21]. The predominant sequence of cellular changes starts with glycogenotic clear and acidophilic (smooth endoplasmic reticulum- rich) hepatocytes and progresses through intermediate phenotypes in mixed cell populations to glycogen-poor, homogeneously basophilic (ribosome -rich) cellular phenotypes prevailing in undifferentiated HCC.
Elevated levels of IGF II gene and protein expression have been found in a wide range of human and rodent tumors [8-10]. IGF II has been claimed to be one of the most important growth factors, playing role during the progression of neoplasm [22]. But the advent of IGF II gene expression during the development of neoplasia has remained to be elucidated. Antisense in situ hybridization technique has been used here to explore the type(s) of foci in which IGF II mRNA is expressed at the stages of neoplastic conversion of hepatocytes during diethylnitorsamine (DENA)-induced and phenobarbital promoted hepatocarcinogensis in Sprague-Dawley rats.
Materials and methods
Experiments were conducted on the hepatic tissues of male Sprague-Dawley rats (purchased from the Indian Institute of Chemical Biology, Kolkata, India). The initial body weights of the animals were approximately 130 g. They were maintained under constant temperature (22°C) and humidity (55%RH) conditions and were fed basal diet [23] and had free access to water. Animals were divided in four groups-Group A (having 15 animals), group B, C and D (having 10 animals in each of the three groups). Group A animals were normal untreated animals. Group B, C and D animals were carcinogen treated groups. All carcinogen treated animals were treated with DENA, 200 mg/kg body weight, intraperitoneally at the start of the experiment, i.e., day zero. Group C and D animals were promoted with 0.05 % phenobarbital in basal diet [24]. Group B animals were sacrificed at the 8th week to study earlier preneoplastic lesions, group B animals were sacrificed at 22nd weeks to study mixed and basophilic lesions and group C animals were sacrificed at 36th week as HCC development are reported to take place between 32–36 weeks [25]. In each case five normal control animals were sacrificed with the respective treated groups B, C and D for comparison. All animals were unfed for 12 h before they were sacrificed by cervical dislocation and the animals were sacrificed between 9 and 10 a.m. The livers were removed, sliced (5–10 mm thick) and were snap-frozen in isopentane precooled with liquid nitrogen at – 150°C. The tissues were stored at – 80°C until further used.
Serial sections were prepared from the frozen materials with a Reichert-Jung microtome. They were studied histochemically by periodic acid Schiff reaction, toluidine blue and hematoxylin-eosin to detect different hepatic altered foci (26, 27). Preneoplastic lesions were eventually divided into three categories in accordance with their respective size and the total area of the occupied liver parenchyma (<1 mm, >1 mm-<3 mm, >3 mm) as described previously [28]. These serial sections were used for in situ hybridization with sense and anitsense digoxigenin labeled mRNA probes of IGF II and immunohistochemical reaction for proliferating cell nuclear antigen (PCNA).
In vitro transcription of Digoxigenin (DIG)-labeled sense and antisense IGF II mRNA
A 860 base pair (bp) coding region fragment from rIGF II cDNA extended between EcoRI- EcoRI ploylinker site of pGEM-4 vector (Source: H. Hacker, DKFZ, Germany, obtained from A Riccio, Universita degli studi di Napoli Federico II, Napoli, Italy) was transformed into XL1-blue E.coli competent cells. The plasmid was also religated and propagated in the competent cells to obtain the opposite orientation of the 860 bp coding fragment of rIGF II cDNA. SP6-RNA polymerase was used to generate both sense and anitsense riboprobes from Xba I linearized plasmids of the opposite orientation. In vitro transcription was conducted, using digoxigenin (DIG)-RNA labeling Kit from Boehringer-Mannheim, Mannheim, Germany. The concentrations of the sense and antisense probes were estimated by using Nucleic Acid Detection Kit (Boehringer- Mannheim, Mannheim, Germany). The method is based upon the comparison between the colour intensities of slot blots and DIG labeled standard (Boehringer- Mannheim, Mannheim, Germany) (Figure 1) and the experimental riboprobes. By electrophoresing the probes on formaldehyde containing 1% agarose gel the purity of the labeled probes were checked (Figure 2).
Figure 1 Quantification of DIG-labeled sense and antisense IGFII mRNA by spot test against standard labeled mRNA (Boehringer-Mannheim RNA detection kit).
Figure 2 Detection of DIG-labeled IGFII sense and antisense mRNA (860 bp) using gel electrophoresis. Lane 1; DIG-labeled sense IGFII mRNA, Lane 2; DIG-labeled antisense IGFII mRNA.
Detection of IGF II mRNA by In situ hybridization
In situ hybridization was conducted on 6 μm cryosections of the liver samples by using a modified procedure of Braissant and Walhi [29]. The cryosections were first fixed at 50°C for 30 seconds on a hot plate and delipidified with chloroform for 5 min whenever necessary. The sections were then fixed in 4% paraformaldehyde in phosphate buffer saline, pH 7.4, (PBS) for 1 hr. The slides were then treated with PBS containing active DEPC (diethyl pyrocarbonate) for 30 min followed by a wash of 4 × SSC of 15 min. Prehybridization was conducted using hybridization buffer (mentioned below) except dextran sulphate and Denhardt's solution for 1 hr. at 52°C. This was followed by hybridization at 50°C overnight in a humid chamber. Every 10 ml of hybridization buffer contained 5 ml deionized formamide, 2.4 ml DEPC treated water, 2 ml 20 × SSC, pH 7.0, 1 g dextran sulphate, 200 μl 50 × Denhardt's solution, denatured yeast t-RNA (10 mg/ml) 400 μl and denatured salmon sperm DNA (10 mg/ml) μl along with respective probes denatured by heating at 80°C for 3 min., followed by immediate cooling on ice. After hybridization, sections were washed in 2 × SSC at room temperature (RT) for 30 min, then again with 2 × SSC at 60°C, for 30 min. This was followed by 1 × SSC for 30 min and 0.1 × SSC for 1 hr, at 60°C with a frequent moderate shaking. After the stringency wash, the slides were incubated with Tris-NaCl buffer, pH 7.5 containing 0.5% blocking agent (Boehringer-Mannheim, Mannheim, Germany) at RT on a shaking platform with a low speed. Sections were then incubated with anti-DIG-polyclonal antibody conjugated with alkaline phosphatase (Boehringer-Mannheim, Mannheim, Germany) (1:5000) in Tris-NaC1 buffer, pH 7.5 for 2 hr at RT on the shaking platform with a low speed. After incubation, sections were washed with washing buffer (Tris-NaC1 buffer, pH 7.5, containing 0.3% tween 20) for 10 min at RT. This was followed by two 15 min washes with Tris-NaC1 buffer, pH 7.5. the section were equilibrated with detection buffer (100 mM Tris-HC1, 100 mM NaC1 and 50 mM MgC12,6H2O), pH 9.5 for 5 min and staining was done using nitroblue tetrazolium/ bromochloroindolylphosphate (NBT/BCIP) in detection buffer. Counter staining was also done occasionally.
Immunohistochemical detection for proliferating cell nuclear antigen
PCNA was demonstrated by following the methods of Enzmann et al. [30] and Bannasch et al [31]. Briefly 6 μm cryostat sections were air-dried after fixation in acetone for 10 min followed by a treatment with 2% H2O2 in methanol to block endogenous peroxidases. Then the sample was incubated in 3% bovine serum albumin for 30 min to prevent unspecific reactions followed by incubation with mouse monoclonal antibody to PCNA, diluted 1:1600 (PC10 Dako), overnight at 4°C. PCNA positive nuclei were identified by using alkaline phosphatase labeled streptavidine-biotin complex (Dako LASB2 Kit, AP). Then the sections were stained with new-fuschin and Myer's haemetoxylin.
IGF II mRNA positive foci were estimated in randomly selected areas of each section containing hepatic altered foci from experimental animals. The results were statistically evaluated using ANOVA followed by Fisher's exact Test otherwise mentioned.
Results
The morphometric determination of the focal lesions showed that incidences of preneoplastic lesions (i.e. glycogen storage foci/acidophilic foci [Figure 3], mixed cell focal lesions and basophilic focal lesions [Figure 4] were 100% in group B, C and D animals. The numerical densities and size distributions of preneoplastic lesions were depicted in Table 1. Average numerical density was found to be maximum (1498 ± 33 per cm3 of hepatic tissue) in case of group D animals, which was followed by group C (968 ± 16 per cm3 of hepatic tissue) and group B (412 ± 28 per cm3 of hepatic tissue) animals. The preneoplastic lesions were categorized with respect to their sizes and their areas of the occupied liver parenchyma. It was observed that group B animals in which carcinogenesis had been initiated with DENA and animals were sacrificed after 8 weeks, had 81% focal lesions having size less than 1 mm i.e. smaller lesions and 19% focal lesions were in the size range of 1 mm-3 mm. Group C had 52% lesions less than 1 mm. 44% lesions were between 1 mm and 3 mm and sizes of 4% lesions were greater than 3 mm. In group D animals, seven animals out of ten were found to develop HCC (Table 2) and they had more intermediate (1 mm-3 mm) and large (greater than 3 mm) lesions as compared to smaller lesions.
Table 1 Incidences, numbers and size distributions of preneoplastic lesions (glycogen-storage, mixed cell and basophilic type) in the hepatic tissues of male Sprague-Dawley rats in different experimental groups (B, C and D).
Group No. of rats with preneoplastic lesions/Total numbers of rats Incidences (%) Numerical densities of preneoplastic lesionsa,b Size distributions of preneoplastic lesions (% of total numbers)
<1 mm >1 mm-<3 mm >3 mm
B 10/10 100 412 ± 28 81 19 -
C 10/10 100 968 ± 16 52 44 4
D 10/10 100 1498 ± 33 20 48 32
a Numerical density, expressed in number per cm3 of hepatic tissue
b Each value represent mean of 10 rats per group ± SD
Table 2 Expression of IGF II gene in different types of preneoplastic hepatic altered focal lesions (HAF) as well as carcinoma in hepatic tissues of the experimental animals (group B, C and D)
Types of HAF which expressed IGF II gene Numbers of rats with HAF and/or HCC per total no of animals Number of IGF II expressed lesions/100 respective HAF lesions
Glycogen-storage focal lesion 10/10 98 ± 2.98a,*
Mixed cell focal lesion 10/10 74 ± 8.26a,*
Basophilic cell focal lesion 10/10 05 ± 2.68a,NS
Hepatocellular carcinoma 7/10 95 ± 4.02%b,*
a Mean ± SD (n = 10)
b Mean ± SD (n = 7)
* P < 0.001
NS- Non-significant
Figure 3 Glycogen storage foci (shown by yellow arrows) in rat hepatic tissue with Periodic Acid Schiff reaction, ×100.
Figure 4 Mixed cell focal lesion (shown by green arrows) and basophilic lesion (shown by blue arrows) in DENA treated rat liver, ×100.
When IGFII gene expressed, preneoplastic lesions were compared to the various type of hepatic altered foci (HAF) such as glycogen storage focal lesions, mixed cell focal lesions, and basophilic focal lesions, it was found that 98% glycogen storage focal lesions expressed IGFII [Figures 5, 6, 7]. Again 74% of the mixed cell focal lesions expressed IGFII where once again mainly glycogen storage hepatocytes were found to express IGFII not the basophilic cells. Very few basophilic cell focal lesions with IGFII gene expression were detected. In those lesions, there were, some glycogen containing cells (as seen with PAS staining). As the lesions containing predominantly basophilic cells, they were categorized under basophilic lesions. Thus it may be believed that glycogen storage cells showed IGFII expression in those basophilic lesions. Only 5% of the basophilic hepatic altered foci were found to express IGFII although the results are statistically non-significant. As described earlier in group D animals, seven out of ten animals were found to develop HCC and they had some high expression in the peripheral hepatocytes of tumor. There was also some expression within the tumor areas [Figures 8, 9, 10, 11].
Figure 5 Hepatic section of DENA treated experimental animals showing glycogen storage foci with PAS staining, ×100.
Figure 6 IGFII mRNA expressed glycogen storage lesion detected with Digoxigenin-labeled antisense IGFII, mRNA ×100.
Figure 7 Glycogen storage lesion shown after Digoxigenin-labeled sense IGFII mRNA treatment during in situ hybridization method ×100. Pv- portal vein. Bd – bile duct.
Figure 8 Section of hepatic tumor (shown by arrows) in HCC; using toluedine blue ×40.
Figure 9 Section of hepatic tumor (shown by arrows) in HCC; using PAS reaction ×40.
Figure 10 Section of hepatic tumor (shown by arrows) in HCC; IGFII mRNA expression detected with DIG-labeled IGFII antisense mRNA in tumor as well as peripheral tissue. ×40.
Figure 11 IGFII mRNA expression in hepatocytes in tumor area during HCC in experimental animals (see protocols) ×97.5.
Hepatocytes proliferation was established by demonstration of PCNA (Proliferating Cell Nuclear Antigen). PCNA positive hepatocytes and non-hepatocytes were detected in carcinogen treated groups as well as normal untreated groups. Five fold higher numerical value of PCNA positive cells was detected in the carcinogen treated animals as compared to normal untreated animals (Table 3) and PCNA positive cells in carcinogen treated animals were mostly (about 85%) confined to lesion area. Interestingly the counts were greater in glycogen storage lesions as well as in HCC (data not shown). This indicates that proliferation was higher during the early stage i.e. at the immediate post initiation stage as well as in HCC.
Table 3 Number of PCNA (Proliferating Cell Nuclear Antigen) positive liver cells per 103 cells
Labeling Indexa
Group HAF Hepatocytes Non focal-hepatocytes
Carcinogen-treated 667 ± 28* 122 ± 17*
Normal Control 132 ± 12 88 ± 14
a results show mean ± SD (n = 10)
* P < 0.001 as compared to normal control animals, assessed by Student's t-test
In the hepatic tissue of group D animals some scattered spongiotic pericytoma have been noticed. This spongiotic pericytoma are also believed to be the precursor of perisinusoidal cell sarcomas [32]. They were very few in numbers but were found in the hepatic tissue of all the animals in group D. But they were not found in every tissue slide of the same animals. When IGFII expression of the subsequent slides were studied it was observed that IGFII gene did not express in the area of spongiotic hepatis. But high IGFII expression was noticed at the peripheral hepatocytes [Figures 12, 13, 14, 15]. PCNA positive cells were also detected predominantly in the peripheral cells of spongiosis hepatis.
Figure 12 Serial sections of experimental rat hepatic tissue showing spongiosis hepatis with haematoxylin and eosin (shown by white arrows) ×40.
Figure 13 Section of experimental rat hepatic tissue showing spongiosis hepatis with toluedine blue ×120.
Figure 14 Section of experimental rat hepatic tissue showing spongiosis hepatis with PAS reaction) ×40.
Figure 15 IGFII mRNA expression detected by DIG- labeled antisense IGFII mRNA in the peripheral cells of the spongiocyst hepatis ×120.
Discussion
A predominant sequence of cellular changes starting from the appearance of glycogenotic foci to the glycogen-poor basophilic foci, leading to HCC via various mixed cell populations, has been observed in rodents during the development of HCC induced by various chemical carcinogens, hormones, radiation, chronic viral infection and transgenic manipulation [19,20]. During this process several phenotypic changes appear which suggest a progressive cellular dedifferentiation.
The DIG detection system used here has been claimed to offer a much higher sensitivity and detectability than radioactive detection systems [33,34]. Using this system, we tried to localize the IGF II gene expression in different type(s) of foci during DENA-induced hepatocarcinogenesis. Predominant expression of the gene was located in hepatic altered glycogen-storage foci which has been claimed to be early preneoplastic lesions [32] along with some expressions in foci of mixed cell populations. High expression was also detected in HCC. It was higher at the periphery as well as the glycogen storage cells of the tumor. This observation is supported by the observation of Sohda et.al who demonstrated that the localization of IGF II mRNA in areas consisting of less differentiated tumor were present at the periphery of the tumour nest [35]. Higher expression was detected in the foci of glycogen storage altered hepatocytes as compared to the tumor cells. This suggests that there is a differential expression of IGF II mRNA during the development of preneoplasia to neoplasia. Fetal transcripts of IGF II express high in man and rodent liver cancer [6]. In human, fetal transcripts express highly in the HCC, suggesting that IGF II-expression is late event during the development of liver cancer [36]. On the other hand, an inverse correlation was observed between IGF II-expression and cellular differentiation in rat [37]. Again, it has been found that the IGF II expression is predominant at the fetal and neonatal life of rat and it declines with the days after birth when IGF I is found to be increased and believed to take up the role of IGF II [6]. Simultaneously increased pattern of expressions of IGF I and TGFα, during some preneoplastic and neoplastic conditions [35] also suggests that the IGF I and TGFα may counterbalance the activity of IGF II.
Amongst the various methods available, two important methods of studying cancer cell proliferation are measuring the rate of DNA synthesis and the immunohistochemical reaction for proliferating cell nuclear antigen (PCNA) [30,31]. Demonstration of PCNA has been reported to be a valuable non-tracer method for the detection of proliferating cells. Report suggests the method is used to detect proliferating cells in S phase, G2 and late G1 phases of the cell cycles [32]. PCNA positive liver cell counts were found to be significantly higher (p < 0.001) in this study in carcinogen treated animals as compared to normal control animals. Higher PCNA labeling index were noticed in IGF II expressed lesions (mostly glycogen-rich lesions) as compared to the other IGF II non-expressed or little-expressed hepatic lesions (mixed cells and basophilic cell lesions) in hepatic areas studied. This suggests that the IGF II expressed cells in all the lesions are more proliferating in nature as compared to the other focal lesions and of course the extra-focal hepatic areas. The labeling index was found to be higher again in HCC. This indicates that cancer development is faster during glycogen-rich focal development (they predominantly expressed IGF II) and during the stage of HCC.
Spongiosis hepatis have been found interestingly only in-group C animals (in six out of ten animals). They were found to be very scattered in nature. In whole observation they were only 18 in numbers (in 50 different hepatic tissue samples of the seven animals of group C). Thus it seems that they appear in progression phase of cancerous process, at least during DENA-induced hepatocarcinogenesis. Spongiosis hepatis also described as spongiotic pericytoma are believed to be the possible precursor of malignant tumors [32]. It has also been classified as perisinusoidal (ito) cell sarcomas [33]. Lesions classified as spongiosis hepatis have been reported to progress to pericytomas which have been related to perisinusoidal cells (ito cells). Thus the study of these cells may help to understand neoplastic transformation of them. In present study efforts have been made to characterize various preneoplastic hepatic altered foci and hepatic cellular carcinoma along with spongiosis hepatis. Spongiosis hepatis have been detected using haematoxylin and eosin, toluidine blue and PAS staining. Consecutive sections were compared with the IGFII expressions and it has been observed that the predominant expression of IGFII were noticed in the peripheral hepatocytes but not in the holes of the spongiotic formations which are believed to be filled with a finely flocculent material that shows metachromasia (Acid mucopolysaccharide deposit) [32]. The IGFII expression seems to be dominant more in the peripheral hepatocytes of the spongiocyst hepatis. PCNA positive cells were also detected in the surrounding areas where the IGFII expressions were noticed. Again IGFII expression has been claimed to be associated with the development of hepatocellular carcinoma (HCC) by several authors [34,35]. Thus the study further supports that the spongiosis hepatis are possibly the origin of malignant tumor [32]
The claims related to an amount of over expression of IGF II gene in liver tumors and pre-neoplastic hepatic lesions during hepatocarcinogenesis in animal models and in human HCC were highly variable. A possible involvement of an IGF II autocrine loop in the pathogenesis of hepatic preneoplastic and neoplastic lesions have been linked with the association of allelic imbalance with an increased expression of the IGF II gene in pre-cancerous lesions [36,42].
The work has demonstrated that the allelic imbalance of the IGF II gene expression was seen in the early pre-cancerous lesions during hepatocarcinogenesis; but was not observed in well and moderately differentiated HCC.
Thus, some workers considered IGF II gene expression as an early event. Mixed reports are found which claimed the expression of IGFII either in preneoplasia [37,38] or during HCC [34,35]. The over expression of IGF II in HCC has been claimed to be associated with re-expression pattern of IGF II transcript occurring through activation of fetal promoters p2 – p4 with a loss of activity of the adult promoter p1 [36]. Interestingly, the over expression of IGF II was found in the preneoplastic glycogen storage focal lesions maximally as compared to that in the basophilic lesions.
In mixed cell focal lesions as well as basophilic lesions, the expressions were low and in the mixed cell focal lesions the expression was particularly confined to the glycogen rich cells. When the expression was predominant almost throughout a lesion, this has been considered as an IGF II expressed lesion in this study. So, the trend of IGF II over expression in our study was in the sequence of "high-low-high" in "glycogen storage foci – mixed cell foci – basophilic lesion – HCC". This suggests that the IGF II over expression is not only the early event, but the late event too. The gene expression amongst others appears to be responsible for cancer initiation process as well as in the progression of HCC. But its role during the progression from preneoplasia to neoplasia remains to be elucidated. Thus this gene over expression or may be the protein over expression can be used to detect the cancer at the early stage or at HCC and hence it may be considered as a future diagnostic means for detection of hepatic cancer and its stages and IGFII gene 33 expression could be considered as one of the best positive markers for early detection of putative preneoplastic cell as well as HCC in chemically induced hepatocarcinogenesis. Further investigation of this gene expression is warranted in the interest of better cancer detection. Thus this may lead to better cancer prevention. Again in contrast to many other gene or protein expressions IGFII has particular advantage for its use in screening of potential carcinogens and promoters since it is not expressed in lesions with some non-xenotoxic chemicals [39]. Further IGFII gene was reported to express at high levels in both spontaneous [40] and induced [41] hepatic lesions at the early stages of development [37,38] and in HCC [32,33].
Acknowledgements
The work has been carried out with the financial grants (UGC, grant no.7-5/2004(SR), ICMR grant no. 45/11/2003-PHA/BMS and DAF grant no. A/97/00771)
==== Refs
Daughaday WH Rotwein P Insulin-like-growth factors I and II peptide, messenger ribonucleic acid and gene structures, serum and tissue concentrations Endocrinol Rev 1989 10 68 91
Rechler MM Nissley SP The nature and regulation of the receptors for insulin-like-growth factors Ann Rev Physiol 1985 47 425 442 2986537 10.1146/annurev.ph.47.030185.002233
Brown AL Graham DE Nissley SP Hill DJ Strain AJ Rechler MM Developmental regulation of insulin like growth factor II mRNA in different rat tissues J Biol Chem 1986 261 445 462
Humbel RE Insulin like growth factor I and II Eur J Bioche 1990 190 445 462 10.1111/j.1432-1033.1990.tb15595.x
Cariani E Lasserre C Seurin D Hamelin B Kemeny F Franco D Czech MP Ullrich A Brechot C Differential expression of insulin like growth factor II mRNA in human primary liver cancer, benign tumors and liver cirrhosis Cancer Res 1998 48 6844 6849 3180092
Moses AC Nissley SP Short PA Rechler MM White RM Knight AB Higa OZ Increased level of multiplication-stimulating activity or insulin-like-growth factor in fetal rat serum Proc Natl Acad Sci USA 1998 77 354 365
Soares MB Ishii DN Efsratiadis A Development and tissue specific expression of transcripts related to the rat insulin-like-growth factor II mRNA Nucleic Acid Re 1985 13 1119 1134
Ueno T Takahashi K Tetsuya T Matsuguchi T Ikejiri K Endo H Yamamato M Reactivation of rat insulin like growth factor II gene during hepatocarcinogenesis Carcinogensis 1988 9 1779 1783
Li SL Xu ZDX Kimura G Sun Y Kawachi MH Wilson TG Wilczynk S Yamaguchi YE Expression of insulin like growth factor (IGF)-II in human prostate, breast, bladder and paraganglioma tumors Cell Tissue Res 1998 291 469 479 9477303 10.1007/s004410051016
Uchida K Kondo M Takeda S Osada H Takahashi T Nakado A Takahashi T Altered transcriptional regulation of the insulin like growth factor 2 gene in human hepatocellular carcinoma Mol Carcinogensis 1997 18 193 198 10.1002/(SICI)1098-2744(199704)18:4<193::AID-MC2>3.0.CO;2-F
Schirmacher P Held WA Yang D Chisari FV Rustum Y Rogler C Reactivation of insulin like growth factor II during hepatocarcinogensis in transgenic mice suggests a role of malignant growth Cancer Res 1992 52 2549 2556 1568224
Stewart C Rotwein P Growth, differentiation and survival; multiple physiological functions for insulin like growth factors Physiol Rev 1996 76 1005 1026 8874492
Cullen KJ Smith HS Hills Lippman ME Growth factor messenger RNA expression by human breast fibroblast from benign and malignant lesions Cancer research 1991 51 4978 85 1893385
Zapf J Schoenle E Froeasch ER Effect of insulin-like growth factor I and II: Some biological actions and receptors binding characteristics of two purified constituents of nonsuppressible insulin-like activity in human serum Eur J Biochem 1978 87 285 296 668694 10.1111/j.1432-1033.1978.tb12377.x
Blatt J White C Dienes S Friedman H Foley T Production of an insulin-like growth factor by osteosarcoma Biochem Biophysics Res Commun 1984 123 373 379
Yang D Rogler CE Analysis of insulin-like growth factor II (IGF II) expression in neoplastic nodules and hepatocellular carcinoma of woodchucks utilizing in situ hybridization and immunohistochemistry Carcinogenesis (Lond) 1991 12 1893 1901 1657428
Modan-Moss D Janicot M Mclenithan JC Lane MD Casella SJ Expression and Function of insulin-like growth factor I hybrid receptors during differentiation of 3T3-L1 preadipocytes Biochem J 1998 333 825 831 9677346
Faber E The sequential analysis of liver cancer induction Biophys Biochim Acta 1980 605 149 166
Bannasch P Zerban H Hacker HJ Jones TC, Popp J, Mohr U Foci of altered hepatocytes, rat Monogrpahs on pathology of laboratory animals Digestive system 1997 Berlin, Heidelberg, New York; Tokyo, Springer 3 37
Su Q Benner A Hofmann WJ Otto G Pichlmayr R Bannasch P Human hepatic preneoplasia : phenotypes and proliferation kinetics of foci and nodules of altered hepatocytes and their relationship to liver cell dysplasia Virchows Arch 1997 431 391 406 9428927 10.1007/s004280050116
Bannasch P Boyer JL, Ockner RK Pathogensis of hepatocellular carcinoma : sequential cellular, molecular and metabolic changes Progress in liver disease 1996 14 Phildelphia, W.B. Saunders 1161 1197
Fiorentino M Grigioni WF Baccarini P Errico D De Mitri MS Pisi E Mancini AM Different in situ expression of insulin-like-growth factor type II in hepatocellular carcinoma : an in situ hybridization and immunohistochemical study Diagn Mol Pathol 1994 3 59 65 8162257
Pence CB Dietary Selenium and anti-oxidant status; toxic effects of 12 demethyl hydrazine in rats J Nut 1991 121 138 44
Mukherjee B Basu M Chaterjee M Effect of selenomethionine on N-methyl nitroso guanidine-induced colonic aberrant crypt foci in rats Europian Journal of Cancer Prevention 2001 10 347 355 10.1097/00008469-200108000-00008
Bannasch P The cytoplasm of hepatocytes during carcinogensis Recent Results in Cancer Research 1968 19 Heidelberg, Springer 1 10
Bannasch P Jahn UR Hacker HJ Hofmann W Pichlmayr R Otto G Focal hepatic glycogenosis: a putative preneoplastic lesion associated with neoplasia and cirrhosis in explanted human livers Int J of Oncology 1991 10 261 268
Yang D Alt E Rogler CE Coordinate expression of N-myc2 and insulin-like growth factor II in precancerous altered hepatic foci in woodchuck hepatis virus carriers Cancer Res 1993 53 2020 7 8481904
Moreno FS Rizzi MBSL Dagil MLA Inhibitory effect of B-Carotene on preneoplastic lesions induced in Wiser rats by the resistant hepatocyte model Carcinogenesis 1991 12 1817 1822 1934263
Braissant O Wahli A simplified in situ hydridization protocol using non-radioactively labeled probes to detect abundant and rare mRNAs on tissue sections Biochemica 1998 1 10 16
Enzmann H Zerban H Kopp-Schneider A Loser E Bannasch P Effects of low doses of N- Nitrosomorpholine on the development of early stages of hepatocarcinogenesis Carcinogenesis 1995 16 1513 1518 7614684
Bannasch P Blosch M Zerban H Spongiosis hepatis: Specific changes of the perisinusoidal liver cells induced in rats by N- Nitrosomorpholine Lab Invest 1981 44 252 264 7464049
Stroebel P Mayer F Zerban H Bannasch P Neoplasm Deriving from the perisinusoidal (ito) cells in rat liver Am J Path 1995 146 903 913 7717457
Amann G Breitschopf H Lassmann H Suchanck Gm Heniz-Erian P Cellular localization of insulin- like growth factor II mRNA in human fetus and the placenta : detection with a digoxigenin labeled cRNA probe and immunocytochemistry Pediatr Res 1998 43 614 620 9585007
Nivet V Hajduch E Hainault I Delattre J Lavau M Hanique B Sensitive northern blot by hybridization using digexigenin RNA probes for the mRNA detection of two gulcose transporter isoforms Cell Mol Biol 1997 41 979 984 8595377
Sohda T Oka Y Iwata K Gunn J Kamimura S Shijo H Okumura M Yun K Co-localization of insulin-like growth factor II and the proliferation marker-MIBI in hepatocellular carcinoma cells J Clin Pathol 1997 50 135 137 9155694
Cariani E Lasserre C Seurin D Hamelin B Kemeny F Franco D Czech MP Allurich A Brechot C Differential expression of insulin-like-growth factor II mRNA in human primary liver cancers, benign tumors and liver cirrhosis Cancer Res 1988 48 6844 6849 3180092
Norstedt GA Levinewit A Moller C Eriksson LC Andersson G Expression of insulin-like-growth factor I (IGF I) and IGF II mRNA during hepatic development, proliferation and carcinogensis in rat Carcinogensis 1988 9 209 203
Lin YZ Ding L Chen JY Expression of IGF II and its relation to the differentiation of preneoplastic hepatocytes in rats Chung-hua-Ping-Li-Hsuch-Tsa-Chih 1993 22 9 12
Chung CK Antoniades HN Expression of c-cis/platelet-derived growth factor β, insluin-like growth factor I, and transforming growth factor alpha messenger RNAs and their respective receptor messenger RNAs in primary human gastric carcinomas : in vivo studies with in situ hybridization and immunocytochemsitry Cancer Res 1992 52 3453 3459 1317752
Wang YZ Wong YC Sex hormone-induced prostatic carcinogenesis in the noble rat : The role of insulin-like-growth factor-I (IGF I) and vascular endothelial growth factor (VEGF) in the development of prostate cancer Prostate 1998 35 165 177 9582085 10.1002/(SICI)1097-0045(19980515)35:3<165::AID-PROS2>3.0.CO;2-G
Dietrich DR Candrian R Marsman DS Popp JA Kanfmann WK Swenberg JA Retrospective assessment of liver cell proliferation via PCA: a comparison with tritiated thymidine Cancer Letter 1994 79 45 51 10.1016/0304-3835(94)90061-2
Aihara T Noguchi S Miyoshi Y Sasaki Y Nakamura Y Monden M Imaoka S Allelic imbalance of insulin-like-growth factor II gene expression in cancer in cancerous and precancerous lesions of the liver Hepatology 1998 28 86 89 9657100 10.1002/hep.510280113
|
16092956
|
PMC1199609
|
CC BY
|
2021-01-04 16:39:20
|
no
|
J Carcinog. 2005 Aug 10; 4:12
|
utf-8
|
J Carcinog
| 2,005 |
10.1186/1477-3163-4-12
|
oa_comm
|
==== Front
Cell ChromosomeCell & Chromosome1475-9268BioMed Central London 1475-9268-4-11608351010.1186/1475-9268-4-1ReviewChromosomal changes in uroepithelial carcinomas Fadl-Elmula Imad [email protected] Al Neelain Medical Research Center, Faculty of Medicine, Al Neelain University, Khartoum, Sudan2005 7 8 2005 4 1 1 30 12 2003 7 8 2005 Copyright © 2005 Fadl-Elmula; licensee BioMed Central Ltd.2005Fadl-Elmula; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
This article reviews and summarizes chromosomal changes responsible for the initiation and progression of uroepithelial carcinomas. Characterization of these alterations may lead to a better understanding of the genetic mechanisms and open the door for molecular markers that can be used for better diagnosis and prognosis of the disease. Such information might even help in designing new therapeutic strategies geared towards prevention of tumor recurrences and more aggressive approach in progression-prone cases.
The revision of 205 cases of uroepithelial carcinomas reported with abnormal karyotypes showed karyotypic profile characterized by nonrandom chromosomal aberrations varying from one or few changes in low-grade and early stage tumors to massively rearranged karyotypes in muscle invasive ones. In general, the karyotypic profile was dominated by losses of chromosomal material seen as loss of entire chromosome and/or deletions of genetic materials. Rearrangements of chromosome 9 resulting in loss of material from 9p, 9q, or of the entire chromosome were the most frequent cytogenetic alterations, seen in 45% of the cases. Whereas loss of material from chromosome arms 1p, 8p, and 11p, and gains of chromosome 7, and chromosome arm 1q, and 8q seem to be an early, but secondary, changes appearing in superficial and well differentiated tumors, the formation of an isochromosome for 5p and loss of material from 17p are associated with more aggressive tumor phenotypes. Upper urinary tract TCCs have identical karyotypic profile to that of bladder TCCs, indicating the same pathogenetic mechanisms are at work in both locales. Intratumor cytogenetic heterogeneity was not seen except in a few post-radiation uroepithelial carcinomas in which distinct karyotypic and clonal pattern were characterized by massive intratumor heterogeneity (cytogenetic polyclonality) with near-diploid clones and simple balanced and/or unbalanced translocations. In the vast majority of cases strong correlation between the tumors grade/stage and karyotypic complexity was seen, indicating that progressive accumulation of acquired genetic alterations is the driving force behind multistep bladder TCC carcinogenesis. Although most of these cytogenetic alterations have been identified for many years, the molecular consequences and relevant cancer genes of these alterations have not yet been identified. However, loss of TSG(s) from chromosome 9 seems to be the primary and important event(s) in uroepithelial carcinogenesis
==== Body
Introduction
Uroepithelial cancer e.g., cancer of the urinary bladder, ureter, and renal pelvis, is the 4th and 8th most common malignant neoplasm in men and the women respectively [1]. In areas where infestation with Schistosoma haematobium is endemic, uroepithelial cancer is even more common and accounts for 25% of all cancers in men. The incidence, the etiologies, and the histology types vary considerably among countries. In Europe and the USA, chemicals, including smoking, are the main etiology of dominant transitional cell carcinoma (TCC) (90%) of the bladder. The picture is reversed in the Middle East and Africa, where urinary Schistosoma haematobium is associated with an increased risk of developing BC, especially of the SCC type. In these areas BC is the most common cancer in men (25% of all cancers) [2].
Uroepithelial cancer is characterized by unique natural history that is characterized by heterogeneity of the clinical course seen as variation in clinical-course of tumors [3]. Even tumors of the same pathologic stage, may follow very different clinical courses. The prognostic markers currently used are tumor grade, multiplicity, tumor shape, location, and presence of carcinoma in situ (Cis); however all are of limited value [4,5]. The difficulties encountered in predicting the clinical course, especially for superficial bladder tumors may suggest that TCC has at least two main subgroups with distinct genetic make-ups and, consequently, clinical behavioral patterns [6]. The natural history is also characterized by high incidence of recurrence seen in 70% of the cases after the initial treatment even in superficial bladder tumors [3,7]. Ureteral and renal pelvis tumors also have a high risk (40%–75%) of having recurrent bladder tumor and 50% of these cases develop metastasis [8]. Of these, 80% remain superficial throughout the patient's life, whereas 16% to 25% eventually recur as higher grade and/or stage tumor(s) [9]. The last clinical characteristics of uroepithelial carcinomas is the multifocal nature of the disease, seen in around 30% of urinary tract TCC at the time of diagnosis [5]. Although the mechanism behind the polychronotopicity of uroepithelial carcinomas is settling, since the possibility of new tumor(s) arising as the result of field cancerization was ruled out by cytogenetic analysis of synchronous and/or metachronous tumors [10], the heterogeneity of the clinical course justify the need for the study of prognostic genetic marker. Such information when made available might help in the design of new therapeutic strategies geared toward a more aggressive multimodal treatment approach in progression-prone cases.
Materials and methods
A total of 205 uroepithelial tumors (188 bladder tumors, 15 ureteral, one renal pelvis and one prostatic urethra tumor) were obtained from 153 male and 48 female patients. They were processed for short-term culturing and chromosomal analysis after staining with Wright stain according to the standard protocols. All cases revealed cytogenetic abnormality and all were reported previously [Appendix 1]. Of these only 3 tumors were SCC, whereas the remaining were of TCC type. The histopathology examination revealed superficial disease (Ta-1/GI-II) in 109 tumors and muscle invasion (T2-or more) in 96 tumors. For each case the histology grading, the tumor staging, and the karyotypic descriptions were reviewed in order to understand chromosomal changes and cytogenetic profile associated with biological parameters of uroepithelial tumors.
Maps of breakpoints and imbalances
To help identify chromosomal changes in a more comprehensive manner, the data were also presented as breakpoint and karyotypic imbalance maps (Figures 1 and 2). In order to report the basic and most representative range of cytogenetic changes and to avoid cytogenetic noise brought about by possible ploidy shifts, we included only 2n-4n clones for each tumor in the maps. In case of duplicated aberrations found within a clone or in more than one related clone, the breakpoints were plotted once only. If any aberrant chromosome was involved twice in the same clone or in a related clone, only additional breakpoints were considered. When the same chromosome was involved in both numerical and structural aberrations, we recorded only the total net imbalance, and if the same imbalance was found in more than one related clone, it was recorded only once. If additional gains or losses of the same chromosome or chromosomal segment were found in related clones, only the largest imbalance was recorded.
Figure 1 Distribution of 677 breakpoints observed in structural chromosomal aberrations in 205 uroepithelial carcinomas.
Figure 2 Karyotypic imbalances caused by numerical and structural aberrations in 205 uroepithelial carcinomas. Losses are shown in red, gains in blue.
Results
A total of 205 uroepithelial carcinomas have been reported with chromosomal changes.(Appendix 1). Of these, 40 (18%) had grossly incomplete karyotypes, 58 (26%) displayed altogether 157 unidentified marker chromosomes as well as 18 unidentified ring chromosomes and a total of 208 uncertain breakpoint, as is evident from how the karyotypes were written [11].
All chromosomes were involved in numerical and/or structural aberrations at least once. The most commonly involved were chromosomes 9 (in 45%) of cases, 1 (in 29%), 11 (in 27%), 8 (in 22%), 7 (in 20%), Y (in 20%), 5 (in 18%), 17 (in 18%), 3 (in 17%), 15 (in 15%), 6 (in 13%), 14 (in 13%), and 18 (in 12%).
A total of 15 unbalanced recurrent chromosomal rearrangements were described: del(1)(q21), i(1)(q10), i(5)(p10), del(6)(q21), i(11)(q10), del(1)(p11), add(3)(q21), add(5)(q11), del(6)(q13), i(13)(q10), del(14)(q24), and i(17)(q10).
A total of 677 chromosomal breakpoints were identified in the 205 uroepithelial TCC showing structural chromosomal abnormalities (Figure 1), with the highest number of breakpoints (95) seen in chromosome 1 and more than 40 seen in chromosomes 2, 3, 5, 6, 8, 9, 11, and 13. The chromosome bands most frequently involved were 1p12, 1q12, 2q21, 1q25, 2q27, 5p10, 6q21, 9p10, 9q22, 10q22, 11p15, 11p11 and 13q12 (each involved at least 10 times). The karyotypic imbalances resulting from the structural and numerical chromosomal aberrations (Figure 2) were dominated by loss of an entire copy of chromosomes 4, 9, 10, 14, 15, 16, 18, 21, 22, X, and Y, and gain of entire copy of chromosomes 7, 16, 19, and 20. Losses of the entire arms or parts of 1p, 5q, 8p, 9p, 11p, 13p, 15p and 17p and gain of 1q, 3q, 5p, 8q, 13q, and 17q were also common.
Discussion
Although uroepithelial carcinomas are among the most common malignancies, their karyotypic characteristics and genetic pathway remain poorly understood. Most of the reported data are quantitatively limited and lacking karyotypic precision [11]
Although no specific chromosomal aberration has been identified for uroepithelial carcinomas, a clearly nonrandom pattern of chromosomal changes has emerged, albeit with considerable karyotypic heterogeneity among cases ranging from the presence of sole anomalies in early tumors to very complex karyotypes in advanced ones. Translocations are rarely seen, at least in early stages, and seem to play no important role in the initiation of uroepithelial carcinomas. Instead, the cytogenetic profile is dominated by nonrandom chromosome gains and, especially, losses, the latter indicating that loss of tumor suppressor gene(s) may be the most crucial event in the pathogenesis of uroepithelial carcinomas.
The fact that chromatin losses dominated the imbalance pattern indicates that loss of tumor suppressor gene(s) is the most important pathogenetic consequence of the chromosomal aberrations associated with uroepithelial carcinoma.
Changes involving chromosome 9 [-9, del(9p), del(9q)] are the most common chromosomal aberrations in uroepithelial carcinomas. Rearrangements of chromosome 9 are seen as the sole change in cases with simple karyotypes in early and superficial carcinomas but also persisted in the massively complex karyotypes of advanced muscle invasive tumors [12]. Besides loss of the entire chromosome copy, losses of material from either arms is often seen, which may indicates the presence of at least one pathogenetically important TSG in each arm [13,14] (Figures 3 and 4). Loss of chromosome 9 material had therefore been widely accepted as an early ubiquitous, pathogenetically important and early event in urinary tract transitional cell carcinogenesis [15,16]. Recent data suggest that 9q abnormalities are more common in Ta compared with T1 tumors, in which a mixture of aberrant 9p and 9q genotypes are seen [17]. These observations indicate that loss of 9p material may be associated with the development of tumors with more aggressive biological behavior or, alternatively, they may be related to early disease progression [6]. Although several attractive candidates such as p16/CDKN2 in 9p21 and TSC1 in 9q34 have been reported to be homozygously deleted in superficial TCC of the bladder [18] the crucial gene-level consequences of these chromosomal aberrations remain unknown.
Figure 3 Representative karyotype from a well differentiated transitional cell carcinoma of the bladder. The arrowhead indicates the breakpoint in a deleted chromosome 9 {46,XY,del(9)(p11)}.
Figure 4 Representative karyotype from a well differentiated transitional cell carcinoma of the bladder. The arrowhead indicates the breakpoint in a deleted chromosome 9 {46,XY,del(9)(q12q22)}.
Trisomy 7 has been frequently described in near-diploid karyotypes and as the sole chromosomal change in bladder and ureteral tumors [19] (Figure 5). The importance of trisomy 7 as a tumor-associated aberration and its molecular consequence remain controversial since it has been found repeatedly also in some unquestionably nonneoplastic lesions [20]. However, the important molecular consequences of trisomy 7 may be an increased number of alleles for the epidermal growth factor (EGF) receptor gene.
Figure 5 Representative karyotype from a well differentiated transitional cell carcinoma of the bladder. The chromosome indicated by "mar" represents unidentified marker, "r" represents ring chromosome. Arrowheads indicate breakpoints.
In addition to chromosomes 9 and 7, chromosomes 1, 8, and 11 seemed to be preferentially involved and were found to be rearranged in superficial tumors. Alterations involving chromosome 1 had been seen in 35% of the investigated BC cases [11]. The alterations are diverse and include deletions, translocations, duplications, and isochromosome formations. However, regardless of the type of changes, the most common consequences of the changes in chromosome 1 were gain of 1q material and loss of 1p material. Changes involving chromosome 1 are rarely if ever seen as the sole aberration but nevertheless may form part of relatively simple karyotypes. Support for the view that changes of chromosome 1 are secondary in BC tumorigenesis also comes from molecular cytogenetic investigations showing that gain of 1q material is more frequent in pT1 than in pTa tumors [21]. Regardless of when they occur, the molecular consequences and biological significance of these changes in uroepithelial carcinomas remain uncertain, although some investigators suggest that they may be implicated in tumor progression and in vitro immortalization of human cells [22].
Aberrations involving chromosome 8 usually resulting in loss of material from the short arm, have been repeatedly detected in urinary tract tumors. Losses from 8p and gain of 8q were the dominant net result in, almost all, changes affecting this chromosome. Although often found in simple karyotypes of low-grade tumors, the aberrations involving chromosomes 1, 8 and/or 11 were never the sole change, but typically accompanied by chromosome 9 rearrangements [11]. Similar aberrations occur frequently also in many other tumor types as carcinoma of the lung[23]. Studies of patients with non-muscle-invasive bladder cancers have shown more frequent loss of 8p in minimally invasive (pT1) tumors than in non-invasive (pTa) ones [24]. These findings are consistent with the view that loss of a putative TSG on 8p plays important role in the progression and tumor invasion in BC.
Several cytogenetic studies have revealed nonrandom involvement of chromosome 11 in BC, mostly leading to loss of genetic material from 11p [11]. Additional evidence to the same effect has come from CGH analyses[25]. The frequency of 11p loss may be higher in pT1 than in pTa tumors, and even higher in pT2-4b tumors [26]. In contrast, some FISH studies using centromeric probes have shown an increased copy number of chromosome 11[27]. However, these studies did not assess the tumors' ploidy level and the seemingly increased copy number may therefore have reflected general polyploidization. Loss of 11p therefore appears to be an early but secondary change, associated with tumor progression. The putative TSG on 11p lost through the chromosomal rearrangements remain unknown.
Chromosomal changes associated with aggressive biological tumor behavior
The formation of an isochromosome for 5p, i(5)(p10), leading to net gain of material from the short arm of chromosome 5 has been reported in bladder TCCs, making it the single most common recurrent structural chromosomal abnormality in Uroepithelial tumors [11]. In most cytogenetic reports the aberration was associated with aggressive tumor phenotypes, muscle-invasive and poorly differentiated tumors. Considering the fact that this marker has not been consistently linked to any other malignancy and that it is seldom seen in association with rearrangements of chromosome 17, it appears to characterize a subgroup of advanced TCC of the bladder arising via a unique pathogenetic pathway. A strong correlation between chromosome 5 involvement and tumor grade and stage was shown in several studies [6,28].
Rearrangement of chromosome 17 resulting in loss of material from the short arm is rarely seen in superficial BC, but is common in more advanced and aggressive tumors. The molecular target of 17p material loss could be the TP53 gene, which resides at 17p12, and is known as the most generally important tumor suppressor gene in human neoplasia. Cytogenetic analyses revealed involvement of 17p in less than 10% of the analyzed tumors, and molecular genetic studies have shown LOH in up to 42% of the tumors. While cytogenetic data probably underestimate 17p involvement. However, this finding indicates that most deletions of 17p are submicroscopic and thus cannot be detected using conventional cytogenetic technique. On the other hand, several FISH analyses using centromeric probes have revealed frequent chromosome 17 copy-gain in bladder carcinomas of high grade and stage. Such apparent gains of whole chromosome copies in FISH analyses may reflect polyploidization in advanced tumors rather than a specific role of chromosome 17. Although rearrangement of chromosome 17 resulting in loss of material from the short arm was seen frequently in high-grade, muscle-invasive tumors, loss of 17p material is not restricted only to advanced TCC [12]. The loss of 17p material in some superficial tumors fits the prognostic heterogeneity seen even in superficial BC in which some tumors take a more aggressive clinical course. For these ones it has been suggested that early loss of 17p, combined with mutational inactivation of the TP53 allele on the other chromosome 17, could be responsible for the aggressive behavior seen in 16–25% of superficial BCs.
Rearrangement of chromosome 3 resulting in loss of material from the short arm had been seen in 44% of the reported BC with abnormal karyotypes, usually in advanced tumors with complex karyotypes [29]. Deletion of 3p is also frequent in other carcinomas such as renal cell cancer, breast cancer, and small cell lung cancer [11]. Although many potential target genes are located on 3p, the crucial molecular-level consequence of these chromosomal aberrations in BC carcinogenesis remains unknown.
Aberrations involving chromosome 13 are known to occur in muscle invasive tumors and often result in net loss of material from the long arm [11]. LOH studies seem to be more informative than cytogenetic studies in this prospect as LOH at the RB locus in band 13q14 was seen in 90% of muscle invasive tumors.
Loss of the Y chromosome is common in bladder tumors of male patients. It has been reported in all stages and even as the sole change in several reports [30]. In contrast, bladder tumors obtained from female patients do not reveal the same incidence of X chromosome loss. FISH studies have shown that loss of the Y is infrequent in normal urothelial cells obtained from healthy males [31]. This observation does not exclude, however, the possibility that loss of chromosome Y in cultured cells could reflect changes in stromal elements, in particular since-Y has been demonstrated in non-neoplastic disease lesions of several tissues and organs such as kidney, bone marrow, and brain [32,33]. In conclusion, one cannot be certain that loss of the Y chromosome really signifies a pathogenetically important event in neoplastic cells.
Correlation between karyotype and tumor grade and stage
Strong correlation was seen between the grade/stage of the tumors and the karyotypic profile. Most superficial and well-differentiated tumors (TaG1) were pseudo-or near-diploid and exhibited simple karyotypes (5 or fewer chromosomal changes) (Figures 3, 4, 5). A progressive increase in the number of chromosome aberrations with tumor grade and/or stage was evident in most large reported series, with TaG1 tumors showing less abnormal karyotypes than did those that were T1G2, which, in their turn, were less abnormal than T2G3 tumors. This is in agreement with the view that the uroepithelial carcinomas follow the multistep carcinogenesis and that their clinical progression is steered by the synergistic effect of accumulated genetic alterations.
Upper urinary tract carcinomas (UUTC)
Very limited information is available on cytogenetics of UUTC; only 16 cases with abnormal karyotype, have been published [34,35]. Because of the great similarities between upper urinary tract carcinomas (UUTC) and bladder TCC with regard to etiology, histology, and natural disease history, one would assume that the tumorigenic mechanisms, including the karyotypic profile, are more or less identical.
Loss of material from chromosome 9 was seen in most informative UUTC [12]. The ubiquity of chromosome 9 involvement thus seems to be no less pronounced in UUTC than BC; this is in agreement with the view that this is an early and crucial event in the genesis of nearly all urinary tract TCC, regardless of their site and stage. The high frequency of chromosome 9 involvement in upper urinary tract transitional cell carcinogenesis makes this a logical target for a possible genetic marker to be used in the follow-up of patients with such tumors, something that would be particularly useful considering how frequent downstream disease spreading is in these patients.
Post-radiation uroepithelial carcinomas
Post-radiation tumors or second-primary radiation-induced cancers have distinct clinical and cytogenetic characteristics. These tumors develop in an irradiated field and should have a different histological type from that of the primary tumors. Usually they are infiltrative and of high grade at the time of diagnosis. Only 2 cases of post-radiation uroepithelial carcinoma with cytogenetics abnormality have been reported [35,36]. The karyotypic profile in both cases revealed near-diploid, karyotypically abnormal clones characterized by rather simple balanced and/or unbalanced translocations in each tumor. The intratumor karyotypic heterogeneity seen in these cases indicates massive polyclonality in contrast to the cytogenetic monoclonality consistently demonstrated in urinary tract TCC. This reflects the likelihood of the previous radiation carcinogenic effect and illustrating that post-radiation urinary tract tumors are distinct from other urinary tract TCC not only etiologically but also with regard to the pathogenetic mechanisms involved.
Supplementary Material
Additional File 1
Clinical and cytogenetic data on 205 cases of uroepithelial carcinomas
Click here for file
Acknowledgements
This work was supported by grants from the Medical Faculty of Al Neelain University and Sudanese cancer research group.
==== Refs
Silverman DT Hartge P Morrison AS Devesa SS Epidemiology of bladder cancer Hematol Oncol Clin North Am 1992 6 1 30 1556044
Badawi AF Mostafa MH Probert A O'Connor PJ Role of schistosomiasis in human bladder cancer: evidence of association, aetiological factors, and basic mechanisms of carcinogenesis Eur J Cancer Prev 1995 4 45 59 7728097
Herr HW Natural history of superficial bladder tumors: 10- to 20-year follow-up of treated patients World J Urol 1997 15 84 8 9144896
Aprikian AG Sarkis AS Reuter VE Cordon-Cardo C Sheinfeld J Biological markers of prognosis in transitional cell carcinoma of the bladder: current concepts Semin Urol 1993 11 137 44 8210835
Holmang S Hedelin H Anderstrom C Johansson SL The relationship among multiple recurrences, progression and prognosis of patients with stages Ta and T1 transitional cell cancer of the bladder followed for at least 20 years J Urol 1995 153 1823 6 7752327 10.1097/00005392-199506000-00022
Hoglund M Sall T Heim S Mitelman F Mandahl N Fadl-Elmula I Identification of cytogenetic subgroups and karyotypic pathways in transitional cell carcinoma Cancer Res 2001 61 8241 6 11719456
Lee R Droller MJ The natural history of bladder cancer. Implications for therapy Urol Clin North Am 2000 27 1 13 10696240 10.1016/S0094-0143(05)70229-9
Steffens J Nagel R Tumours of the renal pelvis and ureter. Observations in 170 patients Br J Urol 1988 61 277 83 3289673
Kiemeney LA Witjes JA Heijbroek RP Verbeek AL Debruyne FM Predictability of recurrent and progressive disease in individual patients with primary superficial bladder cancer J Urol 1993 150 60 4 8510276
Fadl-Elmula I Gorunova L Mandahl N Elfving P Lundgren R Mitelman F Heim S Cytogenetic monoclonality in multifocal uroepithelial carcinomas: evidence of intraluminal tumour seeding Br J Cancer 1999 81 6 12 10487605 10.1038/sj.bjc.6690643
Mitelman F www.Mitelmansdatabase.org 2008
Fadl-Elmula I Gorunova L Mandahl N Elfving P Lundgren R Mitelman F Heim S Karyotypic characterization of urinary bladder transitional cell carcinomas Genes Chromosomes Cancer 2000 29 256 65 10992300 10.1002/1098-2264(2000)9999:9999<::AID-GCC1034>3.0.CO;2-O
Ohgaki K Minobe K Kurose K Iida A Habuchi T Ogawa O Kubota Y Akimoto M Emi M Two target regions of allelic loss on chromosome 9 in urinary-bladder cancer Jpn J Cancer Res 1999 90 957 64 10551324
Simoneau AR Spruck CH Gonzalez-Zulueta M Gonzalgo ML Chan MF Tsai YC Dean M Steven K Horn T Jones PA Evidence for two tumor suppressor loci associated with proximal chromosome 9p to q and distal chromosome 9q in bladder cancer and the initial screening for GAS1 and PTC mutations Cancer Res 1996 56 5039 43 8895761
Cairns P Shaw ME Knowles MA Initiation of bladder cancer may involve deletion of a tumour-suppressor gene on chromosome 9 Oncogene 1993 8 1083 5 8096074
Simoneau M LaRue H Aboulkassim TO Meyer F Moore L Fradet Y Chromosome 9 deletions and recurrence of superficial bladder cancer: identification of four regions of prognostic interest Oncogene 2000 19 6317 6323 11175346 10.1038/sj.onc.1204022
Gonzalez-Zulueta M Shibata A Ohneseit PF Spruck CH Busch C Shamaa M El-Baz M Nichols PW Gonzalgo ML Elbaz M al. High frequency of chromosome 9p allelic loss and CDKN2 tumor suppressor gene alterations in squamous cell carcinoma of the bladder [published erratum appears in J Natl Cancer Inst 1995 Dec 6;87(23):1807] J Natl Cancer Inst 1995 87 1383 93 7658499
Nishiyama H Takahashi T Kakehi Y Habuchi T Knowles MA Homozygous deletion at the 9q32-33 candidate tumor suppressor locus in primary human bladder cancer Genes Chromosomes Cancer 1999 26 171 5 10469456 10.1002/(SICI)1098-2264(199910)26:2<171::AID-GCC10>3.0.CO;2-B
Berrozpe G Miro R Caballin MR Salvador J Egozcue J Trisomy 7 may be a primary change in noninvasive transitional cell carcinoma of the bladder Cancer Genet Cytogenet 1990 50 9 14 2253193 10.1016/0165-4608(90)90232-Y
Johansson B Heim S Mandahl N Mertens F Mitelman F Trisomy 7 in nonneoplastic cells Genes Chromosomes Cancer 1993 6 199 205 7685621
Richter J Wagner U Schraml P Maurer R Alund G Knonagel H Moch H Mihatsch MJ Gasser TC Sauter G Chromosomal imbalances are associated with a high risk of progression in early invasive (pT1) urinary bladder cancer Cancer Res 1999 59 5687 91 10582685
Terracciano L Richter J Tornillo L Beffa L Diener PA Maurer R Gasser TC Moch H Mihatsch MJ Sauter G Chromosomal imbalances in small cell carcinomas of the urinary bladder J Pathol 1999 189 230 5 10547580 10.1002/(SICI)1096-9896(199910)189:2<230::AID-PATH407>3.0.CO;2-8
Wistuba II Behrens C Virmani AK Milchgrub S Syed S Lam S Mackay B Minna JD Gazdar AF Allelic losses at chromosome 8p21-23 are early and frequent events in the pathogenesis of lung cancer Cancer Res 1999 59 1973 9 10213509
Zhao J Richter J Wagner U Roth B Schraml P Zellweger T Ackermann D Schmid U Moch H Mihatsch MJ Gasser TC Sauter G Chromosomal imbalances in noninvasive papillary bladder neoplasms (pTa) Cancer Res 1999 59 4658 61 10493521
Voorter C Joos S Bringuier PP Vallinga M Poddighe P Schalken J du Manoir S Ramaekers F Lichter P Hopman A Detection of chromosomal imbalances in transitional cell carcinoma of the bladder by comparative genomic hybridization Am J Pathol 1995 146 1341 54 7778674
Sauter G Gasser TC Moch H Richter J Jiang F Albrecht R Novotny H Wagner U Bubendorf L Mihatsch MJ DNA aberrations in urinary bladder cancer detected by flow cytometry and FISH Urol Res 1997 25 37 43 10.1007/BF00942046
Hopman AH Moesker O Smeets AW Pauwels RP Vooijs GP Ramaekers FC Numerical chromosome 1, 7, 9, and 11 aberrations in bladder cancer detected by in situ hybridization Cancer Res 1991 51 644 51 1985781
von Knobloch R Bugert P Jauch A Kalble T Kovacs G Allelic changes at multiple regions of chromosome 5 are associated with progression of urinary bladder cancer J Pathol 2000 190 163 8 10657014 10.1002/(SICI)1096-9896(200002)190:2<163::AID-PATH509>3.0.CO;2-0
Knowles MA Elder PA Williamson M Cairns JP Shaw ME Law MG Allelotype of human bladder cancer Cancer Res 1994 54 531 8 8275491
Smeets W Pauwels R Laarakkers L Debruyne F Geraedts J Chromosomal analysis of bladder cancer. III. Nonrandom alterations Cancer Genet Cytogenet 1987 29 29 41 3311350 10.1016/0165-4608(87)90028-8
Bentz M Plesch A Stilgenbauer S Dohner H Lichter P Minimal sizes of deletions detected by comparative genomic hybridization Genes Chromosomes Cancer 1998 21 172 5 9491330 10.1002/(SICI)1098-2264(199802)21:2<172::AID-GCC14>3.0.CO;2-T
Elfving P Cigudosa JC Lundgren R Limon J Mandahl N Kristoffersson U Heim S Mitelman F Trisomy 7, trisomy 10, and loss of the Y chromosome in short-term cultures of normal kidney tissue Cytogenet Cell Genet 1990 53 123 5 2369838
Heim S Mandahl N Jin Y Stromblad S Lindstrom E Salford LG Mitelman F Trisomy 7 and sex chromosome loss in human brain tissue Cytogenet Cell Genet 1989 52 136 8 2630185
Batata M Grabstald H Upper urinary tract urothelial tumors Urol Clin North Am 1976 3 79 86 936370
Fadl-Elmula I Gorunova L Mandahl N Elfving P Lundgren R Rademark C Heim S Cytogenetic analysis of upper urinary tract transitional cell carcinomas Cancer Genet Cytogenet 1999 115 123 7 10598145 10.1016/S0165-4608(99)00075-8
Fadl-Elmula I Bonaldi L Gorunova L Mandahl N Elfving P Heim S Cytogenetic heterogeneity in a second primary radiation-induced bladder carcinoma: ten karyotypically unrelated clones Cancer Genet Cytogenet 1998 105 134 7 9723030 10.1016/S0165-4608(98)00028-4
|
16083510
|
PMC1199610
|
CC BY
|
2021-01-04 16:38:33
|
no
|
Cell Chromosome. 2005 Aug 7; 4:1
|
utf-8
|
Cell Chromosome
| 2,005 |
10.1186/1475-9268-4-1
|
oa_comm
|
==== Front
Cell ChromosomeCell & Chromosome1475-9268BioMed Central London 1475-9268-4-11608351010.1186/1475-9268-4-1ReviewChromosomal changes in uroepithelial carcinomas Fadl-Elmula Imad [email protected] Al Neelain Medical Research Center, Faculty of Medicine, Al Neelain University, Khartoum, Sudan2005 7 8 2005 4 1 1 30 12 2003 7 8 2005 Copyright © 2005 Fadl-Elmula; licensee BioMed Central Ltd.2005Fadl-Elmula; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
This article reviews and summarizes chromosomal changes responsible for the initiation and progression of uroepithelial carcinomas. Characterization of these alterations may lead to a better understanding of the genetic mechanisms and open the door for molecular markers that can be used for better diagnosis and prognosis of the disease. Such information might even help in designing new therapeutic strategies geared towards prevention of tumor recurrences and more aggressive approach in progression-prone cases.
The revision of 205 cases of uroepithelial carcinomas reported with abnormal karyotypes showed karyotypic profile characterized by nonrandom chromosomal aberrations varying from one or few changes in low-grade and early stage tumors to massively rearranged karyotypes in muscle invasive ones. In general, the karyotypic profile was dominated by losses of chromosomal material seen as loss of entire chromosome and/or deletions of genetic materials. Rearrangements of chromosome 9 resulting in loss of material from 9p, 9q, or of the entire chromosome were the most frequent cytogenetic alterations, seen in 45% of the cases. Whereas loss of material from chromosome arms 1p, 8p, and 11p, and gains of chromosome 7, and chromosome arm 1q, and 8q seem to be an early, but secondary, changes appearing in superficial and well differentiated tumors, the formation of an isochromosome for 5p and loss of material from 17p are associated with more aggressive tumor phenotypes. Upper urinary tract TCCs have identical karyotypic profile to that of bladder TCCs, indicating the same pathogenetic mechanisms are at work in both locales. Intratumor cytogenetic heterogeneity was not seen except in a few post-radiation uroepithelial carcinomas in which distinct karyotypic and clonal pattern were characterized by massive intratumor heterogeneity (cytogenetic polyclonality) with near-diploid clones and simple balanced and/or unbalanced translocations. In the vast majority of cases strong correlation between the tumors grade/stage and karyotypic complexity was seen, indicating that progressive accumulation of acquired genetic alterations is the driving force behind multistep bladder TCC carcinogenesis. Although most of these cytogenetic alterations have been identified for many years, the molecular consequences and relevant cancer genes of these alterations have not yet been identified. However, loss of TSG(s) from chromosome 9 seems to be the primary and important event(s) in uroepithelial carcinogenesis
==== Body
Introduction
Uroepithelial cancer e.g., cancer of the urinary bladder, ureter, and renal pelvis, is the 4th and 8th most common malignant neoplasm in men and the women respectively [1]. In areas where infestation with Schistosoma haematobium is endemic, uroepithelial cancer is even more common and accounts for 25% of all cancers in men. The incidence, the etiologies, and the histology types vary considerably among countries. In Europe and the USA, chemicals, including smoking, are the main etiology of dominant transitional cell carcinoma (TCC) (90%) of the bladder. The picture is reversed in the Middle East and Africa, where urinary Schistosoma haematobium is associated with an increased risk of developing BC, especially of the SCC type. In these areas BC is the most common cancer in men (25% of all cancers) [2].
Uroepithelial cancer is characterized by unique natural history that is characterized by heterogeneity of the clinical course seen as variation in clinical-course of tumors [3]. Even tumors of the same pathologic stage, may follow very different clinical courses. The prognostic markers currently used are tumor grade, multiplicity, tumor shape, location, and presence of carcinoma in situ (Cis); however all are of limited value [4,5]. The difficulties encountered in predicting the clinical course, especially for superficial bladder tumors may suggest that TCC has at least two main subgroups with distinct genetic make-ups and, consequently, clinical behavioral patterns [6]. The natural history is also characterized by high incidence of recurrence seen in 70% of the cases after the initial treatment even in superficial bladder tumors [3,7]. Ureteral and renal pelvis tumors also have a high risk (40%–75%) of having recurrent bladder tumor and 50% of these cases develop metastasis [8]. Of these, 80% remain superficial throughout the patient's life, whereas 16% to 25% eventually recur as higher grade and/or stage tumor(s) [9]. The last clinical characteristics of uroepithelial carcinomas is the multifocal nature of the disease, seen in around 30% of urinary tract TCC at the time of diagnosis [5]. Although the mechanism behind the polychronotopicity of uroepithelial carcinomas is settling, since the possibility of new tumor(s) arising as the result of field cancerization was ruled out by cytogenetic analysis of synchronous and/or metachronous tumors [10], the heterogeneity of the clinical course justify the need for the study of prognostic genetic marker. Such information when made available might help in the design of new therapeutic strategies geared toward a more aggressive multimodal treatment approach in progression-prone cases.
Materials and methods
A total of 205 uroepithelial tumors (188 bladder tumors, 15 ureteral, one renal pelvis and one prostatic urethra tumor) were obtained from 153 male and 48 female patients. They were processed for short-term culturing and chromosomal analysis after staining with Wright stain according to the standard protocols. All cases revealed cytogenetic abnormality and all were reported previously [Appendix 1]. Of these only 3 tumors were SCC, whereas the remaining were of TCC type. The histopathology examination revealed superficial disease (Ta-1/GI-II) in 109 tumors and muscle invasion (T2-or more) in 96 tumors. For each case the histology grading, the tumor staging, and the karyotypic descriptions were reviewed in order to understand chromosomal changes and cytogenetic profile associated with biological parameters of uroepithelial tumors.
Maps of breakpoints and imbalances
To help identify chromosomal changes in a more comprehensive manner, the data were also presented as breakpoint and karyotypic imbalance maps (Figures 1 and 2). In order to report the basic and most representative range of cytogenetic changes and to avoid cytogenetic noise brought about by possible ploidy shifts, we included only 2n-4n clones for each tumor in the maps. In case of duplicated aberrations found within a clone or in more than one related clone, the breakpoints were plotted once only. If any aberrant chromosome was involved twice in the same clone or in a related clone, only additional breakpoints were considered. When the same chromosome was involved in both numerical and structural aberrations, we recorded only the total net imbalance, and if the same imbalance was found in more than one related clone, it was recorded only once. If additional gains or losses of the same chromosome or chromosomal segment were found in related clones, only the largest imbalance was recorded.
Figure 1 Distribution of 677 breakpoints observed in structural chromosomal aberrations in 205 uroepithelial carcinomas.
Figure 2 Karyotypic imbalances caused by numerical and structural aberrations in 205 uroepithelial carcinomas. Losses are shown in red, gains in blue.
Results
A total of 205 uroepithelial carcinomas have been reported with chromosomal changes.(Appendix 1). Of these, 40 (18%) had grossly incomplete karyotypes, 58 (26%) displayed altogether 157 unidentified marker chromosomes as well as 18 unidentified ring chromosomes and a total of 208 uncertain breakpoint, as is evident from how the karyotypes were written [11].
All chromosomes were involved in numerical and/or structural aberrations at least once. The most commonly involved were chromosomes 9 (in 45%) of cases, 1 (in 29%), 11 (in 27%), 8 (in 22%), 7 (in 20%), Y (in 20%), 5 (in 18%), 17 (in 18%), 3 (in 17%), 15 (in 15%), 6 (in 13%), 14 (in 13%), and 18 (in 12%).
A total of 15 unbalanced recurrent chromosomal rearrangements were described: del(1)(q21), i(1)(q10), i(5)(p10), del(6)(q21), i(11)(q10), del(1)(p11), add(3)(q21), add(5)(q11), del(6)(q13), i(13)(q10), del(14)(q24), and i(17)(q10).
A total of 677 chromosomal breakpoints were identified in the 205 uroepithelial TCC showing structural chromosomal abnormalities (Figure 1), with the highest number of breakpoints (95) seen in chromosome 1 and more than 40 seen in chromosomes 2, 3, 5, 6, 8, 9, 11, and 13. The chromosome bands most frequently involved were 1p12, 1q12, 2q21, 1q25, 2q27, 5p10, 6q21, 9p10, 9q22, 10q22, 11p15, 11p11 and 13q12 (each involved at least 10 times). The karyotypic imbalances resulting from the structural and numerical chromosomal aberrations (Figure 2) were dominated by loss of an entire copy of chromosomes 4, 9, 10, 14, 15, 16, 18, 21, 22, X, and Y, and gain of entire copy of chromosomes 7, 16, 19, and 20. Losses of the entire arms or parts of 1p, 5q, 8p, 9p, 11p, 13p, 15p and 17p and gain of 1q, 3q, 5p, 8q, 13q, and 17q were also common.
Discussion
Although uroepithelial carcinomas are among the most common malignancies, their karyotypic characteristics and genetic pathway remain poorly understood. Most of the reported data are quantitatively limited and lacking karyotypic precision [11]
Although no specific chromosomal aberration has been identified for uroepithelial carcinomas, a clearly nonrandom pattern of chromosomal changes has emerged, albeit with considerable karyotypic heterogeneity among cases ranging from the presence of sole anomalies in early tumors to very complex karyotypes in advanced ones. Translocations are rarely seen, at least in early stages, and seem to play no important role in the initiation of uroepithelial carcinomas. Instead, the cytogenetic profile is dominated by nonrandom chromosome gains and, especially, losses, the latter indicating that loss of tumor suppressor gene(s) may be the most crucial event in the pathogenesis of uroepithelial carcinomas.
The fact that chromatin losses dominated the imbalance pattern indicates that loss of tumor suppressor gene(s) is the most important pathogenetic consequence of the chromosomal aberrations associated with uroepithelial carcinoma.
Changes involving chromosome 9 [-9, del(9p), del(9q)] are the most common chromosomal aberrations in uroepithelial carcinomas. Rearrangements of chromosome 9 are seen as the sole change in cases with simple karyotypes in early and superficial carcinomas but also persisted in the massively complex karyotypes of advanced muscle invasive tumors [12]. Besides loss of the entire chromosome copy, losses of material from either arms is often seen, which may indicates the presence of at least one pathogenetically important TSG in each arm [13,14] (Figures 3 and 4). Loss of chromosome 9 material had therefore been widely accepted as an early ubiquitous, pathogenetically important and early event in urinary tract transitional cell carcinogenesis [15,16]. Recent data suggest that 9q abnormalities are more common in Ta compared with T1 tumors, in which a mixture of aberrant 9p and 9q genotypes are seen [17]. These observations indicate that loss of 9p material may be associated with the development of tumors with more aggressive biological behavior or, alternatively, they may be related to early disease progression [6]. Although several attractive candidates such as p16/CDKN2 in 9p21 and TSC1 in 9q34 have been reported to be homozygously deleted in superficial TCC of the bladder [18] the crucial gene-level consequences of these chromosomal aberrations remain unknown.
Figure 3 Representative karyotype from a well differentiated transitional cell carcinoma of the bladder. The arrowhead indicates the breakpoint in a deleted chromosome 9 {46,XY,del(9)(p11)}.
Figure 4 Representative karyotype from a well differentiated transitional cell carcinoma of the bladder. The arrowhead indicates the breakpoint in a deleted chromosome 9 {46,XY,del(9)(q12q22)}.
Trisomy 7 has been frequently described in near-diploid karyotypes and as the sole chromosomal change in bladder and ureteral tumors [19] (Figure 5). The importance of trisomy 7 as a tumor-associated aberration and its molecular consequence remain controversial since it has been found repeatedly also in some unquestionably nonneoplastic lesions [20]. However, the important molecular consequences of trisomy 7 may be an increased number of alleles for the epidermal growth factor (EGF) receptor gene.
Figure 5 Representative karyotype from a well differentiated transitional cell carcinoma of the bladder. The chromosome indicated by "mar" represents unidentified marker, "r" represents ring chromosome. Arrowheads indicate breakpoints.
In addition to chromosomes 9 and 7, chromosomes 1, 8, and 11 seemed to be preferentially involved and were found to be rearranged in superficial tumors. Alterations involving chromosome 1 had been seen in 35% of the investigated BC cases [11]. The alterations are diverse and include deletions, translocations, duplications, and isochromosome formations. However, regardless of the type of changes, the most common consequences of the changes in chromosome 1 were gain of 1q material and loss of 1p material. Changes involving chromosome 1 are rarely if ever seen as the sole aberration but nevertheless may form part of relatively simple karyotypes. Support for the view that changes of chromosome 1 are secondary in BC tumorigenesis also comes from molecular cytogenetic investigations showing that gain of 1q material is more frequent in pT1 than in pTa tumors [21]. Regardless of when they occur, the molecular consequences and biological significance of these changes in uroepithelial carcinomas remain uncertain, although some investigators suggest that they may be implicated in tumor progression and in vitro immortalization of human cells [22].
Aberrations involving chromosome 8 usually resulting in loss of material from the short arm, have been repeatedly detected in urinary tract tumors. Losses from 8p and gain of 8q were the dominant net result in, almost all, changes affecting this chromosome. Although often found in simple karyotypes of low-grade tumors, the aberrations involving chromosomes 1, 8 and/or 11 were never the sole change, but typically accompanied by chromosome 9 rearrangements [11]. Similar aberrations occur frequently also in many other tumor types as carcinoma of the lung[23]. Studies of patients with non-muscle-invasive bladder cancers have shown more frequent loss of 8p in minimally invasive (pT1) tumors than in non-invasive (pTa) ones [24]. These findings are consistent with the view that loss of a putative TSG on 8p plays important role in the progression and tumor invasion in BC.
Several cytogenetic studies have revealed nonrandom involvement of chromosome 11 in BC, mostly leading to loss of genetic material from 11p [11]. Additional evidence to the same effect has come from CGH analyses[25]. The frequency of 11p loss may be higher in pT1 than in pTa tumors, and even higher in pT2-4b tumors [26]. In contrast, some FISH studies using centromeric probes have shown an increased copy number of chromosome 11[27]. However, these studies did not assess the tumors' ploidy level and the seemingly increased copy number may therefore have reflected general polyploidization. Loss of 11p therefore appears to be an early but secondary change, associated with tumor progression. The putative TSG on 11p lost through the chromosomal rearrangements remain unknown.
Chromosomal changes associated with aggressive biological tumor behavior
The formation of an isochromosome for 5p, i(5)(p10), leading to net gain of material from the short arm of chromosome 5 has been reported in bladder TCCs, making it the single most common recurrent structural chromosomal abnormality in Uroepithelial tumors [11]. In most cytogenetic reports the aberration was associated with aggressive tumor phenotypes, muscle-invasive and poorly differentiated tumors. Considering the fact that this marker has not been consistently linked to any other malignancy and that it is seldom seen in association with rearrangements of chromosome 17, it appears to characterize a subgroup of advanced TCC of the bladder arising via a unique pathogenetic pathway. A strong correlation between chromosome 5 involvement and tumor grade and stage was shown in several studies [6,28].
Rearrangement of chromosome 17 resulting in loss of material from the short arm is rarely seen in superficial BC, but is common in more advanced and aggressive tumors. The molecular target of 17p material loss could be the TP53 gene, which resides at 17p12, and is known as the most generally important tumor suppressor gene in human neoplasia. Cytogenetic analyses revealed involvement of 17p in less than 10% of the analyzed tumors, and molecular genetic studies have shown LOH in up to 42% of the tumors. While cytogenetic data probably underestimate 17p involvement. However, this finding indicates that most deletions of 17p are submicroscopic and thus cannot be detected using conventional cytogenetic technique. On the other hand, several FISH analyses using centromeric probes have revealed frequent chromosome 17 copy-gain in bladder carcinomas of high grade and stage. Such apparent gains of whole chromosome copies in FISH analyses may reflect polyploidization in advanced tumors rather than a specific role of chromosome 17. Although rearrangement of chromosome 17 resulting in loss of material from the short arm was seen frequently in high-grade, muscle-invasive tumors, loss of 17p material is not restricted only to advanced TCC [12]. The loss of 17p material in some superficial tumors fits the prognostic heterogeneity seen even in superficial BC in which some tumors take a more aggressive clinical course. For these ones it has been suggested that early loss of 17p, combined with mutational inactivation of the TP53 allele on the other chromosome 17, could be responsible for the aggressive behavior seen in 16–25% of superficial BCs.
Rearrangement of chromosome 3 resulting in loss of material from the short arm had been seen in 44% of the reported BC with abnormal karyotypes, usually in advanced tumors with complex karyotypes [29]. Deletion of 3p is also frequent in other carcinomas such as renal cell cancer, breast cancer, and small cell lung cancer [11]. Although many potential target genes are located on 3p, the crucial molecular-level consequence of these chromosomal aberrations in BC carcinogenesis remains unknown.
Aberrations involving chromosome 13 are known to occur in muscle invasive tumors and often result in net loss of material from the long arm [11]. LOH studies seem to be more informative than cytogenetic studies in this prospect as LOH at the RB locus in band 13q14 was seen in 90% of muscle invasive tumors.
Loss of the Y chromosome is common in bladder tumors of male patients. It has been reported in all stages and even as the sole change in several reports [30]. In contrast, bladder tumors obtained from female patients do not reveal the same incidence of X chromosome loss. FISH studies have shown that loss of the Y is infrequent in normal urothelial cells obtained from healthy males [31]. This observation does not exclude, however, the possibility that loss of chromosome Y in cultured cells could reflect changes in stromal elements, in particular since-Y has been demonstrated in non-neoplastic disease lesions of several tissues and organs such as kidney, bone marrow, and brain [32,33]. In conclusion, one cannot be certain that loss of the Y chromosome really signifies a pathogenetically important event in neoplastic cells.
Correlation between karyotype and tumor grade and stage
Strong correlation was seen between the grade/stage of the tumors and the karyotypic profile. Most superficial and well-differentiated tumors (TaG1) were pseudo-or near-diploid and exhibited simple karyotypes (5 or fewer chromosomal changes) (Figures 3, 4, 5). A progressive increase in the number of chromosome aberrations with tumor grade and/or stage was evident in most large reported series, with TaG1 tumors showing less abnormal karyotypes than did those that were T1G2, which, in their turn, were less abnormal than T2G3 tumors. This is in agreement with the view that the uroepithelial carcinomas follow the multistep carcinogenesis and that their clinical progression is steered by the synergistic effect of accumulated genetic alterations.
Upper urinary tract carcinomas (UUTC)
Very limited information is available on cytogenetics of UUTC; only 16 cases with abnormal karyotype, have been published [34,35]. Because of the great similarities between upper urinary tract carcinomas (UUTC) and bladder TCC with regard to etiology, histology, and natural disease history, one would assume that the tumorigenic mechanisms, including the karyotypic profile, are more or less identical.
Loss of material from chromosome 9 was seen in most informative UUTC [12]. The ubiquity of chromosome 9 involvement thus seems to be no less pronounced in UUTC than BC; this is in agreement with the view that this is an early and crucial event in the genesis of nearly all urinary tract TCC, regardless of their site and stage. The high frequency of chromosome 9 involvement in upper urinary tract transitional cell carcinogenesis makes this a logical target for a possible genetic marker to be used in the follow-up of patients with such tumors, something that would be particularly useful considering how frequent downstream disease spreading is in these patients.
Post-radiation uroepithelial carcinomas
Post-radiation tumors or second-primary radiation-induced cancers have distinct clinical and cytogenetic characteristics. These tumors develop in an irradiated field and should have a different histological type from that of the primary tumors. Usually they are infiltrative and of high grade at the time of diagnosis. Only 2 cases of post-radiation uroepithelial carcinoma with cytogenetics abnormality have been reported [35,36]. The karyotypic profile in both cases revealed near-diploid, karyotypically abnormal clones characterized by rather simple balanced and/or unbalanced translocations in each tumor. The intratumor karyotypic heterogeneity seen in these cases indicates massive polyclonality in contrast to the cytogenetic monoclonality consistently demonstrated in urinary tract TCC. This reflects the likelihood of the previous radiation carcinogenic effect and illustrating that post-radiation urinary tract tumors are distinct from other urinary tract TCC not only etiologically but also with regard to the pathogenetic mechanisms involved.
Supplementary Material
Additional File 1
Clinical and cytogenetic data on 205 cases of uroepithelial carcinomas
Click here for file
Acknowledgements
This work was supported by grants from the Medical Faculty of Al Neelain University and Sudanese cancer research group.
==== Refs
Silverman DT Hartge P Morrison AS Devesa SS Epidemiology of bladder cancer Hematol Oncol Clin North Am 1992 6 1 30 1556044
Badawi AF Mostafa MH Probert A O'Connor PJ Role of schistosomiasis in human bladder cancer: evidence of association, aetiological factors, and basic mechanisms of carcinogenesis Eur J Cancer Prev 1995 4 45 59 7728097
Herr HW Natural history of superficial bladder tumors: 10- to 20-year follow-up of treated patients World J Urol 1997 15 84 8 9144896
Aprikian AG Sarkis AS Reuter VE Cordon-Cardo C Sheinfeld J Biological markers of prognosis in transitional cell carcinoma of the bladder: current concepts Semin Urol 1993 11 137 44 8210835
Holmang S Hedelin H Anderstrom C Johansson SL The relationship among multiple recurrences, progression and prognosis of patients with stages Ta and T1 transitional cell cancer of the bladder followed for at least 20 years J Urol 1995 153 1823 6 7752327 10.1097/00005392-199506000-00022
Hoglund M Sall T Heim S Mitelman F Mandahl N Fadl-Elmula I Identification of cytogenetic subgroups and karyotypic pathways in transitional cell carcinoma Cancer Res 2001 61 8241 6 11719456
Lee R Droller MJ The natural history of bladder cancer. Implications for therapy Urol Clin North Am 2000 27 1 13 10696240 10.1016/S0094-0143(05)70229-9
Steffens J Nagel R Tumours of the renal pelvis and ureter. Observations in 170 patients Br J Urol 1988 61 277 83 3289673
Kiemeney LA Witjes JA Heijbroek RP Verbeek AL Debruyne FM Predictability of recurrent and progressive disease in individual patients with primary superficial bladder cancer J Urol 1993 150 60 4 8510276
Fadl-Elmula I Gorunova L Mandahl N Elfving P Lundgren R Mitelman F Heim S Cytogenetic monoclonality in multifocal uroepithelial carcinomas: evidence of intraluminal tumour seeding Br J Cancer 1999 81 6 12 10487605 10.1038/sj.bjc.6690643
Mitelman F www.Mitelmansdatabase.org 2008
Fadl-Elmula I Gorunova L Mandahl N Elfving P Lundgren R Mitelman F Heim S Karyotypic characterization of urinary bladder transitional cell carcinomas Genes Chromosomes Cancer 2000 29 256 65 10992300 10.1002/1098-2264(2000)9999:9999<::AID-GCC1034>3.0.CO;2-O
Ohgaki K Minobe K Kurose K Iida A Habuchi T Ogawa O Kubota Y Akimoto M Emi M Two target regions of allelic loss on chromosome 9 in urinary-bladder cancer Jpn J Cancer Res 1999 90 957 64 10551324
Simoneau AR Spruck CH Gonzalez-Zulueta M Gonzalgo ML Chan MF Tsai YC Dean M Steven K Horn T Jones PA Evidence for two tumor suppressor loci associated with proximal chromosome 9p to q and distal chromosome 9q in bladder cancer and the initial screening for GAS1 and PTC mutations Cancer Res 1996 56 5039 43 8895761
Cairns P Shaw ME Knowles MA Initiation of bladder cancer may involve deletion of a tumour-suppressor gene on chromosome 9 Oncogene 1993 8 1083 5 8096074
Simoneau M LaRue H Aboulkassim TO Meyer F Moore L Fradet Y Chromosome 9 deletions and recurrence of superficial bladder cancer: identification of four regions of prognostic interest Oncogene 2000 19 6317 6323 11175346 10.1038/sj.onc.1204022
Gonzalez-Zulueta M Shibata A Ohneseit PF Spruck CH Busch C Shamaa M El-Baz M Nichols PW Gonzalgo ML Elbaz M al. High frequency of chromosome 9p allelic loss and CDKN2 tumor suppressor gene alterations in squamous cell carcinoma of the bladder [published erratum appears in J Natl Cancer Inst 1995 Dec 6;87(23):1807] J Natl Cancer Inst 1995 87 1383 93 7658499
Nishiyama H Takahashi T Kakehi Y Habuchi T Knowles MA Homozygous deletion at the 9q32-33 candidate tumor suppressor locus in primary human bladder cancer Genes Chromosomes Cancer 1999 26 171 5 10469456 10.1002/(SICI)1098-2264(199910)26:2<171::AID-GCC10>3.0.CO;2-B
Berrozpe G Miro R Caballin MR Salvador J Egozcue J Trisomy 7 may be a primary change in noninvasive transitional cell carcinoma of the bladder Cancer Genet Cytogenet 1990 50 9 14 2253193 10.1016/0165-4608(90)90232-Y
Johansson B Heim S Mandahl N Mertens F Mitelman F Trisomy 7 in nonneoplastic cells Genes Chromosomes Cancer 1993 6 199 205 7685621
Richter J Wagner U Schraml P Maurer R Alund G Knonagel H Moch H Mihatsch MJ Gasser TC Sauter G Chromosomal imbalances are associated with a high risk of progression in early invasive (pT1) urinary bladder cancer Cancer Res 1999 59 5687 91 10582685
Terracciano L Richter J Tornillo L Beffa L Diener PA Maurer R Gasser TC Moch H Mihatsch MJ Sauter G Chromosomal imbalances in small cell carcinomas of the urinary bladder J Pathol 1999 189 230 5 10547580 10.1002/(SICI)1096-9896(199910)189:2<230::AID-PATH407>3.0.CO;2-8
Wistuba II Behrens C Virmani AK Milchgrub S Syed S Lam S Mackay B Minna JD Gazdar AF Allelic losses at chromosome 8p21-23 are early and frequent events in the pathogenesis of lung cancer Cancer Res 1999 59 1973 9 10213509
Zhao J Richter J Wagner U Roth B Schraml P Zellweger T Ackermann D Schmid U Moch H Mihatsch MJ Gasser TC Sauter G Chromosomal imbalances in noninvasive papillary bladder neoplasms (pTa) Cancer Res 1999 59 4658 61 10493521
Voorter C Joos S Bringuier PP Vallinga M Poddighe P Schalken J du Manoir S Ramaekers F Lichter P Hopman A Detection of chromosomal imbalances in transitional cell carcinoma of the bladder by comparative genomic hybridization Am J Pathol 1995 146 1341 54 7778674
Sauter G Gasser TC Moch H Richter J Jiang F Albrecht R Novotny H Wagner U Bubendorf L Mihatsch MJ DNA aberrations in urinary bladder cancer detected by flow cytometry and FISH Urol Res 1997 25 37 43 10.1007/BF00942046
Hopman AH Moesker O Smeets AW Pauwels RP Vooijs GP Ramaekers FC Numerical chromosome 1, 7, 9, and 11 aberrations in bladder cancer detected by in situ hybridization Cancer Res 1991 51 644 51 1985781
von Knobloch R Bugert P Jauch A Kalble T Kovacs G Allelic changes at multiple regions of chromosome 5 are associated with progression of urinary bladder cancer J Pathol 2000 190 163 8 10657014 10.1002/(SICI)1096-9896(200002)190:2<163::AID-PATH509>3.0.CO;2-0
Knowles MA Elder PA Williamson M Cairns JP Shaw ME Law MG Allelotype of human bladder cancer Cancer Res 1994 54 531 8 8275491
Smeets W Pauwels R Laarakkers L Debruyne F Geraedts J Chromosomal analysis of bladder cancer. III. Nonrandom alterations Cancer Genet Cytogenet 1987 29 29 41 3311350 10.1016/0165-4608(87)90028-8
Bentz M Plesch A Stilgenbauer S Dohner H Lichter P Minimal sizes of deletions detected by comparative genomic hybridization Genes Chromosomes Cancer 1998 21 172 5 9491330 10.1002/(SICI)1098-2264(199802)21:2<172::AID-GCC14>3.0.CO;2-T
Elfving P Cigudosa JC Lundgren R Limon J Mandahl N Kristoffersson U Heim S Mitelman F Trisomy 7, trisomy 10, and loss of the Y chromosome in short-term cultures of normal kidney tissue Cytogenet Cell Genet 1990 53 123 5 2369838
Heim S Mandahl N Jin Y Stromblad S Lindstrom E Salford LG Mitelman F Trisomy 7 and sex chromosome loss in human brain tissue Cytogenet Cell Genet 1989 52 136 8 2630185
Batata M Grabstald H Upper urinary tract urothelial tumors Urol Clin North Am 1976 3 79 86 936370
Fadl-Elmula I Gorunova L Mandahl N Elfving P Lundgren R Rademark C Heim S Cytogenetic analysis of upper urinary tract transitional cell carcinomas Cancer Genet Cytogenet 1999 115 123 7 10598145 10.1016/S0165-4608(99)00075-8
Fadl-Elmula I Bonaldi L Gorunova L Mandahl N Elfving P Heim S Cytogenetic heterogeneity in a second primary radiation-induced bladder carcinoma: ten karyotypically unrelated clones Cancer Genet Cytogenet 1998 105 134 7 9723030 10.1016/S0165-4608(98)00028-4
|
16042762
|
PMC1199611
|
CC BY
|
2021-01-04 18:01:01
|
no
|
Clin Pract Epidemiol Ment Health. 2005 Jul 21; 1:8
|
latin-1
|
Clin Pract Epidemiol Ment Health
| 2,005 |
10.1186/1745-0179-1-8
|
oa_comm
|
==== Front
Front ZoolFrontiers in Zoology1742-9994BioMed Central London 1742-9994-2-131607638910.1186/1742-9994-2-13ResearchRole of early experience in ant enslavement: a comparative analysis of a host and a non-host species Blatrix Rumsaïs [email protected] Claire [email protected] Laboratoire d'Ethologie Expérimentale et Comparée CNRS UMR 7153, Université Paris 13, 99 av. JB Clément 93430 Villetaneuse, France2005 2 8 2005 2 13 13 15 6 2005 2 8 2005 Copyright © 2005 Blatrix and Sermage; licensee BioMed Central Ltd.2005Blatrix and Sermage; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Ants use the odour of the colony to discriminate nestmates. In some species, this odour is learned during the first days following emergence, and thus early experience has a strong influence on nestmate discrimination. Slave-making ants are social parasites that capture brood of other ant species to increase the worker force of their colony. After emerging in the slave-maker nest, slave workers work as if they were in their own colony. We tested the hypothesis that early experience allows the deception of commonly enslaved species, while non-host species use a different mechanism, which does not involve learning.
Results
Pupae of a host species, Temnothorax unifasciatus, and a non-host species, T. parvulus, were allowed to emerge in the presence of workers of one of two slave-maker species, Chalepoxenus muellerianus or Myrmoxenus ravouxi. When T. unifasciatus was exposed to slave-makers for 10 days following emergence, they were more aggressive towards their own sisters and groomed the slave-maker more. T. parvulus gave a less clear result: while workers behaved more aggressively towards their sisters when exposed early to C. muellerianus workers, this was not the case when exposed early to M. ravouxi workers. Moreover, T. parvulus workers allogroomed conspecific nestmates less than T. unifasciatus. Allogrooming activity might be very important for the slave-makers because they are tended by their slaves.
Conclusion
Our findings show that early experience influences nestmate discrimination in the ant T. unifasciatus and can account for the successful enslavement of this species. However, the non-host species T. parvulus is less influenced by the early environment. This might help to explain why this species is never used by social parasites.
==== Body
Background
The early stages of an animal's life constitute a critical period during which the environment can have dramatic and sometimes irreversible effects on adult behaviour. The best example is imprinting, a process by which the individual develops an irreversible reference pattern from a stimulus present during a short period of time during its development. Lorenz [1] was the first to conceptualize the imprinting process and gave one of the most striking examples: during the few minutes after hatching, young birds developed an attachment to their parents if present, or a parental surrogate, such as Lorenz himself. Imprinting is only a special case of the processes by which early experience affects subsequent adult behaviour. For instance, in monkeys and humans, Harlow & Harlow [2] and Bowlby [3] noted the importance of attachment during the early phase after birth for the development of adult behaviour.
Despite a much simpler brain structure, insects display similarly pronounced plastic responses to the early environment. Influences of the preimaginal environment on adult behaviour have been demonstrated as early as the late 1930's in Drosophila [4], and were later documented and shown to be involved in various contexts in many other species (see review in [5]). These contexts can have a social dimension in social insects. Some ants, for example, learn the colony odour during the early phase of imaginal life and later use this template for recognizing brood [6-11] and discriminating among adults [12-15]. This critical period determines who benefits from altruistic behaviours shown by worker ants, which rarely reproduce directly but rely on indirect fitness benefits via reproduction of related individuals [16].
Slave-making ants are social parasites that exploit the labour force of other ant species. Slave-maker workers are specialized for conducting raids, wherein they seize brood from nearby host ant colonies and bring them back to their own nest [17]. When they emerge, the slave ants behave as if they were in their own colony. Among other routine ant tasks, they rear the slave-maker brood, defend the nest, and sometimes feed and groom the slave-maker workers. Altruistic acts of slaves are thus directed toward unrelated individuals. One hypothesis suggests that slave deception is possible because slaves are captured as pupae and learn the slave-maker colony odour after emergence [18-20].
Myrmoxenus ravouxi and Chalepoxenus muellerianus are among the most common slave-makers in Western Europe. They use several related species as slaves but show a preference for Temnothorax unifasciatus [21]. In the Mediterranean area, M. ravouxi, C. muellerianus and T. unifasciatus can be found in sympatry with T. parvulus, though T. unifasciatus seems to prefer more arid environments. T. parvulus has never been found as a slave, but the reason for this is unknown. We postulated that slavery would be prevented if nestmate discrimination in T. parvulus was not achieved through early learning of colony odour, because slave workers emerging in the slave-maker colony would recognize the slave-maker brood and workers as aliens.
Our study had two aims. First, we tested the hypothesis that deception of the host T. unifasciatus ensues from early learning of colony odour as a mechanism for nestmate discrimination. Second, we tested the hypothesis that early experience (within the first 10 days following emergence) affects the discrimination capabilities of T. parvulus workers less than those of T. unifasciatus. After manipulating the early experience of T. unifasciatus and T. parvulus, we tested and compared their discrimination capabilities.
Results
T. unifasciatus workers reared with M. ravouxi (batch A) or C. muellerianus (batch B) were significantly more aggressive towards their unfamiliar sisters than towards the familiar M. ravouxi or C. muellerianus workers (Table 1). Moreover they spent significantly more time allogrooming the familiar M. ravouxi or C. muellerianus workers than their unfamiliar sisters. Similarly, T. unifasciatus workers reared with conspecifics (batch C) were less aggressive towards their familiar sisters than towards the unfamiliar M. ravouxi and C. muellerianus workers. They also spent significantly less time allogrooming the unfamiliar M. ravouxi or C. muellerianus workers than the familiar sisters (Table 1).
Table 1 Discrimination capabilities of two ant species after manipulation of their early experience. Number of agonistic acts and duration (in seconds) of allogrooming behaviours of T. unifasciatus and T. parvulus workers tested during 10 minutes towards a M. ravouxi, C. muellerianus or a homospecific worker. Test workers were reared for 10 days after emergence either in the presence of M. ravouxi (batch A), C. muellerianus (batch B) or nestmate workers (batch C).
species tested treatment (N,C)* species presented aggression (mean ± SE) Z P grooming (mean ± SE) Z p
T. unifasciatus batch A (22,10) M. ravouxi 0.3 ± 0.2 16.5 ± 5.9
3.2 <0.002 3.2 <0.002
T. unifasciatus 14.3 ± 6.4 0.4 ± 0.4
batch B (22,10) C. muellerianus 0.8 ± 0.5 30.8 ± 9.5
3.5 <0.001 3.4 <0.001
T. unifasciatus 14 ± 4 0.03 ± 0.03
batch C (22,10) M. ravouxi 6.3 ± 3.2 0.9 ± 0.9
2.3 0.02 2.4 0.019
T. unifasciatus 0.5 ± 0.3 19.3 ± 13.4
2 0.047 2.1 0.033
C. muellerianus 3.2 ± 1.4 2.5 ± 2.3
T. parvulus batch A (22,5) M. ravouxi 2.8 ± 1.6 2.7 ± 1.9
0.4 0.69 1.6 0.11
T. parvulus 2 ± 0.8 0 ± 0
batch B (23,8) C. muellerianus 1.9 ± 0.7 4.9 ± 3.6
2.3 0.02 2.4 0.018
T. parvulus 4.6 ± 1.4 0 ± 0
batch C (22,8) M. ravouxi 3.4 ± 1.5 3.5 ± 2.4
2 0.049 0.7 0.47
T. parvulus 0.2 ± 0.1 0.1 ± 0.1
2.5 0.013 1.3 0.50
C. muellerianus 3.2 ± 1.8 0 ± 0
* N: number of ants tested; C: number of colonies the test ants originated from.
T. parvulus workers reared with M. ravouxi (batch A) did not show a significant difference in aggressive behaviour or in the time spent allogrooming between their unfamiliar sisters and the familiar M. ravouxi workers (Table 1). Conversely, T. parvulus workers reared with C. muellerianus (batch B) were significantly more aggressive towards their unfamiliar sisters than towards the familiar C. muellerianus workers, and they spent significantly more time allogrooming the familiar C. muellerianus workers than the unfamiliar sisters. T. parvulus workers reared with conspecifics (batch C) were significantly less aggressive towards their familiar sisters than towards the unfamiliar M. ravouxi or C. muellerianus workers, but the time they spent allogrooming the unfamiliar M. ravouxi or C. muellerianus workers and their familiar sisters was much reduced and did not differ significantly (Table 1).
T. unifasciatus and T. parvulus workers reared with conspecifics (batch C) did not show any significant difference in the number of aggressive acts towards M. ravouxi (6.3 ± 3.2 and 3.4 ± 1.5 respectively, Z = 1.2, p = 0.22) or C. muellerianus (3.2 ± 1.4 and 3.2 ± 1.8 respectively, Z = 0, p = 0.99). However, T. unifasciatus workers reared with conspecifics (batch C) spent significantly more time allogrooming familiar sisters than did T. parvulus (19.3 ± 13.4 and 0.1 ± 0.1 respectively, Z = 2.6, p = 0.01).
Discussion
Our results suggest that the environment experienced by T. unifasciatus during the first 10 days after emergence influences nestmate discrimination, at least in our artificial laboratory conditions. Indeed, workers that were exposed early to M. ravouxi and C. muellerianus displayed a very aggressive behaviour towards their own sisters and spent much time grooming the slave-makers. This suggests that the influence of the early environment could potentially account for the fact that T. unifasciatus slaves care for and defend the slave-makers. However, as host workers never emerge only with slave-makers present the mechanisms allowing enslavement could be different in nature. The same result was obtained for T. parvulus workers that were exposed to C. muellerianus. However, when exposed to M. ravouxi, they were not more aggressive toward their sisters than toward the slave-maker. These results partly confirm our hypothesis that the effect of early experience on the discrimination capabilities is reduced in T. parvulus when compared with T. unifasciatus workers.
The use of tethered, but live ants to record the reaction of test workers could have influenced their behaviour through the performance of antennation, biting, stridulation, etc., and the emission of pheromones. Indeed, several species of social parasites are known to use repellent, appeasement, and/or propaganda substances to usurp host nests [22]. The occurrence of such substances in our focal slave-maker species is likely and could have reduced aggressiveness of test workers. Moreover, several pupae of the test species were allowed to emerge in the glass tubes containing slave-makers. The early experience of test workers could then have been influenced by the other emerging test workers. Another potential source of bias in our results is the unequal representation of collected populations of Temnothorax in the experimental groups. The observed differences between our T. parvulus and T. unifasciatus might be due to differences between populations rather than species, if early experience had a smaller influence on nestmate discrimination in the Italian population. However, this possibility seems unlikely because species differences for fundamental mechanisms (such as the determinant of nestmate discrimination) are expected to be unaffected by population differences. The observed differences between the two species cannot be due to a mere difference in aggressiveness, because their levels of aggression towards slave-makers in the control experiment were not significantly different. However the difference between their levels of allogrooming was highly significant. A low basic level of allogrooming activity for T. parvulus workers was probably responsible for the fact that they did not groom their sisters more than the slave-makers in the control experiment. This makes it difficult to test the influence of early experience on allogrooming behaviour of T. parvulus.
Few other studies have demonstrated that early experience with slave-maker workers or brood elicits slave-maker care in the host species [23-25]. These studies dealt with species in the subfamily Formicinae; specifically, the slave-maker Formica sanguinea and its hosts F. fusca and F. cunicularia. Slave-making ants are known from only two subfamilies: the Formicinae and the Myrmicinae [26]. In the only study considering Myrmicinae, Alloway and Hare [27] showed that early learning was not necessary to explain the acceptance of the brood of the slave-maker Protomognathus americanus by enslaved T. longispinosus workers. Indeed, slave-maker larvae were preferentially accepted even when T. longispinosus workers were exposed early to conspecific larvae. Our results confirm that early experience can be important for successful ant enslavement in other myrmicine systems. Together, these studies across subfamilies seem to confirm the hypothesis that early behavioural plasticity of certain ant species has permitted or at least facilitated the evolution of slave-making habits [28], and show that other mechanisms are possible [27] but seem less common. This suggests that early behavioural plasticity could be a general prerequisite for the evolution of slave-making in ants [20]. The fact that early experience is important for the integration of the young slaves into the slave-maker colony might explain why M. ravouxi and C. muellerianus do not capture adult hosts during raids [29,30]. In fact, out of 10 genera displaying interspecific slavery, only one is known to result in the mixing of adult slaves into the slave-maker colony [26]: raids of the slave-making Strongylognathus commonly end in the fusion with the Tetramorium host nest [31]. However, the fusion occurs after a prolonged fight that often results in casualties on both sides. It is noteworthy that intraspecific colony fusion has been reported in the genus Tetramorium as a result of territorial competition [26].
For some slave-making species, host colony take-over by slave-maker queens requires acceptance by adult host workers that have had no previous exposure to slave-makers. The founding queen of many slave-making species has to enter a host nest, be accepted by the host workers, then kill and replace the resident queen [19]. Once the slave-making queen manages to kill the resident queen, the workers care for the new queen and rear her brood. The acceptance of the new queen does not require previous exposure to slave-makers because specialised behavioural and chemical strategies are involved [22]. For example, in the genus Polyergus, the young slave-maker queen does not bear any odour, and thus prevents the host workers from detecting her. The slave-maker queen then acquires the cuticular hydrocarbons of the host queen by physical contact and the host workers accept her as though she were there natal queen [32,33]. The founding queen of M. ravouxi enters the Temnothorax host nest, reaches the queen, and slowly throttles her to death [34]. The chemical mechanism responsible for the acceptation of the M. ravouxi queen by the host workers remains to be elucidated.
Queens of other slave-making species including C. muellerianus, Harpagoxenus and Protomognathus evict all adult ants from the host nest and keep only the host brood. They thus rely completely on the manipulation of early experience even during colony foundation. The life history of slave-making ants suggests that the use of adult slaves without previous exposure to the slave-maker is costly or involves a highly specialised strategy. Manipulation of early experience by the slave-maker appears as the most parsimonious strategy, which reinforces the suggestion that it facilitated the evolution of slavery.
Early behavioural plasticity is a widespread mechanism in animals and is involved in a number of host-parasite systems. However, in most cases, this mechanism is used by the parasite to find its host. The European cuckoo is known to rely on habitat, probably in addition to other cues, which is learned when reared by the host [35], but it does not seem to imprint directly on the host [36]. In parasitoid wasps, preimaginal imprinting is involved in host selection [5]. In slave-making ants, imprinting was also shown to influence host selection during raids and colony foundation [37,38].
A striking result of this study was that manipulating the early experience of T. parvulus had different consequences depending on the slave-maker species to which it was exposed; nestmate discrimination seemed to be influenced by C. muellerianus, but not by M. ravouxi. If T. parvulus was insensitive to experience at emergence due to a strict genetic system of odour discrimination or an earlier sensitive period, we would expect the same outcome for both slave-maker species. A possible explanation is that there is a limited set of odours that T. parvulus can learn and/or perceive at emergence. M. ravouxi and C. muellerianus likely have different chemical profiles and T. parvulus might be chemically more similar to C. muellerianus than to M. ravouxi. Thus, the latter could be out of the range of potentially learned patterns. As we show in our experiments, phylogenetically closely related ant species can display various sensitivities to the early environment. Thus, perhaps the mechanisms of nestmate discrimination differ as well.
Nestmate discrimination has been shown to be less influenced by social environment at emergence in the ant genus Camponotus than in the genus Formica [7]. It is therefore interesting that no species of Camponotus is parasitized by slave-makers, while many Formica species are hosts to slave-makers. Similarly, our results on T. parvulus, the non-slave species, suggest that plasticity at emergence is not the only determinant of nestmate discrimination. A first alternative possibility is that nestmate discrimination is determined during an earlier sensitive period. Indeed, studies in Camponotus floridanus and Cataglyphis cursor demonstrated the importance of larval stages for nestmate discrimination [39,40]. As most slave-makers capture pupae preferentially, a nestmate discrimination mechanism based on larval experience might prevent these species from being enslaved. A second possibility is that nestmate discrimination in T. parvulus has a genetic component or is based on self-referent phenotype matching. T. nylanderi, a species very closely related to T. parvulus, has nestmate discrimination cues based on colony environment, and especially nest site material [41]. Even if T. parvulus can be expected to rely in part on a similar mechanism in the field, it could not have influenced our results because colonies were reared in identical artificial nests. Whatever the determinant of nestmate discrimination in this species, it can be an obstacle to slavery because at least some elements of the template involved in discrimination are established before the slaves are captured. Moreover, T. parvulus workers do not groom their nestmates often, which could be problematic for the slave-makers because they are tended by the slaves. These characteristics might partly explain why T. parvulus has never been found as slave. However, other factors could explain why M. ravouxi and C. muellerianus does not enslave T. parvulus. Social parasites in the tribe Formicoxenini are phylogenetically closely-related to their hosts [42]. It might be impossible for the social parasites to exploit more distant species as hosts because they would not share the same ecological and microhabitat requirements, or the same communication system. Slave-maker species could also be under selection to match the colony odour of the host species. Coevolution occurs between slave-making ants and their hosts [43,44] and it has been shown that in slave-maker colonies both species have similar cuticular hydrocarbon profiles [45,46]. It might therefore be easier for the host species to learn the more similar profile of the parasite, than for the non-host species.
Conclusion
Early experience can have important consequences on the behaviour of adult ant workers and consequently on their inclusive fitness. Manipulation of early experience in an experimental (this study) or natural (social parasitism) context influences nestmate discrimination. However, we showed that the effect varies across species: it was more pronounced in the host than in the non-host species. Social parasitism is not evenly distributed in ants: most of the 200 known species occur in only two subfamilies (the Formicinae and the Myrmicinae), and in these subfamilies it is concentrated in a few genera [47]. The reasons for this distribution are still unknown and elucidating the fine scale mechanisms involved in nestmate discrimination across taxa might help to explain the evolution of slavery in ants.
Methods
Colonies of the two slave-maker species and of both host species were collected in northern Italy (Lago di Garda) and southern France (Vaison-la-Romaine and Grasse, we found no slave-makers near Grasse) in the spring and summer 2003. Because of a sampling bias, significantly more T. unifasciatus originated from southern France and more T. parvulus from northern Italy (Chi-square = 13, p = 0.022). Nests were housed in plastic boxes containing a thick layer of plaster. The print of a microscope slide covered by a glass plate provided a nest site [48]. Colonies were fed twice a week with frozen fruit flies and a mixture of honey and apple. Rearing conditions were as follows: day/night: 14 h-24°C/10 h-17°C. Genus names comply with Bolton's new classification of Formicidae [49]. The experiments were performed in August and September 2003.
Pupae of T. unifasciatus and T. parvulus that were close to emergence were removed from their colonies (unparasitized) and isolated in glass tubes with either M. ravouxi (experimental batch A), C. muellerianus (experimental batch B) or homo-colonial (control batch C) workers. Each glass tube contained one to three pupae (from the same colony) and two to six adult workers. The small numbers of slave-maker workers available precluded larger worker/pupae ratios, but at least two adult workers were used for each pupa. Slave-maker workers of M. ravouxi and C. muellerianus were obtained from 20 and 27 colonies respectively. Pupae close to emergence were recognizable by pigmentation, and emergence took place within two days of uniting pupae with their respective conditioning workers. Amputation of the distal segments of one leg of adult Temnothorax workers from batch C allowed us to distinguish them unambiguously from the newly emerged test workers. Slave-makers from batches A and B were not subjected to this amputation. Workers were tested 10 ± 0.3 (mean ± SE) days after emergence.
Each test worker was introduced into a circular arena of 1.5 cm2 where another worker was tethered by a nylon thread tied between the head and alitrunk. We were only interested in the behaviour of the test worker, thus we tethered the other worker to reduce the probability of interaction. Tethered workers could, however, still emit pheromones and engage in various other behaviours such as antennation, biting, stridulation etc. Tests started at least five minutes after tethering in order to let the tethered worker acclimate, but we cannot totally exclude the influence of pheromones on the behaviour of test workers. Duration of allogrooming and number of agonistic acts (mandibles opening, biting, stinging attempts) towards the tethered worker were recorded over 10 minutes (from the first encounter). Each test worker from batches A and B was tested successively against a slave-maker worker (M. ravouxi or C. muellerianus respectively) from its glass tube (hetero-colonial and familiar) and a worker from its colony of origin (homo-colonial and unfamiliar). Test order was controlled. Duration of allogrooming and number of agonistic acts were compared between the two tests using the Wilcoxon test for paired samples. Each test worker from batch C was tested successively against a worker from its glass tube (homo-colonial and familiar), a M. ravouxi worker, and a C. muellerianus worker (hetero-colonial, unfamiliar). All combinations of test order were equally represented. The test involving the homo-colonial and familiar worker was compared to each of the two tests involving a slave-maker worker using the Wilcoxon test for paired samples. Moreover, the number of agonistic acts towards the slave-makers was compared between T. unifasciatus and T. parvulus with a Mann-Whitney test for independent samples in order to test for a difference in levels of aggression between the two species. Differences in duration of allogrooming between the two species were similarly tested by comparing the performances of test workers from batch C in tests involving a homo-colonial and familiar worker. Tests were computed with Statistica 6.
Authors' contributions
RB conceived the study, carried out part of the experiments and data analysis, and drafted the manuscript. CS participated in the design of the study, carried out part of the experiments and data analysis. Both authors read and approved the final manuscript.
Acknowledgements
We thank Paul Devienne and Rémi Dillys for help with ant collecting, Stéphane Chameron for discussions and Jeremy Bono for correcting the English text. Andreas Schulz helped with species identification.
==== Refs
Lorenz K The companion in the bird's world Auk 1937 54 245 273
Harlow HF Harlow MK Learning to love Am Sci 1966 54 244 272 4958465
Bowlby JA Critical phases in the development of social responses in man and other animals New Biology 1953 14 25 37
Thorpe WH Further studies on preimaginal conditioning in insects Proc R Soc Lond B 1939 127 424 433
Caubet Y Jaisson P Lenoir A Preimaginal induction of adult behaviour in insects Q J Exp Psychol B 1992 44 165 178
Jaisson P Note préliminaire sur l'ontogenèse du comportement de soin au couvain chez la jeune fourmi rousse (Formica polyctena Forst): Rôle d'un mécanisme probable d'imprégnation C R Acad Sci, D, Sci Nat 1972 275 2721 2723
Carlin NF Halpern R Hölldobler B Schwartz P Early learning and the recognition of conspecific cocoons by carpenter ants (Camponotus spp.) Ethology 1987 75 306 316
Hare JF Alloway TM Early learning and brood discrimination in leptothoracine ants (Hymenoptera: Formicidae) Anim Behav 1987 35 1720 1724
Le Moli F Passetti M Olfactory learning phenomena and cocoon nursing behaviour in the ant Formica rufa L Boll Zool 1978 45 389 397
Le Moli F Mori A Early learning and cocoon nursing behaviour in the red wood ant Formica lugubris Zett. (Hymenoptera: Formicidae) Boll Zool 1982 49 93 97
Fénéron R Jaisson P Ontogeny of nestmate brood recognition in a primitive ant, Ectatomma tuberculatum Olivier (Ponerinae) Anim Behav 1995 50 9 14 10.1006/anbe.1995.0215
Morel L Relation entre comportement agressif et privation sociale précoce chez les jeunes immatures de la fourmi Camponotus vagus Scop. (Hymenoptera: Formicidae) C R Acad Sci, D, Sci Nat 1983 296 449 452
Errard C Role of early experience in mixed-colony odor recognition in the ants Manica rubida and Formica selysi Ethology 1986 72 243 249
Carlin NF Hölldobler B The kin recognition system of carpenter ants (Camponotus spp.). I. Hierarchical cues in small colonies Behav Ecol Sociobiol 1986 19 123 134 10.1007/BF00299947
Alloway TM Ryckman D Learned social attachment to queens in Leptothorax ambiguus Emery ant workers Behaviour 1991 118 235 243
Hamilton WD Altruism and related phenomena, mainly in social insects Ann Rev Ecol Syst 1972 3 193 232 10.1146/annurev.es.03.110172.001205
Buschinger A Evolution of social parasitism in ants Trends Ecol Evol 1986 1 155 160 10.1016/0169-5347(86)90044-3
Jaisson P Hepper PG Kinship and fellowship in ants and social wasps Kin recognition 1991 Cambridge, Cambridge University Press 60 93
D'Ettorre P Heinze J Sociobiology of slave-making ants Acta Ethol 2001 3 67 82 10.1007/s102110100038
Le Moli F On the origin of slaves in dulotic ant societies Boll Zool 1980 47 207 212
Buschinger A Kipyatkov VE Socially parasitic formicoxenine ants from Western Europe – a review (Hymenoptera, Formicidae) Proceedings of the International Colloquia on Social Insects 1997 3-4 St. Petersburg, Russian Language Section of the IUSSI, Socium 1 9
Lenoir A D'Ettorre P Errard C Hefetz A Chemical ecology and social parasitism in ants Ann Rev Entomol 2001 46 573 599 11112180 10.1146/annurev.ento.46.1.573
Le Moli F Mori A Why a slave ant species accepts and cares for cocoons of its slave-maker Insectes Soc 1987 34 136 141
Le Moli F Mori A The influence of the early experience of worker ants on enslavement Anim Behav 1985 33 1384 1387
Jaisson P L'impregnation dans l'ontogenèse des comportements de soins aux cocons chez la jeune fourmi rousse (Formica polyctena Forst.) Behaviour 1975 52 1 37
Hölldobler B Wilson EO The ants 1990 Cambridge, Belknap Press 732
Alloway TM Hare JF Experience-independent attraction to slave-maker ant larvae in host-species ant workers (Leptothorax longispinosus; Hymenoptera: Formicidae) Behaviour 1989 110 93 105
Jaisson P L'imprégnation dans l'ontogenèse du comportement de soins aux cocons chez les formicines Proceedings of the 7th International Congress of the IUSSI 1973 London, United Kingdom, 176 181
Buschinger A Ehrhardt W Winter U The organization of slave raids in dulotic ants - a comparative study (Hymenoptera; Formicidae) Z Tierpsychol 1980 53 245 264
Schumann RD Raiding behavior of the dulotic ant Chalepoxenus muellerianus (Finzi) in the field (Hymenoptera: Formicidae, Myrmicinae) Insectes Soc 1992 39 325 333 10.1007/BF01323952
Sanetra M Güsten R The socially parasitic ant genus Strongylognathus Mayr in North Africa (Insecta: Hymenoptera: Formicidae) Zootaxa 2001 20 1 20
D'Ettorre P Errard C Chemical disguise during colony founding in the dulotic ant Polyergus rufescens Latr. (Hymenoptera, Formicidae) Insect Soc Life 1998 2 71 77
Johnson CA Vander Meer RK Lavine B Changes in the cuticular hydrocarbon profile of the slave-maker ant queen, Polyergus breviceps Emery, after killing a Formica host queen (Hymenoptera: Formicidae) J Chem Ecol 2001 27 1787 1804 11545371 10.1023/A:1010456608626
Buschinger A Evolution, speciation, and inbreeding in the parasitic ant genus Epimyrma (Hymenoptera, Formicidae) J Evol Biol 1989 2 265 283 10.1046/j.1420-9101.1989.2040265.x
Teuschl Y Taborsky B Taborsky M How do cuckoos find their hosts? The role of habitat imprinting Anim Behav 1998 56 1425 1433 9933539 10.1006/anbe.1998.0931
Brooke ML Davies NB A failure to demonstrate host imprinting in the cuckoo (Cuculus canorus) and alternative hypotheses for the maintenance of egg mimicry Ethology 1991 89 154 166
Schumann RD Buschinger A Imprinting effects on host-selection behaviour of colony-founding Chalepoxenus muellerianus (Finzi) females (Hymenoptera, Formicidae) Ethology 1994 97 33 46
Schumann RD Buschinger A Imprinting effects on host-selection behavior of slave-raiding Chalepoxenus muellerianus (Finzi) workers (Hymenoptera: Formicidae) Ethology 1995 99 243 251
Isingrini M Lenoir A Jaisson P Preimaginal learning as a basis of colony-brood recognition in the ant Cataglyphis cursor Proc Natl Acad Sci USA 1985 82 8545 8547 16593637
Carlin NF Schwartz PH Pre-imaginal experience and nestmate brood recognition in the carpenter ant, Camponotus floridanus Anim Behav 1989 38 89 95
Heinze J Foitzik S Hippert A Hölldobler B Apparent dear-enemy phenomenon and environment-based recognition cues in the ant Leptothorax nylanderi Ethology 1996 102 510 522
Baur A Sanetra M Chalwatzis N Buschinger A Zimmermann FK Sequence comparisons of the internal transcribed spacer region of ribosomal genes support close relationships between parasitic ants and their respective host species (Hymenoptera: Formicidae) Insectes Soc 1996 43 53 67 10.1007/BF01253956
Foitzik S DeHeer CJ Hunjan DN Herbers JM Coevolution in host-parasite systems: behavioural strategies of slave-making ants and their hosts Proc R Soc Lond B 2001 268 1139 1146 10.1098/rspb.2001.1627
Hare JF Alloway TM Prudent Protomognathus and despotic Leptothorax duloticus: differential costs of ant slavery Proc Natl Acad Sci USA 2001 98 12093 12096 11572933 10.1073/pnas.201397998
Kaib M Heinze J Ortius D Cuticular hydrocarbons profiles in the slave-making ant Harpogoxenus sublaevis and its hosts Naturwissenschaften 1993 80 281 285 10.1007/BF01135915
Brandt M Heinze J Schmitt T Foitzik S A chemical level in the coevolutionary arms race between an ant social parasite and its hosts J Evol Biol 2005 18 576 586 15842487 10.1111/j.1420-9101.2004.00867.x
Wilson EO The insect societies 1971 Cambridge, Harvard University Press 548 7276518
Blatrix R Herbers JM Intracolonial conflict in the slave-making ant Protomognathus americanus: dominance hierarchies and individual reproductive success Insectes Soc 2004 51 131 138 10.1007/s00040-003-0710-4
Bolton B Synopsis and classification of Formicidae Mem Amer Entomol Inst 2003 71 1 370
|
16076389
|
PMC1199612
|
CC BY
|
2021-01-04 16:38:34
|
no
|
Front Zool. 2005 Aug 2; 2:13
|
utf-8
|
Front Zool
| 2,005 |
10.1186/1742-9994-2-13
|
oa_comm
|
==== Front
Global HealthGlobalization and Health1744-8603BioMed Central London 1744-8603-1-131606096110.1186/1744-8603-1-13CommentaryHIV/AIDS: global trends, global funds and delivery bottlenecks Coovadia Hoosen M [email protected] Jacqui [email protected] Victor Daitz Professor of HIV/AIDS Research, Nelson R. Mandela School of Medicine, University of Kwazulu Natal, Private Bag X7 Congella, 4013, South Africa2 AIDS Research Co-Ordinator, Nelson R. Mandela School of Medicine, University of Kwazulu Natal, Private Bag X7 Congella, 4013, South Africa2005 1 8 2005 1 13 13 14 12 2004 1 8 2005 Copyright © 2005 Coovadia and Hadingham; licensee BioMed Central Ltd.2005Coovadia and Hadingham; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Globalisation affects all facets of human life, including health and well being. The HIV/AIDS epidemic has highlighted the global nature of human health and welfare and globalisation has given rise to a trend toward finding common solutions to global health challenges. Numerous international funds have been set up in recent times to address global health challenges such as HIV.
However, despite increasingly large amounts of funding for health initiatives being made available to poorer regions of the world, HIV infection rates and prevalence continue to increase world wide. As a result, the AIDS epidemic is expanding and intensifying globally. Worst affected are undoubtedly the poorer regions of the world as combinations of poverty, disease, famine, political and economic instability and weak health infrastructure exacerbate the severe and far-reaching impacts of the epidemic.
One of the major reasons for the apparent ineffectiveness of global interventions is historical weaknesses in the health systems of underdeveloped countries, which contribute to bottlenecks in the distribution and utilisation of funds. Strengthening these health systems, although a vital component in addressing the global epidemic, must however be accompanied by mitigation of other determinants as well. These are intrinsically complex and include social and environmental factors, sexual behaviour, issues of human rights and biological factors, all of which contribute to HIV transmission, progression and mortality. An equally important factor is ensuring an equitable balance between prevention and treatment programmes in order to holistically address the challenges presented by the epidemic.
==== Body
Introduction
Globalisation, narrowly defined by Joseph Stiglitz as "the removal of barriers to free trade and the closer integration of national economies." [1], has a much wider sweep and also affects the political, cultural and social life of populations across the globe. The health sector is no exception. As Barnett and Whiteside [2] point out, health and wellbeing are international concerns and global goods, and inherent in the epidemic are lessons to be learned regarding collective responsibility for universal human health.
AIDS is a pandemic of unprecedented pervasiveness, spreading to the furthest corners of the world. Globalisation is both midwife to the spread of the disease, as modern travel facilitates rapid dissemination of HIV infection across national borders, and, through concerted global action, triumphant conqueror over its devastating impact and expansion. Despite poorer countries having ever greater access to money, effective and affordable interventions, and technical support, the epidemic continues unabated in many of the resource-constrained regions of the world. A major reason for this continued spread is the numerous constraints within health systems in developing countries, which impact upon government policy, strategic and health policy management and health service delivery.
In this paper, we discuss trends in the global AIDS epidemic as well as the numerous global funds that are available to meet the challenges posed by the disease. We also highlight the need for equal prominence to be given to both treatment and prevention programmes in the global fight against HIV/AIDS. Lastly, we examine how bottlenecks in health systems of developing countries reduce the effectiveness of such aid and suggest ways in which these blockages can be eradicated through systematic strengthening of health systems.
Trends in the global epidemic
Despite increased resources being available to address the global AIDS challenge, the infection continues to spread. Table 1 shows the regional progression in HIV infection rates over the last five years.
Table 1 Trends in HIV Infections By Region
Region No of people living with HIV (end of 1998) [39] No of people living with HIV (end of 2003) [40] % Increase 1998–2003
Sub-Saharan Africa 22,500,000 25,000,000 11%
South & South-East Asia 6,700,000 6,500,000 -3% 1
Eastern Europe & Central Asia 270,000 1,300,000 381%
Western Europe 500,000 580,000 16%
East Asia 560,000 900,000 61%
Oceania 12,000 32,000 167%
North Africa & Middle East 210,000 480,000 129%
North America 890,000 1,000,000 12%
Caribbean 330,000 430,000 30%
Latin America 1,400,000 1,600,000 14%
TOTAL 33,372,000 37,822,000 13%
1 this apparent decrease is due to inconsistencies in data collection methods between earlier and later years, as well as revised estimates by UNAIDS.
HIV prevalence is intensifying in most regions, with sub-Saharan Africa, Eastern Europe and Central Asia being the worst hit, accounting for approximately 79% of new infections between 1998 and 2003. Although the greatest number of people living with HIV are in sub-Saharan Africa, of equal concern is the growing epidemic in Central Asia [3].
The epidemiology of the disease differs between regions. It has been suggested that, due to dissimilar patterns of sexual behaviour between Africa and Asia, the extent of the spread to the heterosexual population in Asia will be circumscribed. In most of sub-Saharan Africa, HIV spreads through an intricate web of relationships from sex workers to male clients to female spouses/partners. According to Peter Piot of UNAIDS, females in Africa generally report more sexual partners than their Asian counterparts [4]. In most of Central Asia transmission is virtually linear, from intravenous drug users to sex workers to male clients to female spouses/partners, with women tending to monogamy [4]. The next decade will attest to the accuracy or error of this prediction. Rising prevalence is, however, not confined to developing countries, as an increase in the number of HIV infections is evident in all other regions except South and South East Asia (where inconsistencies in data collection methods have tended to skew the figures).
Several trends shape the HIV epidemiological curve
• An increasingly mobile global population exacerbates the risk of HIV transmission. The increasing volume of international travel contributes to the spread of sexually transmitted infections, including HIV [5]. Refugee populations arising from areas of conflict, estimated by the United Nations High Commission for Refugees to number 9,7 million worldwide [6], are at higher risk, as are internal migrants within countries, who oscillate between rural and urban milieux. According to the International Labour Organisation, at the beginning of the 21st century, 120 million workers worldwide were migrants [7].
• Females are more at risk of contracting HIV than males. In 1997, women accounted for 41% of people living with HIV worldwide. This figure had risen to almost 50% by 2002. This gender-bias is especially apparent in sub-Saharan Africa, where the majority of those infected are women and girls. Widespread wars and regional conflicts in Africa escalate, by orders of magnitude, the risk of rape of women and girls. The low social status of women, risky sexual practices, and endemic poverty in Africa contribute to the spread of the disease. The impact on women is less marked in Asia (where 28% of those infected are women), although women's low socio-economic status renders them more susceptible to infection. Women's increased vulnerability to HIV infection is not confined to developing countries. Between 2001 and 2003, the percentage of HIV-infected who are women increased in North America from 20% to 25%, and in Oceania from 17% to 19%, suggesting that gender inequalities underpin the transmission of HIV [8].
• The impact of HIV mortality is greatest on people in their 20's and 30's; this severely distorts the shape of the population pyramid in affected societies. Projections indicate that mortality rates will increase: The UN predicts that, in seven selected countries in sub-Saharan Africa, 14 million AIDS-related deaths will occur between 1995 and 2025 [9]. UNAIDS projections indicate that, unless the AIDS response is greatly increased, populations in 38 African countries will decrease by 14% by 2025 [8].
• In sub-Saharan Africa, it is estimated that 12 million children have lost one or both parents to AIDS, a figure which is expected to increase to 18 million by 2010. Even in countries where HIV infections have plateaued, the number of orphans continues to rise due to the time lapse between infection and death of parents [8].
• Agricultural output, the cornerstone of production in agrarian economies, is decreasing as a result of increased mortality in the workforce, resulting in what has been termed "new-variant famine". Studies predict that in the ten most severely affected African countries, the agricultural workforce will decline by 10–26% by 2020 [9]. Bertolt Brecht ascribed these disasters to human greed and folly: "Famines do not simply occur – they are organized by the grain trade." New-variant famine, however, is the consequence of the mutually reinforcing intercessions of human frailty and a social disease. The former from a paucity of timeous responses to the epidemic by the ruling classes, aggravated by communities steeped in stigma, fear and discrimination, and the latter from a mix of biology and human propensity to risky sexual behaviour. The combination of lost production and resulting malnutrition increase susceptibility to disease [10].
• The macroeconomic repercussions of the epidemic vary, depending on the industries underpinning the economy and degree of HIV prevalence. UNAIDS postulates that any deceleration in economic growth (as measured by Gross Domestic Product) will be offset by similar reductions in population numbers due to increased mortality and therefore resource consumption [8]. A faster decline in population size relative to GDP should theoretically result in an increase in per capita GDP. Econometric research, however, has shown that AIDS has either an insignificant impact on per capita GDP, or actually decreases it [11]. The qualitative effects of higher mortality are also considerable: the erosion of social and intellectual capital and decreased investment in populations of the future have far-reaching consequences for society as a whole [9].
• The major economic impact is microeconomic. Individual households are primarily responsible for coping with the repercussions of AIDS, and as such bear the brunt of the epidemic. This translates into increased healthcare expenses, funeral charges and education costs for households. In areas where stigma prevails, the psychological impacts of the disease increase the burden.
• Impact on the workplace is also considerable, translating into productivity losses and increased costs to employers due to staff illnesses and deaths, higher health insurance premiums and low morale [8]. In addition, household demand for goods and services may decline due to lower income and levels of consumption, resulting in the contraction of resource production [9].
Table 2 shows in summary the demographic impacts of the epidemic, while Table 3 shows the impacts on various other aspects of society. The ramifications of an epidemic of this nature and scale will be felt long after incidence of the disease has peaked, predicted in the case of HIV to be in 2040 [12]. By way of comparison, the consequences of the Black Death (1347 – 1351) extended far beyond the life of the epidemic itself, exerting influence for about 150 years in Europe [13]. In order to mitigate these effects, massive investments in prevention, treatment and care programmes and in broad development initiatives must be given priority.
Table 2 Summary of demographic impacts of AIDS
Demography [9] Without AIDS With AIDS Without AIDS With AIDS Without AIDS With AIDS
1995 – 2000 2010 – 2015 2020 – 2025
Life expectancy at birth (years) 63.9 62.4 68.4 64.2 70.8 65.9
Number of deaths (millions) 159 170 174 207 193 231
Crude death rate per 1,000 9.0 9.6 8.1 9.8 8.0 10.1
Infant mortality rate per 1,000 66.4 67.5 49.8 51.3 40.9 42.1
Child mortality rate per 1,000 93.9 98.8 68.9 75.8 56.1 62.3
Population size (millions) 3666 3639 4310 4204 4805 4599
1,UNAIDS Population Division, 2003
Table 3 Summary of sectoral impacts of AIDS
GDP [41, 42] • Annual decrease of between 2 and 4% with AIDS
Households [9] • Decreased household income • Increased expenditure on healthcare
• More women and child-headed households
• More vulnerable to poverty
Firms [9] • Increased healthcare costs
• Greater absenteeism
• Loss of skilled labour and institutional memory
• Decreased demand for goods → decreased income
• Lower staff morale → lower productivity
Agriculture [9] • Loss of agricultural workforce:
• reduction in cultivated land → decreased yields
• smaller harvest size and less crop variety
• loss of agricultural knowledge
• lower remittances sent home
Education [9] • Loss of teachers → reduction in supply and quality of educational facilities and services
• Increased medical and staff training costs
• Reduction in pupil numbers due to non-enrolment /sickness/deaths
• Reversal in progress made in primary education
Health [9] • Absenteeism and deaths of health workers due to illness:
• reduction in supply and quality of health services
• increased training costs
• erosion of knowledge base
• Quality of care may suffer due to stigmatisation of HIV+ patients
• Increased public health expenses → higher burden on private health care system
• Increased demand for donor funding to address HIV/AIDS challenge
• High demand for AIDS treatment crowds out treatment of other diseases
2Dixon, McDonald and Roberts (2002); Cornia and Zagonaria (2002)
Global funds
Various global initiatives and collaborations are addressing the global HIV/AIDS challenge. For example, the United Nations Millennium Development Declaration, signed in 2000 by 189 nations, encompasses eight Millennium Development Goals (MDGs), three of which are health related: reducing child mortality, improving maternal health, and combating HIV/AIDS, malaria and other diseases, by 2015 [14]. Many international organizations have been set up to assist in funding and implementing HIV prevention and care programmes and related health initiatives worldwide. These include the President's Emergency Plan For AIDS Relief (PEPFAR); the Global Fund to fight AIDS, Tuberculosis and Malaria; RollBack Malaria, the Global Alliance for Vaccines and Immunization; the Global Health Council; Médecins sans Frontiers; the Bill and Melinda Gates Foundation; the World Bank Multi Country HIV/AIDS Programme (MAP); the Accelerating Access Initiative and the William J. Clinton Presidential Foundation. These organizations contribute increasing amounts of money to confront AIDS and other pressing global health issues. UNAIDS [8] reports that in 1996, approximately US$330 million was available for HIV/AIDS initiatives worldwide, a figure which had risen to US$4.7 billion by 2003. Although this represents a huge increase in funding, it is still less than half the amount of US$12 billion that is now required, and this exigency is expected to rise to US$20 billion by 2007.
Despite the large amount of aid being made available in addressing the AIDS epidemic, shortfalls in both money and numbers of people being reached are apparent. Of the estimated 6 million people in developing countries who are in need of ART, only 400,000 currently receive it. Of these, 208,000 are in Brazil alone [15]. Even if the World Health Organization's '3 by 5' effort, which aims to provide treatment to 3 million people by the end of 2005, is successful, it will have addressed only 50% of the demand for treatment at the current level of need. The MDGs are unlikely to be met at the current rates of progress, with the worst affected countries likely to make the least headway.
Another issue of concern is that the focus of many of these programmes is on treatment rather than prevention of HIV. Initiatives geared to increasing the delivery of treatment to developing countries has increased substantially since 2001, when the Declaration of Commitment on HIV/AIDS was signed by 189 member states of the United Nations [16]. For example:
• The Global Fund to fight AIDS, Tuberculosis and Malaria has approved funding for the provision of antiretroviral therapy (ART) to 700,000 people [17].
• The World Bank plans to increase financial assistance for ART programmes in eligible countries [17].
• PEPFAR's focus is largely on treatment [18] and plans to deliver ART to 2 million people in sub-Saharan Africa and the Caribbean by 2007 [17].
• The focus of the WHO's "3 by 5" programme is also exclusively on the treatment of HIV [15].
Current data suggests that approximately 33% of funding for AIDS initiativesbe allocated for treatment and care, with approximately 51% for prevention programmes [19]. Schwartländer et al [20] advocate a similar split in fund allocation between treatment and care on the one hand, and prevention initiatives on the other.
During the early stages of the epidemic, programmes designed to prevent HIV had rightly been the prime endeavour of poorer countries; indeed there was little else on offer. Even when the prospects of effective specific antiretroviral treatment improved after 1996, many scientists and health professionals remained committed to a dominant role of prevention over treatment and care. Prevention services, they believed, were not restricted to prophylaxis but included palliative care and the management of opportunistic infections. The latter were inexpensive and cost-effective; the concern was that highly active antiretroviral therapy (HAART), being more costly, would drain money from prevention programmes. But the direct and indirect financial, social, economic, political and security costs of failing to introduce effective prevention measures are undeniably very high. Based on figures from previous studies [21], Marseille et al modelled the cost-effectiveness of HAART against cotrimoxazole prophylaxis, and found that the ratio between the cost-effectiveness of HAART and prevention is US$350:US$12.50. (a ratio of 28:1). In human terms, for every life-year gained through HAART, 28 life-years could have been gained through prevention [22].
Marseille's evaluation, however, disregards the synergy between prevention and treatment interventions. Prevention, although an important component in addressing the epidemic, is inadequate in isolation. The low rates of uptake of preventive measures in many developing countries, which we discuss later, do not diminish this assertion. In addition to prevention programmes, the provision of HAART is not only financially feasible, but morally imperative. The difficulties associated with introducing ART are well known: there is no eradication of the virus, therefore treatment is lifelong; adherence lapses occur; drug formulations are not optimised; drug toxicities are frequent; drug-drug interactions complicate management and drug resistance requires special attention. In addition, there are aspects of HAART management which are still not settled – optimal start time and regimen sequence, the meaning of regime failure, and the sustainable reduction of resistance. The World Health Organization argues that the provision of ART, through its ability to prolong life and alleviate fears about HIV, can both change attitudes to the disease and, in combination with prevention, greatly reduce HIV transmission. It is suggested that resource-constrained countries such as Senegal, Thailand and Brazil, which introduced HAART early, are also the countries with the greatest success in controlling the epidemic. A 70% decline in AIDS-related deaths in affluent countries, where ART is available to the majority of the population, is cited to support this assertion [15].
It is becoming apparent that the advantages of ART might be offset by factors which may, on balance, fail to prevent or reduce transmission of the virus. These include disinhibition of risky sexual behaviour, the spread of drug-resistant strains, and an increased risk of exposure to HIV due to the improved survival rates of infected persons. In the context of the developing world, these putative negative impacts are likely exacerbated for several reasons:
• Early detection of HIV is rare. Patients tend to present in a state of advanced disease when viral load is high and the patient is very ill. This usually follows a period of relative good health during which maximal sexual activity and consequent high transmission of virus has occurred.
• Provision of ARVs may reduce condom use [17]
• ART efficacy may diminish as successive ARV regimes are used [23]
Despite these inherent hazards, given the continued escalation in HIV infections worldwide, it is reasonable and compassionate to attempt to achieve synergies between HAART and prevention services through their simultaneous implementation.
UNAIDS has identified a comprehensive list of prevention, treatment and care services which define standard services for HIV/AIDS control (Table 4). Most of these interventions are affordable by poor countries, either through their own budgets or from donor funds. A key issue is incorporation of applicable interventions into existing health services and programmes.
Table 4 Standard HIV/AIDS Interventions used by UNAIDS to measure resource needs and resource availability in low-and middle-income countries
Prevention Interventions
1. Mass media campaigns
2. Voluntary counseling and testing (VCT)
3. Condom social marketing
4. School-based AIDS education
5. Peer education for out-of-school youth
6. Outreach programmes for sex workers and their clients
7. Outreach programmes for men who have sex with men
8. Harm-reduction programmes for injecting users
9. Blood safety
10. Public sector condom promotion and distribution
11. Treatment of sexually transmitted infections
12. Workplace prevention programmes
13. Prevention of mother-to-child transmission
14. Post-exposure prophylaxis (PEP)
15. Safe injections
16. Universal precautions
17. Policy, advocacy, administration and research
Care Services
1. Palliative care
2. Diagnosis of HIV infection (HIV testing)
3. Treatment for opportunistic infections
4. Prophylaxis for opportunistic infections
5. Antiretroviral (ARV) therapy, including laboratory services for monitoring treatment
Orphan Support
1. Community support for orphan care
2. Orphanages
3. School fee support for orphans
UNAIDS, 2003
Health systems capacity
An over-reliance on donor funds can reduce the long-term sustainability of aid programmes, and the reduced absorptive capacity of recipient countries for such assistance often results in bottlenecks, preventing aid packages from being used where they are most needed. As a result, despite higher levels of acceptance of AIDS by certain governments, a global climate of increased political stability and economic growth, and greater public access to information and advocacy, inequitable access to treatment and prevention persists. While challenges experienced by households and communities in terms of providing resources for home-based care are also significant hindrances to the effective delivery of care, shortcomings inherent in health systems constitute the major blocks in channeling ever-increasing amounts of aid to those most in need. It follows that inequities in the provision of healthcare services may escalate in the coming years unless efficiency is coupled with justice in the construction of national health systems.
Constraints relating to supply within health systems, including finance, information systems, human resources, drugs and logistics [14], as well as those on the demand-side, such as increased patient numbers, and stigma and discrimination among communities [8], hinder progress.
The example of introducing prevention of mother to child transmission (PMTCT) programmes, which are among the simplest and most cost-effective of anti-HIV programmes available, into national health systems, is illustrative of the challenges faced by developing countries. Single dose Nevirapine (a dose each to mother during delivery and to her newborn) is the most widely used regimen for PMTCT, having the advantages of simplicity, affordability, and effectiveness. Most programmes and agencies, including UNICEF, the Elizabeth Glaser Pediatric AIDS Foundation (EGPAF), and state authorities, have found that in developing countries, of the women who should be given ART, only a minority receive the drugs. Even fewer infants are given their prophylactic dose of Nevirapine. Until recently, experience suggested that, despite wide variations between countries, in general, of the HIV positive women attending antenatal clinics, probably < 20% received ARVs. Neff Walker [24] has estimated that, of the 2.1 million pregnant women who are HIV positive in any given year globally (excluding high-income countries), only 200,000 receive PMTCT interventions.
Current information from some centres, however, suggests that uptake is improving. Data from studies undertaken in Kwazulu Natal, South Africa – a region severely affected by the epidemic – show that, for 150,000 deliveries per annum, PMTCT coverage increased from 10% in 2001 to 78% in 2003/04 (Figure 1) [25]. Reasons for such improvements in a number of countries may be attributed to:
Figure 1 Coverage of PMTCT Programme in Kwazulu Natal, South Africa between 2001 and 2004. Kwazulu Natal Dept of Health (2004)
• Increased awareness of HIV due to the expansion of education, information and communication programmes, which results over time in increased acceptance of the disease and its implications. This in turn fosters greater community mobilisation in providing support groups, home-based care initiatives, orphan care and food aid.
• More rapid and reliable testing methods, including 'opt-out' options, better counseling programmes and facilities, and the inclusion of partners in both testing and counseling programmes
• Enhanced record keeping, including improved identification systems for both mothers and infants.
• Advances in drug technology and therapies with resultant wider availability of ARVs for both mothers and infants.
Figure 2, taken from the same study, shows that despite this increase, only 59% of women attending antenatal clinics who test HIV-positive actually receive Nevirapine. Much of this attrition is due to failing health systems, although other factors, such as stigma and discrimination, also have an effect on poor uptake.
Figure 2 PMTCT Uptake at maternity hospitals, clinics and community health centres in Kwazulu Natal, South Africa (June 2001 – August 2004). Kwazulu Natal Dept of Health (2004)
Health system reform
The World Health Report (2004) states that "The 3 by 5 initiative...cannot be implemented in isolation from a regeneration of health systems." [26]. Several studies support this statement, reflecting the unfavourable conditions in the health care systems of developing regions [27,28].
UNAIDS [8] suggests that, in order to build capacity, an approach which incorporates training, technical assistance and access to improved guidelines and tools should be adopted by funders. In order to utilize resources effectively recipient countries need to undertake thorough planning processes whereby goals relevant to that country are set and allocation of funds is made according to need [29,30].
However, constraints may have multiple causes, both within and external to the health system itself, which may themselves be interdependent. Two approaches to overcoming constraints may be identified: dealing with constraints specific to the disease across all aspects of the health system, or addressing specific weaknesses in the health system across all diseases. It has been argued that disease-specific programmes can build skills and develop effective management structures to allow health services to cope with the demands placed on them [31]. The scale and nature of the HIV epidemic is such that it is generally the most pressing health challenge faced by developing countries. As such, an approach specific to the disease itself could be seen as the most effective way of building the capacity of health systems in countries of need, as it may be a more manageable way to address weaknesses in the health system while at the same time delivering short-term returns. This approach can, however, result in parallel systems being set up, and can cause disruptions in day to day healthcare provision. There are multiple overlaps in the health service requirements for HIV/AIDS and those for other diseases, which constitute a compelling argument to avoid as far as possible vertical schemes for HIV prevention and treatment interventions. For example, PMTCT programmes cannot be isolated from adequate antenatal clinic services, family planning, delivery facilities, and ambulatory services for chronic diseases of women and children. Indeed, the inclusion of male partners as an essential component in PMTCT-Plus indicates the broad sweep of interconnected services necessary. The frequent coinfections between HIV and tuberculosis are persuasive reasons for seeking complementarity between services for each. A system-wide response has the advantage that constraints addressed benefit a range of diseases, and draws attention to other health challenges that may be overlooked in the context of HIV/AIDS. Although the results of this approach may not be as quickly seen as in the disease-specific approach, it allows the system in its entirety to be strengthened.
It follows that the health system, rather than the specific disease, should be tackled in order to achieve the effective and holistic delivery of interventions. Such restructuring tends to be effective only in the long term, so immediate interventions may have to be introduced into the health system to deal with the pressing needs of prevention and of HIV/AIDS patients.
Robust health systems play a fundamental role in channelling globally recognised prevention and treatment best practice for the mitigation of HIV/AIDS. However, certain social and biological complexities profoundly affect the transmission, progression and mortality of the disease; these lie beyond the scope of health services. Intrinsically difficult to control, these elements constitute significant obstacles to the prevention and management of the HIV/AIDS epidemic. Biological factors, such as exposure to infected individuals (through sex, contaminated blood products, or perinatally), infectivity (determined by the viral load), and concomitant sexually transmitted infections (STIs) greatly increase susceptibility to infection. Social and environmental determinants, which include socio-economic status (for example, unemployment, poverty, degree of urbanisation and migration) may increase proclivity to risky behaviour (such as unprotected sex or drug use) and heighten the possibility of infection. Another important factor here is gender and age: women's lower status and adolescents' relative youth renders both groups more vulnerable to infection due in part to a consequent lack of power in relationships. [32-38].
It follows therefore that addressing health system constraints alone will not constitute a comprehensive solution to the management of the epidemic. Mitigation of risk factors needs to be an integral part of the response to HIV/AIDS in order for real progress to be made in the propitiation of the disease.
Conclusion
The expansion of the AIDS epidemic across the globe has galvanized the global community into demonstrating a willingness to challenge its unabated spread. The increasing mobilisation of resources aimed at mitigating the impact of the disease in developing regions of the world in particular holds numerous potential benefits on the course of the AIDS epidemic. Whether these benefits are realized or not depends on resources dedicated to addressing the global AIDS challenge being received by those in need.
The large volumes of aid being made available to developing countries has in many instances resulted in bottlenecks in health systems in these regions, which are historically unable to cope with the demands being place upon them by the accelerating spread of HIV and concomitant influx of resources to meet this challenge. Effective delivery of aid is thus hampered. This problem can be addressed through systemic strengthening of health systems in order to build capacity and sustainability, thereby redressing the inequities in healthcare delivery due to historical differences in health systems between and within rich and poor countries. However, other risk factors such as behaviour, socio-economics and biology also contribute to the spread of the disease. It follows that addressing both health systems and these external factors is necessary in order to manage and contain HIV comprehensively.
Another important factor in the management of the epidemic is the balance between prevention and treatment programmes. The present apparent emphasis on treatment to the detriment of prevention needs to be redressed in order to meet the challenges of the disease at all levels.
Globalisation brings with it many benefits in addressing the spread of HIV throughout the world. However, these benefits can only be realized if appropriate programmes are available in areas of need. As part of the generous supply of aid aimed at addressing problems specific to HIV/AIDS, attention needs to be paid to building capacity in recipient countries so that such funds may be effectively disseminated and the epidemic effectively curbed.
Authors' contributions
HC and JH contributed equally to the compilation of information and composition of the paper. Both authors read and approved the final manuscript
==== Refs
Stiglitz J Globalization and its Discontents 2003 New York: WWNorton
Barnett T Whiteside A AIDS in the Twenty-First Century: Disease and Globalization 2000 Basingstoke: Palgrave-Macmillan
Hussain TM HIV/AIDS in central Asia: The issue of a serious concern [Ref E10169] CD publication of XV International AIDS Conference: 11–16 July 2004; Bangkok
Cohen J Asia and Africa: On different trajectories? Science 2004 304 1932 1938 15218138 10.1126/science.304.5679.1932
Rogstad KE Sex, sun, sea, and STIs: sexually transmitted infections acquired on holiday British Medical Journal 2004 329 214 217 15271833
United Nations High Commission for Refugees Refugees By Numbers 2004 Geneva
International Labour Organisation Tripartite Forum on Labour Migration in Southern Africa 2002 Pretoria, South Africa
UNAIDS Report on the global AIDS epidemic – July 2004 2004 Geneva
United Nations Population Division The Impact of AIDS 2003 Geneva
De Waal A Whiteside A New variant famine: AIDS and food crisis in southern Africa The Lancet 2004 362 1234 1237 10.1016/S0140-6736(03)14548-5
Nattrass N The Moral Economy of AIDS in South Africa 2004 Cape Town: Cambridge University Press
Kazatchkine M AIDS: a vaccine to look after? Interview with V Olivier, Le Vif / L'Express, 1–7 October 2004
McNeill H Plagues and People 1976 Anchor Press
Travis P Bennett S Haines A Pang T Bhutta Z Hyder AA Pielemeier NR Mills A Evans T Overcoming health-systems constraints to achieve the Millennium Development Goals The Lancet 2001 364 900 906 10.1016/S0140-6736(04)16987-0
World Health Organization Treating 3 million by 2005: Making It Happen: The WHO Strategy 2003 Geneva
United Nations Declaration of Commitment on HIV/AIDS No. 55, unanimously adopted by the United Nations General Assembly Special Session on HIV/AIDS 2001
Global HIV Prevention Working Group HIV prevention in the era of expanded treatment access 2004
Avert.Org President Bush's Emergency Plan for AIDS Relief (PEPFAR) 2005
Hankins C Gutierrez JP Bertozzi S McGreevey W Bollinger L Greener R Stover J Resource needs for an expanded response to AIDS in low-and middle-income countries
Schwartländer B Stover J Walker N Bollinger L Gutierrez JP McGreevey W Opuni M Forsyth S Kumaranayake L Watts C Bertozzi S AIDS: Resource needs for HIV/AIDS Science 2001 292 2434 2436 11423619 10.1126/science.1062876
Jha P Nagelkerke JD Ngugi EN Public health: reducing HIV transmission in developing countries Science 2001 292 224 225 11305312 10.1126/science.1058187
Marseille E Hofmann PB Kahn GK HIV prevention before HAART in sub-Saharan Africa The Lancet 2002 359 1851 1856 12044394 10.1016/S0140-6736(02)08705-6
Sabin CA Lampe Fiona Matthias R Bhagani S Gilson R Youle MS Johnson MA Fisher M Scullard G Easterbrook P Gazzard B Phillips AN Treatment exhaustion of highly active antiretroviral therapy (HAART) among individuals infected with HIV in the United Kingdom: multicentre cohort study British Medical Journal 2005 330 695 15749728
Walker Neff UNICEF, Personal communication 15th September 2004
Kwazulu Natal Department of Health Implementing prevention of mother to child transmission of HIV (PMTCT) in Kwazulu Natal Unpublished data 2004
World Health Organization World Health Report 2004: changing history 2004 Geneva
Pielemeier N Bhatt P Smith O Wainberg MA et al Lessons from health system strengthening can lead to rapid scale-up of HIV programmes: Case studies from immunization, family planning, and primary health care [Ref WePeE6796] XV International AIDS Conference Abstracts 11–16 July 2004; Bangkok 2004 2 IAS/Medscape 272
Brugha R Donoghue M Starling M Ndubani P Ssengooba F Fernandes B Walt G Wainberg MA et al Tracking global fund policy processes in four African countries [Ref MoOrE1037] XV International AIDS Conference Abstracts 11–16 July 2004; Bangkok 2004 1 IAS/Medscape 228
Chandani Y Allers C Kinzett S Amenyah J Coming up with the numbers: Quantifying commodities to achieve 3 by 5. [Ref E12655] CD publication of XV International AIDS Conference: 11–16 July 2004; Bangkok
Nyaruhirira I Schocken CM Wainberg MA et al Using HIV/AIDS funding to improve health care services for all. [Ref TuPeE5315] XV International AIDS Conference Abstracts 11–16 July 2004; Bangkok 2004 1 IAS/Medscape 531
Toole MJ Harnessing the new global health resources to build sustainable health systems [Ref E10558] CD publication of XV International AIDS Conference: 11–16 July 2004; Bangkok
Tallis VA Cavanagh PD Ewing D Understanding women's vulnerability starts with addressing the realities of women and girls lives [Ref E12889] CD publication of XV International AIDS Conference: 11–16 July 2004; Bangkok
Narayanan R Narang A Dhar S Mathur A Mittal A Reducing women's vulnerability to HIV/AIDS: A multisectoral approach [Ref D11297] CD publication of XV International AIDS Conference: 11–16 July 2004; Bangkok
Cardoso MAC d'Oliviera AFP Couto MT Wainberg MA et al Prevention and vulnerability among women living with HIV/AIDS: stories of women who are partners of injectable drug users. [Ref TuPeE5420] XV International AIDS Conference Abstracts 11–16 July 2004; Bangkok 2004 1 IAS/Medscape 551
Wells VE Wainberg MA et al Empowering women reveals eleven essential steps for HIV/AIDS prevention and mitigation [Ref TuOrD1222] XV International AIDS Conference Abstracts 11–16 July 2004; Bangkok 2004 1 IAS/Medscape 454 455
Mehra S Goswami PK Negi RS Wainberg MA et al Integrating HIV/AIDS prevention and management in public health and education system [Ref MoPeE4240] XV International AIDS Conference Abstracts 11–16 July 2004; Bangkok 2004 1 IAS/Medscape 281
Samad A Wainberg MA et al Epidemiology and prevention of HIV amongst youth, adolescents and mobilized population [Ref TuPeC4740] XV International AIDS Conference Abstracts 11–16 July 2004; Bangkok 2004 1 IAS/Medscape 405
Naseem KVN Gender integration in HIV/AIDS and adolescents [Ref D10985] CD publication at the XV International AIDS Conference: 11–16 July 2004; Bangkok
UNAIDS AIDS epidemic update – December 1998 1998 Geneva
UNAIDS AIDS epidemic update December 2004 2004 Geneva
Dixon S McDonald S Roberts J AIDS and economic growth in Africa: a panel data analysis Journal of International Development 2001 13 411 426 10.1002/jid.795
Cornia GA Zagonari F The HIV /AIDS impact on the rural and urban economy AIDS, Public Policy and Child Well-Being 2002 UNICEF Innocenti Research Centre. Florence, Italy
|
16060961
|
PMC1199613
|
CC BY
|
2021-01-04 16:39:02
|
no
|
Global Health. 2005 Aug 1; 1:13
|
utf-8
|
Global Health
| 2,005 |
10.1186/1744-8603-1-13
|
oa_comm
|
==== Front
Health Qual Life OutcomesHealth and Quality of Life Outcomes1477-7525BioMed Central London 1477-7525-3-461605096010.1186/1477-7525-3-46ResearchHealth measurement using the ICF: Test-retest reliability study of ICF codes and qualifiers in geriatric care Okochi Jiro [email protected] Sakiko [email protected] Tai [email protected] Department of Health Services Coordination, Graduate School of Medical Sciences. Kyushu University. Maedashi 3-1-1 Higashiku, Fukuoka, 812-8586, Japan2 Department of Health Service Management, International University of Health and Welfare, 2600-1 Kita-Kanamaru Ohtawara, Tochigi, 324-0011, Japan2005 29 7 2005 3 46 46 23 2 2005 29 7 2005 Copyright © 2005 Okochi et al; licensee BioMed Central Ltd.2005Okochi et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The International Classification of Functioning, Disability and Health (ICF) was published by the World Health Organization (WHO) to standardize descriptions of health and disability. Little is known about the reliability and clinical relevance of measurements using the ICF and its qualifiers. This study examines the test-retest reliability of ICF codes, and the rate of immeasurability in long-term care settings of the elderly to evaluate the clinical applicability of the ICF and its qualifiers, and the ICF checklist.
Methods
Reliability of 85 body function (BF) items and 152 activity and participation (AP) items of the ICF was studied using a test-retest procedure with a sample of 742 elderly persons from 59 institutional and at home care service centers. Test-retest reliability was estimated using the weighted kappa statistic. The clinical relevance of the ICF was estimated by calculating immeasurability rate. The effect of the measurement settings and evaluators' experience was analyzed by stratification of these variables. The properties of each item were evaluated using both the kappa statistic and immeasurability rate to assess the clinical applicability of WHO's ICF checklist in the elderly care setting.
Results
The median of the weighted kappa statistics of 85 BF and 152 AP items were 0.46 and 0.55 respectively. The reproducibility statistics improved when the measurements were performed by experienced evaluators. Some chapters such as genitourinary and reproductive functions in the BF domain and major life area in the AP domain contained more items with lower test-retest reliability measures and rated as immeasurable than in the other chapters. Some items in the ICF checklist were rated as unreliable and immeasurable.
Conclusion
The reliability of the ICF codes when measured with the current ICF qualifiers is relatively low. The result in increase in reliability according to evaluators' experience suggests proper education will have positive effects to raise the reliability. The ICF checklist contains some items that are difficult to be applied in the geriatric care settings. The improvements should be achieved by selecting the most relevant items for each measurement and by developing appropriate qualifiers for each code according to the interest of the users.
==== Body
Background
The International Classification of Functioning, Disability and Health (ICF) was published by the World Health Organization (WHO) in 2001 to standardize descriptions of health and disability[1]. Not only has the ICF provided a conceptual framework for description of functioning and disability, health professionals can use it as a tool to describe necessary information concerning people with disabilities[2]. Whence the standardization of the language is achieved with the ICF, areas of potential application include description of disability cases[3]; standardization of clinical recording systems, and comparison of disability statistics between countries[4]. As WHO has provided the ICF with its qualifier, existing health measures can be mapped to the ICF [5,6]. It may also be possible to develop measurement scales from the ICF codes[7]. The ICF consists of four domains: body structures, body functions (BF), activities and participation (AP), and environment. The term "disability" is further defined as "impairment" (dysfunction or loss of "body functions or structure"), "limitation" (the difficulty an individual may experience in executing a particular activity) or "restriction" (problems an individual may experience in involvement in life situations). Every domain of the ICF has hierarchical structure, with increasing code values (higher digit items) corresponding to more specific functions or activities. These are the characteristics of the ICF as taxonomy. In addition, WHO applied the ICF to describe the level of disability. For this purpose, WHO developed the qualifiers relevant to each domain, and they were added to the ICF codes. For example, using the Performance qualifier, mild restriction of "d4500 walking short distance" is coded d4500.1. According to WHO, the ICF code without qualifier does not have an inherent meaning when used for individuals or cases[8]; thus the qualifier is indispensable to denote the level of health.
The ICF in its current version consists of 1424 codes. Therefore, it is necessary to select a subset of the codes as needed for any given purpose. One of such activities is the development of the ICF checklist, which is composed of major three digit ICF items, as a practical tool to elicit and record information about an individual's functioning and disability[9,10]. Other such studies involve the development of the ICF core-sets[11]. They are developed to standardize what to measure for each chronic condition.
In addition to these studies, which are aimed at determining what to measure, it is necessary to consider how to describe health and its related status using the ICF codes. One possible approach, as taken in this study, is to apply as many ICF items as possible, as measures to describe health conditions, then to discuss the reliability and applicability of the ICF codes in a health domain such as geriatric care.
In Japan, after the implementation of the long-term care insurance (LTCI) law in 2000[12], accurate assessment of the needs of elderly clients using LTCI services became necessary. In addition, the Ministry of Health, Labor and Welfare today recommends the use of the ICF in rehabilitation care planning. Therefore, it is imperative to assess the accuracy of the application of the ICF codes in geriatric care and rehabilitation settings.
This study examines test-retest reproducibility and the clinical relevance of the ICF codes related to the geriatric care setting in the context of this background. It also aims at evaluating the content validity of the ICF checklist. The research targets the BF and AP domains only, since they contain more easily measurable items and also it was reasonable to lessen the burden of the evaluators.
Methods
788 elderly patients (age> = 65 years) using LTCI services were selected from 5 hospitals, 29 long-term care institutions, 11 day-care centers, and 14 visiting nursing service centers. Candidates were selected without regard to age, gender, or level of function. However the participants were selected on the basis of functional stability during the one week test-retest period, and ability to give informed consent to study participation. Two independent evaluators judged the stability of candidates. Therefore, a randomization approach was not used to select study participants. Written consent was obtained from all study participants except when it was obtained from a family member by proxy in cases where the subject was unable to provide written consent by him/herself. Subjects who exhibited an acute decline in function during the course of the study were excluded from the final analysis.
Between May and October 2003, each subject was independently evaluated for numerous ICF codes by two health care professionals. The two evaluations were performed within a week of each other. In addition, all evaluators concurrently administered the Typology of the Aged with Illustrations (TAI) questionnaire, a simple illustrative assessment tool developed for care-management of the long term care insurance to assess the reproducibility of ICF items[13]. The TAI is a four-scale instrument whose reliability and validity have previously been established [14,15]. Evaluators also documented subjects' chronic medical conditions, health behaviours, and living status. Each evaluator also reported his or her own professional background and years of work experience on the same questionnaire. All evaluators were provided with a comprehensive guides to the ICF codes and qualifiers and to the TAI in advance of the assessment.
Since the ICF checklist is composed of three-digit code items (31 items in BF and 48 items in AP domain), every three digit item in the BF and AP domain was initially selected. For more detailed analysis, additional four-digit items were included, by the consensus recommendations of a panel of physiotherapists, occupational therapists, speech therapists, nurses, social workers and care-managers consulted by the authors. 85 BF items (79 three-digit and 6 four-digit) and 152 AP items (81 three-digit and 71 four-digit) ultimately composed the study instrument.
Eight BF domain chapters included were: 1. "Mental Functions" (21 codes); 2. "Sensory Functions and Pain" (15 codes); 3. "Voice and Speech Functions"(4 codes); 4. "Cardiovascular, Haematological, Immunological and Respiratory Systems" (10 codes); 5. "Function of the Digestive, Metabolic and Endocrine Systems" (10 codes); 6. "Genitourinary and Reproductive Functions" (7 codes); 7. "Neuromusculoskeltal and Movement-Related Functions" (12 codes); 8. "Functions of the Skin and Related Structures" (6 codes).
Nine AP domain chapters included were: 1. "Learning and Applying Knowledge" (15 codes); 2. "General Tasks and Demands" (4 codes); 3. "Communication" (11 codes); 4. "Mobility" (64 codes); 5. "Self-care" (20 codes); 6. "Domestic Life" (14 codes); 7. "Interpersonal Interactions and Relationships" (7 codes); 8. "Major Life Areas" (12 codes) and 9. "Community, Social and Civic Life" (5 codes). Some of the three-digit ICF codes not applicable to Japanese geriatric population, such as "riding animals for transportation" (d460) were nevertheless intentionally included in order to test whether they were correctly identified as irrelevant items.
In response to concerns raised by evaluators, illustrations to each item on the study questionnaire were added to promote efficient comprehension of each ICF item[16]. These illustrations are available on the authors' website [17].
The body functions qualifier is used in BF measurement in this study. Two kinds of qualifiers were used in AP measurement (the performance qualifier and the capacity qualifier). The performance qualifier describes what an individual actually performs in his or her current environment, while the capacity qualifier describes an individual's ability to execute a task or an action. In this study, the performance qualifier was used to evaluate the AP limitation or restriction. According to the WHO definition, the qualifiers were graded as follows: Level 0 indicates "no problem" (0–4% limitation or restriction); Level 1 "mild problem" (5–24% limitation or restriction); Level 2 "moderate problem" (25–49% limitation or restriction); Level 3 "severe problem" (50–95% limitation or restriction) and Level 4 "complete problem" (96–100% limitation or restriction). Levels 8 were used to describe conditions "not specific" meaning the available information does not suffice to quantify the severity of the problem.
Level 9 were used to describe conditions that were "not applicable" For example, the category d760 (Family relationships) is not applicable to a patient with no living family members.
Figure 1 shows the format of the questionnaire used in this study. All the description used in the questionnaire was identical to the WHO publication on the ICF.
Figure 1 Questionnaire used in this study.
The evaluators were given instructions by the authors, using a manual comparable to the ICF checklist [9]. The evaluators are asked to evaluate using all possible information available, including interviews, proxy, and medical records.
To evaluate the reliability of each ICF item, the weighted kappa of each item was calculated using the data obtained by the two independent evaluators [18]. To estimate the reliability between the two, the required number of pairs is 86 with alpha and beta error level of 0.05 and 0.20 respectively, with a minimum required kappa level of 0.4 and acceptable kappa level of 0.6 [19]. Stratification was done so that there were more than 86 relevant data both in terms of measurement settings and evaluators' experience. In this study, the weighted kappa was classified according to Landis et al to moderate (0.41–0.60), substantial (0.61 to 0.8), and almost perfect agreement (above 0.8) [20].
The kappa value was further evaluated by stratifying the experience of the evaluators and care settings. In evaluation by box plot, the kappa statistics that showed negative values were replaced to 0.
As an index of irrelevance of the ICF codes, the immeasurability rate of each item was calculated using the sum of samples judged "non specific" or "not applicable" as the numerator and the total number of evaluations as the denominator. The properties of each item were evaluated by using the values of the weighted kappa and the immeasurability rate of each item. The analyses were performed with STATA (version 8.17).
Results
Evaluations were performed on a total of 788 participants. Among these evaluations, one of two evaluations was not complete in 46 persons (6%). These cases were excluded from the analysis. Thus, two sets of data were independently obtained from 742 geriatric subjects yielding a total of 1484 data sets. 25% of subjects were male (mean age 78.8 years; SD = 9.2 years) and 75% were female (mean age 84.1; SD = 7.6 years); 593 were institutionalized (evaluated at the residential institution) and 149 lived at home (75 evaluated at day-care services and 74 evaluated at home).
289 experienced care professionals served as evaluators: nurses (24%), therapists (26%), care managers (22%) and social workers/caregivers (28%). The average amount of work experience as health professional was 10 years (SD8) with a median of 8 years. Among the measurements, 227 pairs were performed by evaluators who both had 8 years or more experience. Conversely, 205 pair of measurements were performed by evaluators who both had less than 8 years of experience.
Test-retest reliability of the ICF items
The result distribution and rating as non specific (n.s), not applicable (n.a), and weighted kappa of 85 BF items (79 three-digit items and 6 four-digit items) and 152 AP items (81 three-digit items and 71 four-digit items) are shown in Additional files 1 (BF domain) and 2 (AP domain).
Weighted kappa values of BF domain items ranged from 0.13 to 0.72, with an average of 0.46 and a median of 0.44, while that of AP domain items ranged from -0.17 to 0.79, with an average of 0.55 and a median of 0.59.
Table 1 shows the weighted kappa for the BF and AP domains stratified by evaluators' years of experience, and by care settings of the samples, along with the referential weighted kappa value of the TAI scales. The institutionalized care setting showed higher average and median kappa values, compared to the setting of living at home. Average and median of the evaluation performed by a more experienced pair of evaluators exceeded those performed by a less experienced pair. The kappa values of the four TAI scales concurrently measured with the ICF items were as follows: "Mobility" 0.80 (95% C.I. 0.75–0.84); "Mental function" 0.75 (0.70–0.80); "Toileting" 0.76 (0.71–0.82); "Eating" 0.78 (0.73–0.83). The weighted kappa value of the TAI scales did not show marked differences between care settings and evaluators' experience.
Table 1 Average and median weighted kappa values, by care settings and evaluator experience
care setting evaluators experience
Total institutional At home <8 years > = 8 years
BF domain (84 items)
Average 0.46 0.47 0.37 0.41 0.58
SD 0.12 0.12 0.14 0.15 0.10
Median 0.44 0.44 0.35 0.41 0.59
AP domain (137 items)
Average 0.58 0.58 0.51 0.54 0.63
SD 0.09 0.10 0.11 0.13 0.11
Median 0.59 0.59 0.51 0.56 0.64
TAI*
Mobility 0.80 0.80 0.76 0.78 0.83
Mental function 0.75 0.75 0.70 0.81 0.77
Eating 0.76 0.76 0.77 0.82 0.79
Toileting 0.78 0.77 0.79 0.79 0.82
*Typology of the aged with illustrations
The higher average kappa value of measurement in institution is not likely due to the experience of the evaluator, since the measurement performed at home contained more pairs of evaluation by experienced evaluators (69%).
Figure 2 shows the box plot of weighted kappa statistics by chapters of the BF and AP domains respectively. Chapter 2,4,5,6 and 7 in the BF domain and Chapters 8 and 9 in the AP domain showed relatively low reliability. In the BF domain the weighted kappa result by the pair of experienced evaluators showed better measurement reproducibility for all chapters. In AP the domain, experienced evaluators showed better reproducibility in chapters 1,5,6,7,8 and 9,
Figure 2 Box plot of weighted kappa of the ICF BF domain by chapter.
Figure 3 shows the box plot of weighted kappa stratified by care setting. Most of the chapters, except for the chapter 8 and 9 of the AP domain, showed higher reliability. However, caution must be paid when interpreting the weighted kappa result in chapter 8, because kappa values were far less accurate when the results were stratified.
Figure 3 Box plot of weighted kappa of the ICF AP domain by chapter.
Immeasurability rate
The immeasurability rate of BF domain items ranged from 0.00 to 0.96, with an average of 0.06 and a median of 0.03. That of the AP domain items ranged from 0.00 to 0.90, with an average of 0.13 and a median of 0.02.
Figure 4 shows the box plot of the immeasurability stratified by the evaluators' years of experience. In the BF domain, the highest immeasurability item was "sexual functions" (b640): 0.96. Because this is an exceptionally large figure, it was not plotted on Figure 4. Other top five items rated as immeasurable within the BF domain were "menstruation functions" (b650): 0.28, "sensations associated with genital and reproductive functions" (b670): 0.26, "endocrine gland functions" (b555): 0.18, "procreation functions" (b660): 0.18. Except for the item "endocrine gland functions" (b555), all fell within chapter 6 of BF domain, "Genitourinary and Reproductive Functions". The top 5 items rated as immeasurable in AP domain were "preschool education" (d815): 0.90; "school education" (d820): 0.90; "higher education" (d830): 0.89; "producing messages in formal sign language" (d340): 0.89; and "vocational training" (d825): 0.89. Except for the item "communicating in formal sign language" (d340), all fell within chapter 8 of the ICF AP domain, "Major Life Areas". The pattern of immeasurability by chapter did not differ according to the evaluators' experience.
Figure 4 Box plot of immeasurability rate of the ICF BF domain by chapter.
Figure 5 shows the immeasurability rate by care settings of elderly persons. Except for the immeasurability rate of BF domain chapter 6 (domestic life) which showed a lower immeasurability rate compared to the elderly in the institutional settings, there were no marked differences in immeasurability rate by care setting.
Figure 5 Box plot of immeasurability rate of the ICF AP domain by chapter.
Properties of the ICF items
The weighted kappa statistics and the immeasurability rate of three-digit AP and BF domain codes are categorized as shown in Additional file 3 and 4 respectively. Items were classified into 3 categories: high reliability (weighted kappa ≥ 0.6); intermediate reliability (0.4 ≤ weighted kappa<0.6); and low reliability (weighted kappa < 0.4) using all data. Items were secondly categorized by the median value of the immeasurability rate (immeasurability ≥ 0.3 for the BF domain: immeasurability ≥ 0.2 for the AP domain, the median score) or of low requirement (immeasurability <0.3 for the AP domain: immeasurability rate < 0.2 for the AP domain). Additionally, each item was flagged as to whether it was included in the ICF checklist.
High reliability and measurable items included in the study instrument, but not found in the ICF Checklist were: "global psychosocial functions" (b122); "temperament and personality function"(b126); "calculation functions"(b172); "mental function of sequencing complex movements" (b176); "articulation functions"(b320) and "gait pattern functions"(b770) in BF domain, and "focusing attention" (d160); "making decisions" (d177); "transferring oneself" (d420); "Moving around in different location" (d460) in the AP domain. On the contrary, items evaluated as low reliability and immeasurable in the ICF Checklist were: "blood pressure functions"(b420); "Haematological system functions"(b430); "immunological system functions"(b435); "respiration functions"(b440)"; digestive functions" (b515); endocrine gland functions(b555) and "sexual functions" (b640) in BF domain, and "school education" (d820); "apprenticeship"(d840); "religion and spirituality" (d930) and "human rights"(d940) in the AP domain.
Discussion
The clinical application of ICF codes to diverse populations remains an active topic of discussion [21-27], with little consensus as to how each code and qualifier must be utilized for specific populations. There are related previous studies, which deal with the concept of ICF model using different exisiting scales [28-30]. Some studies dealt with the ICF reproducibility to assign ICF categories to extant measures[3]. In geriatric care research, Jette et al. have identified distinct concepts shared by activity and participation[31].
However, still to date, to the best of the authors' knowledge, there is no study that has shown the test-retest reproducibility of the ICF as a scale to evaluate functioning in a specific population.
The ICF is based on a universal model that theoretically can be applied regardless of cultures, age groups or care settings [7,27,32]. However, various codes may have different implications for various care settings in practical terms, and individual ICF items requires validity and reliability studies in application to diverse populations. Such efforts are already underway in the form of development of ICF core-sets for specific medical conditions[11]. Conceptual applications of the ICF to National surveys have also been undertaken[4,33,34].
This study differs from both these approaches, as it does not rely on the experts' opinions to assure face and content validity, but applies the ICF directly as an instrument of geriatric assessment to select more adequate items, while aiming to develop new scales using ICF taxonomy.
It requires a certain level of test-retest reproducibility and measurability, or discard of items which are not appropriate to create new scales.
The authors are now developing the elderly communication performance scale according to the result of test-retest reliability statistics, because AP items related to communications have acceptable level of test-retest reliability.
Items such as d320 and d340, which are related to communication using formal sign language, showed low measurability. These items are not always applicable in the general geriatric care setting, but are pertinent for individuals with hearing loss. Thus, the scale developer can select ICF items with certain reliability and measurability according to the scope of each scale.
The other rationale of testing such a wide range of the ICF codes is that elderly persons hold problems that cover multiple disciplines. This contrasts with the ICF core-set project which is relatively disease focused.
Reliability of the ICF qualifiers
Our findings raise concerns about the low reliability of the ICF items using qualifiers.
Although overall reliability of the ICF items was low, it had improved considerably, when the weighted kappa statistics were stratified by the work experience of the evaluators. As shown in Table 1, the weighted kappa of the TAI scales did not show marked differences compared to the ICF items. A previous study on the TAI scales also indicated that the reliability was not dependent on the experience of the evaluators [15]. It indicates the ICF items and its qualifiers may be too difficult to quantify in some cases.
By stratifying the results by care-settings, it was possible to get better test-retest reproducibility in the institutional setting. This may be because more information, including medical records, are available in the setting.
The result of reliability differs depending on the chapter. As shown in Figures 2 and 3, the low weighted kappa value of chapters 4, 5 and 8 of the BF domain, and chapters 8 and 9 of the AP domain contribute to the overall low reliability of the ICF.
In BF domain, chapter 4("Functions of the Cardiovascular, Haematological, Immunological and Respiratory Systems"), 5 ("Functions of the Digestive, Metabolic and Endocrine Systems") and 8 (Functions of the Skin and Related Structures") are composed of items that can be described with specific medical examination.
For example, "blood pressure functions" (b420) can be described much more easily with blood pressure level measurable with arm cuff than using qualifier levels from 0 to 4.
Immeasurability of the ICF items
What we call immeasurable in this study include level 8 – not specified (available information does not suffice to quantify the severity of the problem) and level 9 – not applicable (e.g., d760, Family relationships is not applicable to an elderly person without family).
For example, in case of the global psychosocial functions (b122:immeasurability rate 2.6%), 38 evaluators could not quantify it because the sufficient information was not available and one evaluator rated it as not applicable as shown in additional file 1. This indicates that items with low immeasurability rate can be easily evaluated.
In contrast, 96% of the measurement was rated as immeasurable in sexual function (b640), and most of them were rated as not applicable, as expected by the target sample of this study. Overall, most of the rating as immeasurable was by level 8 (not specific), although some items such as chapter 8 ("Major life area") of the AP domain showed more level 9 than level 8.
Chapter 8 ("Major life area") is comprised of the categories "education" (d810-d839), "work and employment" (d840-d859), and "economic life" (d860-d879), while Chapter 9 (Community, social and civic life) includes "community life" (d910), "recreation and leisure" (d920), "religion and spirituality" (d930), "human rights" (d940) and "political life and citizenship" (d950). To accurately assign scores in the sub-domains of education, work and employment, community life, and political life in a population of institutionalized elderly patients may be difficult, or even inappropriate. Thus, the large proportion of institutionalized geriatric patients in our study sample may have affected the high immeasurability scores in these two chapters. The measurement of "religion and spirituality" and "human rights" requires multidimensional and subjective assessment. Thus it is difficult to assign either of them into a single code[35,36].
The low reliability shown in this study indicates the difficulty of using the ICF as a measurement tool and is also attributable to the ambiguous nature of the qualifiers. For example, when an evaluator judges the performance level of school education, he or she may assess the subject as level 4 ("complete difficulty"), because of the subject's inability to obtain further education or to attend an institution for learning. However, this item may also be regarded as "not applicable" or "not specified," especially in the context of institutionalized geriatric patient for whom school attendance is not an expected component of daily life.
In contrast, frequently assessed items in the LTCI assessment appeared to have high reliability. Presumably because items such as toileting and self-dressing constitute a part of a standard self-care assessment already widely used by healthcare professionals [37]. This similarity may explain the high reproducibility of self-care item assessments between independent evaluators in our study.
Validity of the ICF Checklist
An additional purpose of this study was to evaluate the validity of the ICF Checklist in geriatric assessment. We have also used the checklist as a training tool for evaluators, because it was the sole available material at the commencement of this study for official training of the ICF. We have found that the existing ICF Checklist lacks several items which we found scored high in reliability and low in immeasurability rate. These items include, "global psychosocial functions" (b122); "temperament and personality function" (b126); "calculation functions" (b172); "mental function of sequencing complex movements" (b176); "articulation functions" (b320) and "gait pattern functions" (b770) in the BF domain, and "focusing attention" (d160); "making decisions" (d177); "transferring oneself" (d420); "Moving around in different locations" (d460) in the AP domain.
The ICF checklist includes less reliable and immeasurable items, e.g. "blood pressure functions" (b420); "haematological system functions" (b430); "immunological system functions" (b435); "respiration functions" (b440)"; digestive functions" (b515); "endocrine gland functions" (b555) and "sexual functions" (b640) in the BF domain, and "school education" (d820); "apprenticeship" (d840); "religion and spirituality" (d930) and "human rights" (d940) in the AP domain. Some of the body function related items could be better described with chronic disease, such as high blood pressure, anemia, and diabetes. Items not relevant to the elderly care settings such as school education; apprenticeship might be just omitted when applying the scheme to those settings.
Importance of participation in religions and spirituality might vary depending on cultural settings. Also, human rights (d940) may play a pivotal role on understanding geriatric domestic violence.
This result should help selecting more useful sets of the ICF items that would reflect evaluators' needs and reliability of items. Some modification to the ICF checklist may also facilitate the use of the ICF.
Study Limitations
There are a few limitations in this study. The samples were selected from various service providers based on the stability of the function during the test-retest period. The kappa statistic is dependent on the samples. Therefore these samples might not fully represent the target population, namely the elderly using long-term care services in Japan. However, the use of a large sample obtained from multiple centers is nevertheless indicative of relatively low reliability of the ICF items measured with the qualifiers.
Also, other possible confounders such as the cultural settings and evaluators' professional backgrounds may influence the ICF measurement values. It is possible that some of the ICF items show different item functioning (DIF) depending on these confounders. The Rasch measurement technique is applicable to answer this question, which remains to be studied[38]. The illustrations added by the authors to clarify the definition of each item could have biased the results. However, our intention in incorporating illustrations was to standardize evaluator assessments. Previous studies have shown that illustrations increase the reliability of assessment instruments[39].
Lastly, the authors used the sum of qualifiers 8 and 9 as a simple index of immeasurability. Items with a high prevalence of level 8 suggested that it was difficult for the evaluator to ask the question or obtain the information from the medical chart. In contrast, assignment of a qualifier of 9, which was more prevalent in chapter 8 of AP domain, suggested these items were not applicable. However, these two qualifiers may convey quite different information, and the study design made it difficult to compare the differences between these two qualifiers. In addition, it was difficult to analyze inter-rater reliability of qualifiers 8 and 9 because of the skewed distribution of the result between these qualifier levels. However, the prevalence of these qualifiers, as shown in Additional files, should help in selecting ICF items for future research.
Conclusion
The reliability of the ICF codes as measured with qualifiers is relatively low, and the ICF Checklist requires modification. Improvements should be achieved by selecting the most relevant items for each measurement and constructing appropriate qualifiers for each code according to the interest of users.
Authors' contributions
JO, SU and TT carried out the study design, data collection, statistical analysis and preparation of this manuscript.
Supplementary Material
Additional File 1
Weighted kappa statistics and immeasurability rate of the ICF BF items
Click here for file
Additional File 2
Weighted kappa statistics and immeasurability rate of the ICF AP items
Click here for file
Additional File 3
Property of the three-digit ICF BF items according to weighted kappa, immeasurability, and the ICF checklist
Click here for file
Additional File 4
Property of the three-digit ICF AP items according to weighted kappa, immeasurability, and the ICF checklist
Click here for file
Acknowledgements
We would like to appreciate the great support of Professor Takashi Hashimoto, Kyushu Rehabilitation College, Dr. Yasuko Arase, Fukuoka City Government, and all healthcare professionals who have cooperated in this study. This research is funded by Health Labor Science Research Grant;1-15-Seisaku-018.
==== Refs
World Health Organization. International classification of functioning, disability and health : ICF 2001 Geneva , World Health Organization
Stucki G Ewert T Cieza A Value and application of the ICF in rehabilitation medicine Disabil Rehabil 2002 24 932 938 12523361 10.1080/09638280210148594
Ogonowski J Kronk R Rice C Feldman H Inter-rater reliability in assigning ICF codes to children with disabilities Disabil Rehabil 2004 26 353 361 15204487 10.1080/09638280410001658658
Madden R Choi C Sykes C The ICF as a framework for national data: the introduction of ICF into Australian data dictionaries Disabil Rehabil 2003 25 676 682 12959342 10.1080/0963828031000137171
Perenboom RJ Chorus AM Measuring participation according to the International Classification of Functioning, Disability and Health (ICF) Disabil Rehabil 2003 25 577 587 12959331 10.1080/0963828031000137081
Brockow T Wohlfahrt K Hillert A Geyh S Weigl M Franke T Resch KL Cieza A Identifying the concepts contained in outcome measures of clinical trials on depressive disorders using the international classification of functioning, disability and health as a reference J Rehabil Med 2004 49 55 15370748 10.1080/16501960410015380
Simeonsson RJ Leonardi M Lollar D Bjorck-Akesson E Hollenweger J Martinuzzi A Applying the International Classification of Functioning, Disability and Health (ICF) to measure childhood disability Disabil Rehabil 2003 25 602 610 12959334 10.1080/0963828031000137117
Ustun B The international classification of functioning, disability and health-a common framework for describing health states
World Health Organization. The ICF checklist
Ewert T Fuessl M Cieza A Andersen C Chatterji S Kostanjsek N Stucki G Identification of the most common patient problems in patients with chronic conditions using the ICF checklist J Rehabil Med 2004 22 29 15370744 10.1080/16501960410015362
Stucki G Grimby G (Eds) ICF core sets for chronic conditions J Rehabil Med 2004 44 5 141 15370740 10.1080/16501960410022300
Matsuda S The health and social system for the aged in Japan Aging Clin Exp Res 2002 14 265 270 12462371
Takahashi T Okochi J Takamuku K Matsuda S The introduction of typology of the aged with illustrations. Casemix Quarterly 2001 3 3 14
Okochi J Increase of mild disability in Japanese elders: A seven year follow-up cohort study BMC Public Health 2005 5 55 15924625 10.1186/1471-2458-5-55
Okochi J Takahashi T Takamuku K Matsuda S Takagi Y Reliability of a geriatric assessment instrument with illustrations Geriatr Gerontol Int 2005 5 37 47 10.1111/j.1447-0594.2005.00268.x
Sutch S ICF illustration library Bulletin of the World Health Organization 2004 82 550 551
Takahashi T ICF illustration library
Fleiss J Levine B Cho Paik M Statistical Method for Rates and Proportion. 2003 New Jersey , Wiley and Sons,Inc
Walter SD Eliasziw M Donner A Sample size and optimal designs for reliability studies Stat Med 1998 17 101 110 9463853 10.1002/(SICI)1097-0258(19980115)17:1<101::AID-SIM727>3.3.CO;2-5
Landis JR Koch GG The measurement of observer agreement for categorical data Biometrics 1977 33 159 174 843571
von Wild KR New development of functional neurorehabilitation in neurosurgery Acta Neurochir Suppl 2003 87 43 47 14518522
Ustun TB Chatterji S Kostansjek N Bickenbach J WHO's ICF and functional status information in health records Health Care Financ Rev 2003 24 77 88 12894636
Ustun TB Chatterji S Bickenbach J Kostanjsek N Schneider M The International Classification of Functioning, Disability and Health: a new tool for understanding disability and health Disabil Rehabil 2003 25 565 571 12959329 10.1080/0963828031000137063
Stucki G Cieza A Ewert T Kostanjsek N Chatterji S Ustun TB Application of the International Classification of Functioning, Disability and Health (ICF) in clinical practice Disabil Rehabil 2002 24 281 282 12004974 10.1080/09638280110105222
Steiner WA Ryser L Huber E Uebelhart D Aeschlimann A Stucki G Use of the ICF model as a clinical problem-solving tool in physical therapy and rehabilitation medicine Phys Ther 2002 82 1098 1107 12405874
Heerkens Y van der Brug Y Napel HT van Ravensberg D Past and future use of the ICF (former ICIDH) by nursing and allied health professionals Disabil Rehabil 2003 25 620 627 12959336 10.1080/0963828031000137135
Battaglia M Russo E Bolla A Chiusso A Bertelli S Pellegri A Borri G Martinuzzi A International Classification of Functioning, Disability and Health in a cohort of children with cognitive, motor, and complex disabilities Dev Med Child Neurol 2004 46 98 106 14974634 10.1017/S0012162204000209
Beckung E Hagberg G Neuroimpairments, activity limitations, and participation restrictions in children with cerebral palsy Dev Med Child Neurol 2002 44 309 316 12033716 10.1017/S0012162201002134
Noreau L Desrosiers J Robichaud L Fougeyrollas P Rochette A Viscogliosi C Measuring social participation: reliability of the LIFE-H in older adults with disabilities Disabil Rehabil 2004 26 346 352 15204486 10.1080/09638280410001658649
Rusch M Nixon S Schilder A Braitstein P Chan K Hogg R Impairments, activity limitations and participation restrictions: Prevalence and associations among persons living with HIV/AIDS in British Columbia Health and Quality of Life Outcomes 2004 2 46 15350202 10.1186/1477-7525-2-46
Jette AM Haley SM Kooyoomjian JT Are the ICF Activity and Participation dimensions distinct? J Rehabil Med 2003 35 145 149 12809198 10.1080/16501970310010501
Imrie R Demystifying disability: a review of the International Classification of Functioning, Disability and Health Sociol Health Illn 2004 26 287 305 15043600 10.1111/j.1467-9566.2004.00391.x
McDougall J Miller LT Measuring chronic health condition and disability as distinct concepts in national surveys of school-aged children in Canada: a comprehensive review with recommendations based on the ICD-10 and ICF Disabil Rehabil 2003 25 922 939 12857580 10.1080/0963828031000122249
Roussel P Barral C Reference to ICIDH in French surveys on disability Disabil Rehabil 2003 25 659 664 12959340 10.1080/0963828031000137153
Bussing A Ostermann T Matthiessen PF Role of religion and spirituality in medical patients: Confirmatory results with the SpREUK questionnaire Health Qual Life Outcomes 2005 3 10 15705195 10.1186/1477-7525-3-10
Meeks-Sjostrom D A Comparison of Three Measures of Elder Abuse. Journal of Nursing Scholarship 2004 36 Blackwell Publishing Limited 247 250 15495494 10.1111/j.1547-5069.2004.04045.x
Tsutsui T Muramatsu N Care-Needs Certification in the Long-Term Care Insurance System of Japan Journal of the American Geriatrics Society 2005 53 522 527 15743300 10.1111/j.1532-5415.2005.53175.x
Kucukdeveci AA Sahin H Ataman S Griffiths B Tennant A Issues in cross-cultural validity: example from the adaptation, reliability, and validity testing of a Turkish version of the Stanford Health Assessment Questionnaire Arthritis Rheum 2004 51 14 19 14872450 10.1002/art.20091
Hadorn DC Hays RD Uebersax J Hauber T Improving task comprehension in the measurement of health state preferences : A trial of informational cartoon figures and a paired-comparison task Journal of Clinical Epidemiology 1992 45 233 243 1569420 10.1016/0895-4356(92)90083-Y
|
16050960
|
PMC1199614
|
CC BY
|
2021-01-04 16:38:15
|
no
|
Health Qual Life Outcomes. 2005 Jul 29; 3:46
|
utf-8
|
Health Qual Life Outcomes
| 2,005 |
10.1186/1477-7525-3-46
|
oa_comm
|
==== Front
Hum Resour HealthHuman Resources for Health1478-4491BioMed Central London 1478-4491-3-51605351910.1186/1478-4491-3-5ResearchHuman resources for emergency obstetric care in northern Tanzania: distribution of quantity or quality? Olsen Øystein Evjen [email protected] Sidney [email protected] Ole Frithjof [email protected] Center for International Health, University of Bergen, Norway. For correspondence: DBL – Institute for Health Research and Development and Primary Health Care Institute, Iringa, Tanzania; P.Box 105297, Dar Es Salaam, Tanzania2 Center for Educational Development and Health (CEDHA), Arusha, Tanzania3 Section for Medical Ethics and Philosophy of Science, Department of Public Health and Primary Care, University of Bergen, Norway, and Center for International Health, University of Bergen, Norway2005 29 7 2005 3 5 5 15 6 2004 29 7 2005 Copyright © 2005 Olsen et al; licensee BioMed Central Ltd.2005Olsen et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Health care agencies report that the major limiting factor for implementing effective health policies and reforms worldwide is a lack of qualified human resources. Although many agencies have adopted policy development and clinical practice guidelines, the human resources necessary to carry out these policies towards actual reform are not yet in place.
Objectives
The goal of this article is to evaluate the current status of human resources quality, availability and distribution in Northern Tanzania in order to provide emergency obstetric care services to specific districts in this area. The article also discusses the usefulness of distribution indicators for describing equity in the decision-making process.
Methods
We conducted a quantitative facility survey in six districts of Northern Tanzania. We collected data from all 129 facilities that provide delivery services in the study area. The data includes information on the emergency obstetric care indicators, as described by the WHO/UNICEF/UFPA guidelines for monitoring the provision of obstetric care. The inventory also includes information on the numbers of qualified health personnel at the basic and comprehensive emergency obstetric care level. We analysed the distribution and workload of the available human resources in a wider policy context with a particular focus on equity, use and quality, by means of descriptive statistics and the Spearman's correlation test.
Results
We determined that there are adequate human resources allocated for health care provision in Tanzania, according to national standards. Compared to similar countries however, Tanzania has a very low availability of health care staff. Most qualified staff are concentrated in a few centralized locations, while those remaining are inequitably and inefficiently distributed in rural areas and lower-level services. Rural districts have restricted access to government-run health care, because these facilities are understaffed. In fact, voluntary agency facilities in these districts have more staff than the government facilities. There is a statistical correlation between availability of qualified human resources and use of services, but the availability of qualified human resources does not automatically translate into higher availability of qualified emergency obstetric care services.
Conclusion
National guidelines for human resources for health care in Tanzania require focused revisions in order to reflect the quality indicators more adequately when monitoring and setting criteria for HR distribution. Availability of qualified personnel as well as institutional management and capacity determine the quality of emergency obstetric care services and personnel. The current wide distribution of staff of inadequate quality should be reconsidered. The use of distribution indicators alone is not useful to properly monitor equity. This article suggests increasing access to high-quality health care instead of distributing low-quality services widely.
==== Body
Background and introduction
Simms and co-workers [1] predict that the health care system infrastructure in developing countries such as in Africa will collapse in the very near future. Recent attention has focused on the capacity of health care systems to provide adequate and timely services [1]. The lack of qualified human resources for health care is a major limiting factor in implementing health policies and health reforms in the developing world [2]. One of the major challenges is securing the availability and effective use of qualified human resources (HR). Thus, it is important for health care research to provide those who make and implement health care decisions with up-to-date information regarding the status of current resource allocation, shortcomings and planning strategies.
As a tracer policy, this paper attempts to identify the HR available to implement reformed reproductive health policy in Tanzania. We have chosen to identify personnel who are able to implement policies for reducing maternal mortality by focusing on emergency obstetric care (EmOC) services.
De Brouwere and co-workers [3] point out that maternal mortality in itself is not a good indicator for assessing maternal health care programmes and maternal health. Rather, it is important to assess the unmet obstetric needs and demonstrate the relative importance of adequate health care provision. Therefore, monitoring maternal health now tends to use process measures rather than impact measures as an accepted proxy [4].
Bertrand and Tsui [4], Nirupam and Yuster [5] and WHO are among many researchers and organizations providing comprehensive efforts to select useful process indicators to evaluate reproductive health programs. This article is based on the framework proposed by UNICEF, WHO and UNFPA [6]. This framework asks six basic questions:
1. Are there enough facilities providing emergency obstetric services (EOC)?
2. Are they well distributed?
3. Are enough women using these facilities?
4. Are the right women (those with obstetric complications) using these facilities?
5. Are sufficient quantities of critical services being provided?
6. Is the quality of the services adequate?
The questions in this framework are used to describe equity of access to quality care in terms of availability of EmOC units, distribution of EmOC services and use and quality of these services. These guidelines also provide a tool for defining and monitoring the emergency obstetric care units
We and other researchers as well as government agencies are using the United Nations guidelines more and more to evaluate the availability of obstetric care of good quality. [7-15].
Securing human resources for maternal health services is a key component in achieving the Millennium Development Goals (MDGs) by 2015. Specifically, many governments and organizations recognize that HR have not been targeted sufficiently for previous global development initiatives such as the Sector Wide Approaches and Poverty Reduction Strategy programmes. This trend is changing, and important initiatives are being established to attract policy-makers to this challenge. Examples are the Africa Working Group on Human Resources (part of the Joint Learning Initiative) and the Health Systems Trust focus on HR in southern Africa. There is also a working group in Tanzania comprising representatives from different ministries and partnerships. Results from this working group are not yet available.
It is vital that policy-makers have access to evidence on key aspects of HR management. Diallo and co-workers [16] have provided an overview of potentially important and useful monitoring and evaluation tools. Indicators include the available HR stock, skill mix, working location and equity issues.
The goal of this article is to adapt this framework to determine HR quality, availability and distribution for providing emergency obstetric care services in Northern Tanzania. We analyse the distribution and workload of these available HR in the wider policy context of urban and rural settings, public or private ownership and facility levels in the health care pyramid, with a particular focus on equity, use and quality.
Methods
Setting
The study area included two districts in the Kilimanjaro Region and four districts in the Arusha Region (two of which now are in the newly formed Manyara Region) of Tanzania, with a total population of about 1.5 million. We selected these districts to reflect different stages of health sector reform implementation, urban and rural settings and public and private service mixtures. Within these regions we identified and surveyed all facilities (n = 129) providing delivery services at all levels of service (dispensary, health centre, first referral hospital, secondary referral hospital). These included government (G), voluntary agencies (VA) and private, for-profit (PFP) facilities.
Data collection
The study represented a combination of comprehensive facility survey and document review. We conducted the facility survey through the structured analysis of facility documents (health management information system – (HMIS)) with the aid of a facility manager interview. Additional policy documents were reviewed at the district and regional health authority levels.
We determined the EmOC status of each facility by asking the following questions during the interview with the facility managers:
Were the following services performed at least once during the last 3 months (Yes/No):
1. parenteral antibiotics?
2. parenteral oxytocics?
3. parenteral sedatives/anticonvulsants?
4. manual removal of placenta?
5. removal of retained products?
6. assisted vaginal delivery?
7. blood transfusion?
8. caesarean section?
If the facility had performed functions 1 through 6 listed on the questionnaire, we considered them as basic emergency obstetric care units (BEmOC). If they also performed tasks 7 and 8, we considered them to be comprehensive emergency obstetric care units (CEmOC).
The United Nations guidelines audit established the level of services provided at each facility using the provisional EmOC indicators. These indicators include basic emergency obstetric care (BEmOC) units per 500 000 people, the comprehensive emergency obstetric care (CEmOC) units per 500 000 people, the number of deliveries at the EmOC facilities, the number of complications handled at the EmOC facilities (met need indicator), the caesarean section rates and case fatality rates in the six districts. From facility and district documents we collected a wide range of information including deliveries, complications (both for workload analysis), infrastructure, equipment and staffing resources data.
In addition to listing the total number of employees and their distribution across cadres, the facility managers provided the number of employees capable of conducting procedures at the basic emergency obstetric care and comprehensive emergency obstetric care levels. A pragmatic expectation for this health care provider includes handling a normal delivery with only minor complications (BEmOC capable staff) and handling complicated deliveries, depending upon their ability to perform a caesarean section (CEmOC capable staff).
The study did not attempt to conduct a clinical capability audit of each staff member, but instead relied on the facility manager to impart the number of staff qualified at the time of the survey. Every participant willingly provided the requisite information. Additional information at the district and regional health authority levels included the district health plans, the district processing files (HMIS summary) and the annual regional reports. All activity information pertains to the year 2000. This project employed one competent field assistant in addition to the principal investigator, to help with the data collection process.
Analysis
We calculated the expected number of births using the population figures from the most recent population census, that of 2002 [17]. From the census we gathered crude birth rates (CBR) of 34.4/1000 in urban districts and 43.5/1000 in rural districts, and growth estimates of 4.0% in the Arusha Region, 1.4% in the Kilimanjaro Region and 3.8% in the Manyara Region [18].
From our own data we compiled and analysed HR availability according to different analytical contexts. These were total HR availability at the district level, availability at each level of health services (dispensary, health centre, first referral hospital and secondary referral hospital) and availability according to ownership of the facilities. The availability of HR in the study area was analysed in terms of workload and output measures by calculating the number of deliveries per qualified worker within the different analytical contexts.
This study uses a complete dataset that includes all facilities providing delivery services and therefore represents a comprehensive overview of all HR for health care in these districts. In this regard, and for the purposes of this article, we conducted descriptive statistical analyses. We used a non-parametric analysis between availability rankings for qualified HR per one unit population (500 000) and United Nations guidelines utilization indicators and quality indicators. The utilization indicators are met need and percentage deliveries in EmOC facilities and the quality indicators are the number of qualified BEmOC and CEmOC facilities per 500 000 people in the six districts (n = 6).
Specifically, we compared direct or inverse correlations between: 1) the availability of qualified HR (BEmOC and CEmOC) and services used by the delivering mothers, and 2) the availability of qualified HR and the quality of the facilities. We used a Spearman's correlation table to find the critical values for the Spearman's ranking coefficient (rs) at n = 6.
We used the statistical software package SPSS for Windows, version 11.0.0. Research clearance was obtained from relevant institutions in Tanzania and Norway.
Results and discussion
In all, our study sample encompassed the six Tanzanian districts and data from all 129 facilities providing delivery services, with 34 756 deliveries, for a quantitative analysis. A summary of the results using the United Nations guidelines process indicators audit is shown in Table 1. Detailed discussions of these results are published elsewhere [14,15].
Table 1 Overview of United Nations guidelines process indicators for the survey area
Minimum recommended level (United Nations guidelines) Total Moshi urban Hai Arusha urban Arumeru Hanang Mbulu
BEmOC per 500 000 4 1.6 3.6 0.0 1.9 1.0 2.6 2.3
CEmOC per 500 000 1 4.6 10.7 2.0 11.5 2.1 0.0 4.5
Percentage of facility deliveries in:
Not EmOC 19.0 4.1 38.6 36.2 14.3 7.6 10.5
BEmOC 2.2 2.7 0.0 4.3 0.2 8.9 0.8
CEmOC 15 34.1 124.1 6.4 112.5 3.7 0.0 42.4
Percentage of total expected complicated deliveriesa
EmOC (met need) 100 59.8 319.3 8.2 164.2 8.9 2.1 50.4
Not EmOC 21.4 0.1 30.5 84.1 9.1 2.7 5.8
Caesarean section rate
At CEmOC 5–15 4.6 23.2 0.9 14.0 0.7 c 3.6
Case fatality rateab 1 1.46 1.28 2.09 1.54 1.78 c 1.38
aAdjusted complications – As described in Plan 3 in the United Nations guidelines
bOnly at CEmOC facilities recording deaths.
cNo CEmOC facilities in the district
Comparing our results with the national standards
Table 2 shows that the aggregated average number of qualified health personnel available at both dispensary and health centre levels is adequate compared to the national standards. Tanzanian HR policy requires that for every 50 000 people there should be five dispensaries and one health centre. For each dispensary there should be two qualified staff and for each health centre, four qualified staff [19]. This means that for a population of 50 000 there should be 14 qualified BEmOC staff, the equivalent of 28 staff per 100 000 people. The HR policies intend for the dispensaries and health centres to be capable of performing BEmOC activities.
Table 2 Availability and workload of BEmOC and CEmOC qualified staff at facility level according to public-private mix, levels of services and urban rural contexts
Available qualified staff Average qualified staff per facility
Number of facilities BEmOC CEmOC BEmOC CEmOC Facility deliveries per qualified BEmOC staff (workload) Percentage CEmOC staff of BEmOC staff
Public-private mix
Private for-profit 11 50 5 4.5 0.45 6.6 10.0
Voluntary agencies 41 576 36 14.0 0.87 17.1 6.3
Government 73 597 37 8.2 0.51 41.1 6.2
Level of services
Dispensaries 93 174 0 1.9 (2)a 0.0 27.9 0.0
Health centres 18 131 0 7.3 (4)a 0.0 38.4 0.0
First referral hospitals 15 479 38 31.9 2.53 20.3 7.9
Secondary referral hospitals 3 439 40 146.3 13.3 34.5 9.1
Urban-rural context
Urban (2 districts) 20 639 54 16.0 1.35 31.5 8.5
Rural (4 districts) 109 584 24 1.3 0.06 25.0 4.1
Total 129 1223 78 9.5 0.60 27.2 6.1
a Tanzanian national guidelines target
Table 2 shows that the required staffing levels at dispensaries are indeed available in Tanzania, with the number of qualified staff being only marginally less than that of the national standard, and those at the health centre level being nearly twice that of the national requirements. This is a surprising finding, since we expected a severe deficit of qualified personnel at these facility levels [20]. The importance of this finding is twofold. First, it provides optimism, considering the critical need for qualified personnel to achieve the MDGs in time. Second, however, it spurs a deeper investigation into the adequacy of national requirements, the distribution of personnel with respect to equity, the context within which they work, the use of the services by the expectant mothers and the role of qualified personnel in institutional quality.
Describing the context
The selected study area represents four rural and two urban districts of Tanzania. The country has a reported maternal mortality rate (MMR) of 1100 deaths per 100 000 live births [21]. This places the country in a position of having among the six highest maternal mortality rates globally for the year 2000. The facilities in the study area are distributed between private for-profit, voluntary agencies and government services (Table 2). Most of the facilities are dispensaries (98), a few are secondary referral hospitals (3), and there are almost equal numbers of health centres and first referral hospitals (18 and 15). The number of facilities per district in rural districts is more than twice that of urban districts.
The majority of EmOC services are provided by voluntary agencies (private not-for-profit, often faith-based organizations) and government services. This is also reflected in the availability of the EmOC staff. The data show that voluntary agencies generally staff each facility with about twice as many qualified personnel as in government and private for-profit services (BemOC staff per facility are 14.0 to 8.2 and 4.5 for VA, G and PP, respectively, and CEmOC staff per facility are 0.87 to 0.51 and 0.45 for VA, G and PP, respectively). These figures confirm the findings of other audits in Tanzania [22].
The availability of qualified staff for each facility is much higher per district in urban areas compared to rural districts for both BEmOC (16.0 to 1.3) and CEmOC (1.35 to 0.06) staff. The same indicator shows that private for-profit facilities have a slightly higher availability per facility compared to voluntary agencies and government facilities (6.3% and 6.2%, respectively). There is little difference in the percentages of CEmOC personnel to the total EmOC personnel available at the facilities, except between the urban and rural districts (8.5% compared to 4.1%).
Workload differences among qualified staff
Table 2 includes an overview of the workload differences among ownership and service levels. It should be noted that the data represent aggregate numbers at facility level and do not parse out details regarding the numbers of deliveries within facilities conducted by qualified or unqualified staff members. Table 2 shows that qualified government staff carry more than twice the burden of voluntary agency staff and more than six times the burden of private for-profit staff (41.1, 17.1 and 6.6, respectively). The corresponding figures across service levels and urban rural locations do not show very large differences, although the workload tends to be higher at the health centres and secondary referral levels, as well as in urban districts.
These findings demonstrate distinct location differences for HR, as well as differences in their workloads. It seems evident that private for-profit services bear smaller workloads, although the proportion of CEmOC to BEmOC staff is higher. This is not surprising, given their objectives of providing services to a smaller but wealthier segment of the population with higher expectations relative to available qualifications.
Furthermore, it seems clear that voluntary agencies staff each facility with a higher number of qualified personnel. We know that these services are often located in very remote rural areas, and it is somewhat surprising that they manage to recruit this level of qualified personnel to these areas. This could be due to factors such as better and more flexible HR policies, active training of local staff and employment contracts that include educational stipends. The fact that they have a reduced workload per qualified staff could be due to their remote location, but also due to the higher number of staff at each facility sharing the burden. Of significant importance is the flexibility with which each facility recruits personnel, based on the workload at the facility.
Government services differ from voluntary agencies as government facilities cannot easily recruit more personnel as the workload increases, and they are also subject to less personnel redistribution flexibility. The high workload per qualified staff ratio at government facilities could be due to the relatively low number of staff per facility, but also to the large number of deliveries conducted at government services (secondary referral hospitals) in urban areas
The higher workload encountered at the health centres could be explained by the relatively high number of qualified staff at health centres compared to dispensaries, assuming mothers prefer to deliver at a facility with a higher number of qualified staff [23]. This issue will be discussed later in the article.
Differences in the distribution of qualified staff
One can analyse differences in qualified staff distributions by examining their service levels, populations at the district level and urban and rural characteristics of the district.
There are large variations in availability of qualified BEmOC and CEmOC staff in the different health service levels across the districts. This difference is notable at dispensary and first referral hospital levels. As shown in Table 3, the dispensary level figures range from 5.0 to 1.1 for qualified BEmOC staff per facility. The variations at the health centre level are much smaller. Compared to minimum national standards (as described earlier), almost all the rural districts are understaffed at the dispensary level, while at health centre level this is true for the facilities in only one rural district.
Table 3 Distribution of qualified human resources across districts according to population and health service levels
Average qualified BEmOC staff per dispensary Average qualified BemOC staff per health centre Average qualified BemOC staff per first referral hospital Average qualified CemOC staff per first referral hospital BemOC qualified personnel per 100 000 CemOC qualified personnel per 100 000
Moshi 5.0 8.5 8.5 1.5 257 18
Hai 2.3 7.5 38.5 3.5 78 3
Arusha 3.0 9.3 18.8 1.8 107 11
Arumeru 1.7 3.0 36.7 3.0 33 2
Hanang 1.3 7.3 a 0.0 21 0
Mbulu 1.1 4.7 81.0 4.0 88 4
Urban 4.0 8.9 13.7 1.7 159.6 13.5
Rural 1.6 5.6 39.0 2.6 51.3 2.1
Total 1.8 7.3 31.9 2.5 79.5 5.1
aHanang district did not have a first referral hospital.
In terms of available qualified staff per population, Table 3 shows a greater than tenfold difference in the availability of BEmOC qualified staff across the districts (21/100 000 to 257/100 000). The relative difference is even higher in terms of CEmOC qualified staff, although comparisons across districts is difficult, given the presence of secondary referral hospitals in the urban districts. Nevertheless, only one rural district does not comply with national guidelines in terms of available BEmOC personnel per 100 000 residents.
Relating the distribution of HR to their EmOC quality does not allow a direct comparison with other studies because of different methods used. Nonetheless, it is useful to compare the availability of BEmOC and CEmOC staff to the availability of nurses and doctors, respectively. This comparison reveals that the staff availability described for the entire study area (5.1 CEmOC staff per 100 000 and 79.5 BEmOC staff per 100 000) is very similar to the previous WHO estimates in Tanzania (4.1 doctors per 100 000 and 85.2 nurses per 100 000), and that Tanzania has the lowest qualified staff availability compared to other African countries [24], as shown in Table 4. Thus, it is reasonable to assume the national guidelines probably accept an unreasonably low number of qualified personnel per facility and thus per population, compared to other countries with similar HR characteristics.
Table 4 Estimates of health personnel per 100 000 population in selected African countries
Country Year Physicians Nurses
Angola 1997 7.7 114.5
Botswana 1994 23.8 219.1
Democratic Republic of Congo 1996 6.9 44.2
Ghana 1996 6.2 72.0
Lesotho 1995 5.4 60.1
Kenya 1995 13.2 90.1
Namibia 1997 29.5 168.0
South Africa 1996 56.3 471.8
Swaziland 1996 15.1 ...
Tanzania 1995 4.1 85.2
Zambia 1995 6.9 113.1
Zimbabwe 1995 13.9 128.7
Source: WHO estimates of health personnel; 1998
Interestingly, Table 3 does not confirm the expected finding of a uniformly higher qualified staff availability in urban districts. On the contrary, the availability of qualified staff is higher in the rural areas in the first referral hospitals of both staff categories. The higher staffing levels at voluntary agency facilities, as shown in Table 2, explain this discrepancy.
It has already been demonstrated in previous articles that the rural areas in Northern Tanzania depend heavily on voluntary agency EmOC services [14,15]. These findings are interesting for two reasons. First, it is clear that the total availability of qualified HR is much lower in rural districts than in urban districts. Secondly, the available qualified rural district staff are located at higher levels of the health care pyramid. This observation might be perfectly logical given the low level of available resources while at the same time the service providers are striving for adequate quality services.
Availability of human resources
It is also useful to examine whether there exists a relationship between qualified staff levels and health care distribution equity. Health care equity is assessed in terms of service use. First, there is the question of whether a relationship between expected workload and actual workload exists. Second is the question of whether an increased number of qualified staff translates into a higher use of health care services measured with EmOC indicators, and finally if a higher availability of qualified staff translates into an increased number of qualified EmOC facilities, both at facility and district level.
Expected and actual workload at district level
Using census data, we anticipate that an increased number of expected deliveries should be followed by an increased allocation of qualified HR. The figures, however, show that urban districts exhibit consistently higher numbers of qualified staff. Table 5 summarizes the large variations in expected workloads among the districts, ranging from 13 to 212 anticipated deliveries per qualified BEmOC. We observed the same variations in expected deliveries per qualified CEmOC personnel, ranging from 192 to 2302. Both variables indicate a higher expected workload per qualified staff member in rural districts compared to those in urban districts (85 per BEmOC, and 2063 per CEmOC in rural districts; 22 per BEmOC and 255 per CEmOC in urban districts). This discrepancy is likely explained by the fact that EmOC staff members at the secondary referral hospitals are included in these figures. By definition, this staff category is allocated not only to the district within which they are placed, but also to the entire study area.
Table 5 Expected and actual workload per qualified personnel in terms of deliveries, complications and Caesarean Sections across districts
Expected deliveries per total qualified BEmOC personnel Deliveries per qualified BemOC personnel Expected deliveries per total qualified CemOC personnel Deliveries per qualified CemOC personnel Complications per qualified BemOC personnel Complications per qualified CemOC personnel Caesarean sections per qualified CemOC personnel
Moshi 13 18 192 252 6.4 92 40
Hai 56 21 1564 591 3.2 91 11
Arusha 32 50 309 478 11.9 115 46
Arumeru 131 26 2302 450 3.6 62 17
Hanang 212 34 a a 1.5 a a
Mbulu 50 26 1197 635 4.2 101 41
Urban 22 32 255 373 8.8 105 44
Rural 85 25 2063 609 3.5 86 23
Total 52 28 817 446 6.3 99 37
aHanang district had no qualified CemOC facilities or CemOC staff
We might expect a similarly large variation in workload per qualified EmOC personnel, given the large variation in staff allocation, as outlined above. No such variation exists, however. While the anticipated workload of qualified BEmOC staff is nearly 400% in rural districts compared to urban districts, the actual workload in rural districts is only 80% of that in urban districts.
The same is found for CEmOC qualified personnel. The anticipated workload of qualified CEmOC staff is more than 800% higher in rural districts compared to urban, while the actual workload is only about 160% higher. Again, this shows that either there is a very low percentage of facility deliveries, or the mothers travel across district boundaries to urban districts for obstetric health care. Although the answer is likely complex, the figures support the latter: that the actual workload in urban districts is higher than the expected workload (32 actual to 22 expected for BEmOC personnel and 373 actual to 255 expected for CEmOC personnel). These excess obstetric deliveries likely originate in neighboring rural districts.
A more detailed discussion about these usage patterns was presented previously [14,15]. The workload, in terms of deliveries per qualified personnel, is nevertheless still higher in rural districts. The other actual workload indicators (complications and caesarean sections per qualified personnel) are higher in the urban districts. This could indicate that the referral system is functional, such that complicated deliveries and caesarean sections are conducted in areas where available qualified personnel are located.
Availability of qualified HR and use
The Spearman's rank analysis shows a significant positive correlation between resource availability and use (met need rs = .943, percent deliveries in EmOC facilities rs = .829). The BEmOC per 500 000 statistic is used as the indicator of available HR, as BEmOC staff are present at all levels of the health care pyramid, while the CEmOC staff are only present at the top of the pyramid.
Table 6 illustrates the association between the availability of qualified HR and HR use and the availability of qualified HR and the number of qualified EmOC facilities. There is an increased likelihood that pregnant women choose facilities with qualified personnel. This is verified by publications on Northern Tanzanian clinical data showing that patients voluntarily bypass low-quality services in favour of high-quality services [23].
Table 6 Spearman's rank correlation analysis between the rankings of available qualified BemOC staff, met need and percent deliveries in EmOC facilities (as measures of utilization) and number of qualified BemOC and CemOC facilities (as measures of quality) in the districts
BEmOC qualified personnel per 500 000
Met need Correlation coefficient .943(**)
Sig. (2-tailed) .005
Percent deliveries in EmOC facilities Correlation coefficient .829(*)
Sig. (2-tailed) .042
BEmOC facilities per 500 000 Correlation coefficient .314
Sig. (2-tailed) .544
CEmOC facilities per 500 000 Correlation coefficient .886(*)
Sig. (2-tailed) .019
* Correlation is significant at the 0.05 level. ** Correlation is significant at the 0.001 level.
Availability of qualified HR and qualified EmOC facilities
Interestingly, there is no correlation (rs = 0.314) between the availability of qualified HR and the number of qualified BEmOC facilities. There is, however, a correlation (rs = .886) between the availability of qualified HR and the number of CEmOC-providing facilities. These data imply that the availability of qualified HR could translate into a higher number of qualified services provided at the higher levels of the health care pyramid, but not at the lower levels of the health care pyramid. A possible explanation for this finding could be the importance of increased training levels needed in the complex process of service provision.
The finding emphasizes the significance of maintaining a health policy focus on process and context parameters necessary for the translation of resources into quality services. These include important issues such as that health care providers must implement more effective management, motivation and policy relevance strategies and increase equipment and drug availability. In a health reform context, this analysis supports the claim that HR must be treated as more than merely a resource, and should be afforded a larger voice in policy formulation [25-29].
It is not possible, using only the Spearman's correlation test, to firmly conclude the direction of the direct correlation found between availability of qualified HR and use. It could be that increased availability leads to increased use, and that increased use leads to increased availability. We have, however, argued previously in this article that HR allocation is inflexible in Northern Tanzania. Increased use of health care services would inevitably necessitate increased HR availability, which would also require a flexible pool of qualified personnel within the target population. The direction of the correlation is most likely due to increased availability of qualified HR, leading to increased HR use.
Similarly, we could not definitively confirm the direction of the correlation between the availability of qualified HR and the number of higher level EmOC facilities. We argue that it is unlikely that the correlation is due to an increased number of facilities leading to higher availability of HR, for the same reasons as above. Rather, it is more likely that there is a correlation between an increased number of qualified HR leading to an increased number of qualified facilities.
In the Tanzanian setting, given the lack of a flexible qualified staff pool, the above described correlation most likely shows that a lower availability of qualified staff per district leads to lower numbers of qualified facilities. We argue this on the basis of the low overall availability of qualified staff not allowing adequate staffing of newly-developed health care facilities. It is more likely that as new facilities are built, HR are distributed more sparsely across these facilities, reducing the likelihood of each unit's becoming a qualified EmOC facility
Services of increased quality are needed before an increased quantity of services can ensue
Based on the results and discussion above, a priority should be to provide an increased number of qualified BEmOC staff to the understaffed districts in order to improve quality and use of existing facilities. Ideally, and importantly, increased numbers of qualified HR are urgently needed. Assuming that no more resources are made available, we need a redistribution of resources.
Although many commentators suggest a shift from higher to lower health care system levels, we believe our data suggest that present high level services are needed to provide the requisite comprehensive emergency obstetric care services and other referral services, train other health personnel, provide contingency services to Northern Tanzanian patients and respond to the demand, rights and trust of the population. Our data suggest that the HR available to these high level services are already at a minimum and should not be reduced.
Our analysis therefore suggests a shift of HR between and among health centre and dispensary levels in almost all districts as a measure to improve access to qualified HR. We should increase the availability of qualified staff at these facilities rather than maintain substandard quality. An example of how to implement this would be to reduce by half the number of available dispensaries and use the released resources to upgrade the remaining facilities to health centre levels. The location of these new health centres should be selected based on equity measures, such as geographical position, to maximize equity concerns.
This scenario should increase quality service availability at the expense of inadequate quality coverage. Such a situation can occur successfully if the managerial capacity and sustainable policy environment are present to translate these resources into good-quality services. Increasing the geographical range in which good services are provided necessitates improving patient access to these services. Thus, we must adopt and implement policies for improving transportation and communications infrastructure to facilitate remote access to health care services as well as to other types of important services.
Reducing the number of nonqualified facilities to the benefit of qualified facilities is useful for the supply of emergency obstetric care services. It remains to be investigated if it also applies for other services. It is likely that it at least applies to all services requiring higher levels of qualifications.
Applying this proposed solution to our study illustrates the potential of such a redistribution mechanism. There are 174 qualified BEmOC personnel distributed across 93 dispensaries. Further investigation is needed to find the adequate number of qualified staff needed to ensure a qualified facility. However, to provide the requisite four qualified personnel per health centre (as proposed by Tanzanian HR policies), these 174 health workers should be distributed across 43 health centres rather than the 93 dispensaries in the study area. If all these 43 new health centres provided high-quality services at the BEmOC level, according to the United Nations guidelines, the coverage of these services would be equivalent to 14 BEmOC facilities per 500 000 people, compared to the current 1.6 BEmOC facilities per 500 000 people reported in the study area (Table 1). The accepted minimum according to the United Nations guidelines is four BEmOC facilities per 500 000 people.
Figure 1 illustrates the distribution of deliveries per facility on a log scale. More than 72% of the facilities conduct fewer than 100 deliveries per year. Almost all these facilities are dispensaries. Reducing the number of facilities would increase the number of deliveries per staff in these facilities, and thus also the quality of services. This example illustrates the potential of redistributing HR for quality rather than quantity care, at least for services requiring higher levels of qualifications, such as emergency obstetric care services.
Figure 1 Distribution of deliveries per facility
Although this article demonstrates distribution differences in qualified HR across different contextual levels (urban/rural, ownership and health service level), it is useful to discuss briefly whether these differences are as inequitable as often described.
Agencies and commentators frequently criticize the relative, and thus perceived, excess HR working at higher levels of the health care pyramid. This excess, however, is understandable, given the extremely low number of qualified personnel in Tanzania. Indeed, Tanzanian national health authorities have few alternatives to placing qualified personnel in locations easily accessible by many people, and with adequate equipment and resources. It is better to provide adequate quality in higher-level facilities covering more people, than it is to provide inadequate quality in lower-level facilities covering fewer people, as argued previously in this article.
Neither should it be considered inequitable that much of the qualified HR are found in the private not-for-profit sector, given that this sector shares the same visions and objectives (equity, efficiency and quality) as the government. We should instead view this perspective in the wider context of the role of the state, in which the regulatory role is more important than that of the state as provider. Of concern is the total provision of good-quality services, accessible to all but not necessarily with the same overall coverage, given the severe resource constraints.
The issue at stake is not ownership or coverage, however, but rather health care quality, accessibility and trust. A high coverage of inadequate quality is not pro-poor. On the contrary, it has been demonstrated that low-quality services contribute to increased poverty [30]. The priority-setting issue with regard to HR in the study area should be to secure access to a determined threshold of quality, with increased availability of this level of quality provided as resources and managerial capacity are made available.
Conclusion
We reveal in this article that there are adequate numbers of EmOC staff at the aggregated district level compared to Tanzanian HR requirements. However, there are large variations in the availability of qualified staff within these districts and severe understaffing at the dispensary levels in rural districts. The total number of available staff in Tanzania is also very low compared to other African countries. We also demonstrate that the availability of staff is concentrated within urban districts and in voluntary agency and government facilities.
Furthermore, there is a much higher workload per qualified staff at the government facilities. This is likely due to systematic understaffing at these facilities, given the expected workload. A significant reason for this may include the lack of flexibility in government facilities to adjust staffing levels to mirror the actual workload. The distribution of the qualified HR has important equity implications in that the pregnant women have to travel long distances to reach the qualified facilities.
Furthermore the article discusses possible relationships between higher availability of qualified staff, and the use and quality of services. The article argues that the availability of qualified staff is an important determinant of use of the services by the women, but that increased availability of qualified staff does not automatically translate into higher numbers of qualified facilities. We recommend that this relationship be investigated further, since developing more facilities necessitates greater management capability and enhanced clinical routines and motivation, as well as a critical exploration of the policy environment within which the services are provided.
The article encourages new HR targets for countries like Tanzania in which availability of quality services and not coverage alone should be the main focus. Increasing the availability of qualified HR must be a priority. Assuming however, that this increase will not happen in the near future, and given the present availability of resources at the different health care levels, we recommend a shift of resources within the dispensary level to reduce the total number of accessible dispensaries, rather than to redistribute HR from higher to lower quality levels. The remaining dispensaries should be upgraded to health centre-level quality, using the liberated resources, based on geographical and demographic parameters for equity maximization. The overall goal should be to improve access (transport and communications) to and coverage of quality services. Whether this is true for all types of services requires further investigation, but it is likely that it is relevant for all services requiring some form of higher qualifications.
The article also questions the criticism of uneven HR distribution between high and low levels of services often voiced, given the extremely low level of available resources. There are many good reasons for this distribution profile. The issue at stake is not ownership or coverage, but rather quality, accessibility, efficiency and trust of the services provided. The priority-setting discussions should therefore focus on defining an accepted threshold of quality under the present circumstances, determining how to distribute these services as widely as possible and deciding how to increase the availability of the current resources to optimize coverage of quality services. Service coverage indicators alone are not very useful for this process.
List of abbreviations
BEmOC basic emergency obstetric care
CEmOC comprehensive emergency obstetric care
CBR crude birth rate
EmOC emergency obstetric care
G government services
HMIS health management information systems
HR human resources
MDGs Millennium Development Goals
PFP private, for-profit services
UNFPA United Nations Population Fund
UNICEF United Nations Children's Fund
VA voluntary agencies services
WHO World Health Organization
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
ØEO was the principal investigator involved in all stages of the project as well as main author of the manuscript. SN participated in the conceptual outline of the project, methodological review and review of the manuscript. OFN supervised all stages of the project and participated in the analysis of the data and review of the manuscript. All authors read and approved the final manuscript.
Acknowledgements
The authors wish to thank the Tanzanian health authorities and the Center for Educational Development and Health, Arusha, for their willingness to facilitate the study. We are further indebted to our colleagues at the Center for International Health and specifically thank Professor Gunnar Kvåle, Dr Sven G. Hinderaker and Dr Bjørg E. Olsen for their important comments. Finally, we wish to thank our dedicated Tanzanian field assistant, Mr John Gwaha.
==== Refs
World Health Organization The World Health Report 2003 – shaping the future 2003 Geneva
World Bank World Development Report 2004: Making Services Work for Poor People 2003 Washington DC
De Brouwere V Laabid A Van Lerberghe W De Brouwere V, Van Lerberghe W Unmet obstetrical need: an operationally relevant concept Networking for Research in Reproductive Health 1996 Brussels: ITM, Antwerp & European Commission, DGXII 6 27
Bertrand J Tsui A Indicators For Reproductive Health Program Evaluation The Evaluation Project 1995 Chapel Hill: Carolina Population Center, University of North Carolina at Chapel Hill
Nirupam S Yuster EA Emergency obstetric care: measuring availability and monitoring progress Int J Gynaecol Obstet 1995 50 S79 S88 10.1016/0020-7292(95)02492-U
Maine D Wardlaw T Ward V McCarthy J Birnbaum A Akalin M Brown J Guidelines for Monitoring the Availability and Use of Obstetric Services 1997 New York: UNICEF/WHO/UNFPA
AMDD Working Group on Indicators Using UN process indicators to assess needs in emergency obstetric services: Pakistan, Peru and Vietnam Int J Gynaecol Obstet 2002 78 275 282 12384277 10.1016/S0020-7292(02)00190-X
AMDD Working Group on Indicators Program note. Using UN process indicators to assess needs in emergency obstetric services: Bhutan, Cameroon and Rajasthan, India Int J Gynaecol Obstet 2002 77 277 284 12065143 10.1016/S0020-7292(02)00021-8
AMDD Working Group on Indicators Program note: using UN process indicators to assess needs in emergency obstetric services: Morocco, Nicargua and Sri Lanka Int J Gynaecol Obstet 2003 80 222 230 12566202 10.1016/S0020-7292(02)00390-9
AMDD Working Group on Indicators Program note: using UN process indicators to assess needs in emergency obstetric services: Niger, Rwanda and Tanzania Int J Gynaecol Obstet 2003 83 112 120 14511885 10.1016/S0020-7292(03)00187-5
Bailey PE Paxton A Program note: using UN process indicators to assess needs in emergency obstetric services Int J Gynaecol Obstet 2002 76 299 305 11880134 10.1016/S0020-7292(01)00592-6
Hussein J Goodburn EA Damisoni H Lema V Graham W Monitoring obstetric services: putting the 'UN Guidelines' into practice in Malawi: 3 years on Int J Gynaecol Obstet 2001 75 63 73 discussion 74 11597621 10.1016/S0020-7292(01)00474-X
Ronsmans C Achadi E Sutratikto G Zazri A McDermott J Use of hospital data for safe motherhood programmes in south Kalimantan, Indonesia Trop Med Int Health 1999 4 514 521 10470344 10.1046/j.1365-3156.1999.00429.x
Olsen OE Ndeki S Norheim OF Complicated deliveries, critical care and quality in emergency obstetric care in Northern Tanzania Int J Gynaecol Obstet 2004 87 98 108 15464791 10.1016/j.ijgo.2004.07.002
Olsen OE Ndeki S Norheim OF Availability, distribution and use of emergency obstetric care in northern Tanzania Health Policy Plan 2005 20 167 175 15840632 10.1093/heapol/czi022
Diallo K Zurn P Gupta N Dal Poz M Monitoring and evaluation of human resources for health: an international perspective Human Resources for Health 2003 1 3 12904252 10.1186/1478-4491-1-3
National Bureau of Statistics United Republic of Tanzania, Sensa 2002 2002 Population Census: Preliminary Report 2003 Dar es Salaam: Ministry of Finance, Economic Affairs and Planning
National Bureau of Statistics Tanzania Reproductive and Child Health Survey 1999 2000 Dar es Salaam: United Republic of Tanzania
Ministry of Health Tanzania National District Health Planning Guidelines 1998 Dar es Salaam: Primary Health Care Secretariat
Kurowski C Human Resources for Health: Requirements and Availability in the Context of Scaling Up Priority Interventions Case Studies from Tanzania and Chad 2002 London: United Kingdom Department for International Development
World Health Organization Maternal Mortality in 1995: Estimates Developed by WHO, UNICEF, UNFPA 2001 Geneva Report No.: WHO/RHR/01.9
National Bureau of Statistics Tanzania Reproductive and Child Health Facility Survey, 1999 1999 Dar es Salaam: Ministry of Health and MEASURE Evaluation
Leonard KL Mliga GR Haile Mariam D Bypassing health centres in Tanzania: revealed preferences for quality J Afr Econ 2002 11 441 471 10.1093/jae/11.4.441
World Health Organization WHO Estimates of Health Personnel 1998 Geneva
Rigoli F Dussault G The interface between health sector reform and human resources in health Hum Resour Health 2003 1 9 14613523 10.1186/1478-4491-1-9
Hongoro C McPake B Human resources in health: putting the right agenda back to the front Trop Med Int Health 2003 8 965 966 14629761 10.1046/j.1360-2276.2003.01118.x
Dussault G Dubois C-A Human resources for health policies: a critical component in health policies Human Resources for Health 2003 1 1 12904254 10.1186/1478-4491-1-1
Editorial The catastrophic failures of public health Lancet 2004 363 745 15016480 10.1016/S0140-6736(04)15712-7
Nolan T Angos P Cunha AJLA Muhe L Qazi S Simoes EAF Tamburlini G Weber M Pierce NF Quality of hospital care for seriously ill children in less-developed countries Lancet 2001 357 106 110 11197397 10.1016/S0140-6736(00)03542-X
Meessen B Zhenzhong Z Damme WV Devadasan N Criel B Bloom G Editorial: Iatrogenic poverty Trop Med Int Health 2003 8 581 584 12828538 10.1046/j.1365-3156.2003.01081.x
|
16053519
|
PMC1199615
|
CC BY
|
2021-01-04 16:37:43
|
no
|
Hum Resour Health. 2005 Jul 29; 3:5
|
utf-8
|
Hum Resour Health
| 2,005 |
10.1186/1478-4491-3-5
|
oa_comm
|
==== Front
J NeuroinflammationJournal of Neuroinflammation1742-2094BioMed Central London 1742-2094-2-181611149410.1186/1742-2094-2-18ResearchMicroglial responses to amyloid β peptide opsonization and indomethacin treatment Strohmeyer Ronald [email protected] Carl J [email protected] Diego [email protected] Brian [email protected] Andrew [email protected] Joseph [email protected] L.J. Roberts Center, Sun Health Research Institute, 10515 West Santa Fe Drive, Sun City, AZ 85351 USA2005 19 8 2005 2 18 18 18 6 2005 19 8 2005 Copyright © 2005 Strohmeyer et al; licensee BioMed Central Ltd.2005Strohmeyer et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Recent studies have suggested that passive or active immunization with anti-amyloid β peptide (Aβ) antibodies may enhance microglial clearance of Aβ deposits from the brain. However, in a human clinical trial, several patients developed secondary inflammatory responses in brain that were sufficient to halt the study.
Methods
We have used an in vitro culture system to model the responses of microglia, derived from rapid autopsies of Alzheimer's disease patients, to Aβ deposits.
Results
Opsonization of the deposits with anti-Aβ IgG 6E10 enhanced microglial chemotaxis to and phagocytosis of Aβ, as well as exacerbated microglial secretion of the pro-inflammatory cytokines TNF-α and IL-6. Indomethacin, a common nonsteroidal anti-inflammatory drug (NSAID), had no effect on microglial chemotaxis or phagocytosis, but did significantly inhibit the enhanced production of IL-6 after Aβ opsonization.
Conclusion
These results are consistent with well known, differential NSAID actions on immune cell functions, and suggest that concurrent NSAID administration might serve as a useful adjunct to Aβ immunization, permitting unfettered clearance of Aβ while dampening secondary, inflammation-related adverse events.
==== Body
Background
Chemotactic and phagocytic responses of microglia to amyloid β peptide (Aβ) have been inferred from postmortem autopsy evaluations [1-3], animal studies [4,5], and an in vitro model in which cultured rodent microglia were placed directly on Alzheimer's disease (AD) cortical sections [5,6]. Although these valuable experiments confirm that microglia cluster around and may help clear Aβ deposits, new questions have arisen concerning the effects of various agents on these microglial interactions with Aβ. In particular, several studies have indicated that the opsonization of Aβ deposits with anti-Aβ antibodies facilitates microglia-mediated Aβ clearance [6,7]. Here, binding of the antibodies to the Aβ target presumably enhances microglial recognition of and subsequent responses to the target through Fc receptors expressed by the microglia [6,7]. Based on these results, it has been suggested that microglial responses to Aβ might represent so beneficial an inflammatory action that anti-inflammatory drugs might actually be detrimental as a treatment for AD [8]. Alternatively, multiple epidemiologic studies [9,10] have reported decreased risk for AD in persons who take common nonsteroidal anti-inflammatory drugs (NSAIDs).
Over the last decade, our laboratory has developed reliable methods for culturing microglia from rapid (< 4 hour) brain autopsies of AD patients [11,12]. These cultures uniquely match the species, developmental stage, and disease state of AD subjects, and provide the ready experimental manipulability that is helpful in assessing complex physiologic processes such as chemotaxis, phagocytosis, secretory activity, and drug responses. In order to quantitatively assay these processes in the context of microglial interactions with Aβ, we seeded AD microglial cultures into wells containing pre-aggregated Aβ1-42 spots dried down to the well floor. Subsequent experiments measured migration of the cells to the Aβ spots, phagocytosis of the Aβ spots, pro-inflammatory cytokine secretion, and the effects on these processes when Aβ spots were opsonized with an anti-Aβ antibody or when microglia were treated with a common nonsteroidal anti-inflammatory drug (NSAID), indomethacin. Overall, opsonization with Aβ antibody enhanced microglial migration to and phagocytosis of Aβ. Indomethacin had little to no effect on these responses, but did significantly inhibit microglial secretion of IL-6.
Methods
AD microglia cultures
Cultures of microglia from rapid (< 4 hours) autopsies of six antemortem-evaluated, neuropathologically-confirmed AD patients were prepared using our previously published methods [11,12]. By immunoreactivity, these cultures are consistently negative for neuron, astrocyte, oligodendrocyte, and fibroblast markers, consistently positive for multiple markers of activated microglia, and readily maintained at purities of 98% or higher [11,12]. Microglia cultures from all six AD patients were used for biochemical assays. Additional cultures from one of these patients were used for quantitative evaluation of chemotaxis and phagocytosis, and additional cultures from two more of these patients were used for qualitative replication of the chemotaxis and phagocytosis results. At 3–7 days post-plating, the microglia were trypsinized and replated at 50,000 cells/well in 12-well plates. Prior to replating, 2 μl of a 1 mM solution of Aβ1-42 (Bachem) in PBS (pH 7.4) was dried down to the well floor. Each well received two such Aβ spots, and there were three wells per experimental condition, so that a total of six Aβ spots were quantified per experimental condition. Serum-free medium was used throughout the experiments. Control wells containing no Aβ or no microglia were also prepared.
Treatment with anti-Aβ antibody
Prior to seeding with microglia, selected wells were pretreated with vehicle (medium) only or with 10 μg/ml 6E10 (Signet Laboratories), a mouse monoclonal antibody directed against the first 17 (N-terminal) amino acids in the Aβ sequence. In some experiments, a 2 μg/ml concentration of 6E10 was included in order to evaluate effects at a lower dose.
Treatment with indomethacin
Prior to seeding with microglia, selected wells were pretreated with vehicle (medium) only or with 1.0 μg/ml indomethacin. Indomethacin, at 1.0 μg/ml, and vehicle were also replenished at Days 3, 6, and 9 in the course of medium changes. The 1.0 μg/ml indomethacin concentration is at the upper end of the physiologically normal range achieved in blood after therapeutic doses of the drug [13], and was chosen to insure that any failure of indomethacin to affect chemotaxis to or phagocytosis of Aβ was not due to inadequate drug dosage. In some experiments, a 0.1 μg/ml concentration of indomethacin, which is at the lower end of the physiologically normal range achieved in blood after therapeutic doses, was included in order to evaluate effects at a lesser concentration.
Cytochemistry and immunocytochemistry
For qualitative evaluations of microglial responses to Aβ, microglial cultures were briefly fixed with 4% buffered paraformaldehyde, then immunoreacted overnight with 1:1000 (0.5 μg/ml) LN3 antibody (MP Biomedical) directed against the major histocompatibility complex type II cell surface glycoprotein, using our previously published methods [11,14,15]. Vectastain ABC kits (Vector Laboratories) were employed using the manufacturer's protocols to detect immunoreactivity with bright field optics. Aβ spots could be sufficiently resolved under these conditions by their modest opaqueness under bright field optics. To visualize Aβ spots in phagocytosis experiments, the wells were washed gently in distilled water (3 × for 5 min each), incubated with 0.1% Thioflavine S (Sigma) for 10 min, washed once in distilled water (5 min), then dehydrated and fixed with 4% buffered paraformaldehyde. In additional experiments, Aβ immunocytochemistry was applied in selected wells so as to graphically illustrate Aβ removal and microglial uptake of Aβ. In these studies, microglial cultures with Aβ spots were briefly fixed with 4% buffered paraformaldehyde and incubated overnight with 1:1000 (1 μg/ml) anti-Aβ antibody 4G8 (Signet Laboratories). Detection of immunoreactivity was accomplished using Vectastain ABC kits (Vector Laboratories) and the manufacturer's suggested protocols.
Microglial migration to Aβ spots
Microglial cultures were assessed on Day 3 and Day 9 after initial plating. Each Aβ spot was visualized under phase contrast optics at 100 × (10 × objective), and photomontages were made of the spot and surrounding area out to a radius of 2 mm from the spot perimeter. A grid was then placed over the photomontages. The number and percentages of microglia within four 500 μm × 500 μm (0.25 mm2) grid squares centered on the Aβ spot and within sets of four 500 μm × 500 μm squares at progressively greater distances from the spot were recorded. The distance intervals for the grid squares were 0, 500, 1000, 1500, and 2000 μm from the Aβ spot, and each distance interval was measured in quadruplicate (Fig. 1). A total of 141,455 microglia were individually hand-counted in this way. Chemotaxis was evaluated by changes in the distributions of microglia relative to the Aβ spots over time, with relatively flat distributions indicative of little or no chemotaxis, and increasingly negative slopes to the distributions indicative of migration toward the Aβ spots (Fig. 1). Slopes of the distributions (m) were operationally defined as the "chemotactic index" [15] for each condition, and the statistical reliability of the measures was assessed with Pearson's Product Momentum (R) statistic and with analysis of variance (ANOVA) techniques. The simplest ANOVAs assessed, for each treatment condition, significant differences in the distributions of microglia over the progressive distance intervals from the Aβ spot, with percentage of microglia at a particular distance (grid square) as the dependent variable and distance from the Aβ spot (0, 500, 1000, 1500, and 2000 μm) as the single factor. Pearson's R Statistic was then run to confirm that the alterations in microglial distributions were consistent with chemotaxis (i.e., showed a significant negative correlation with distance from Aβ) rather than some other response pattern. Dose dependence was evaluated using two-way ANOVAs, with percentage of microglia as the dependent variable, distance from the Aβ spot as the first factor, and drug dose as the second factor. Significant interactions of distance with drug dose thereby provided statistical evidence that the different drug doses differentially affected microglial distributions. A similar approach was taken for comparisons of different treatment conditions (e.g., anti-Aβ antibody exposure ± indomethacin treatment). All data collection was by a technician blind to experimental condition.
Figure 1 Paradigm for estimation of microglial chemotaxis to Aβ. Upper left panel shows a hypothetical example at Day 1, when microglia (black dots) are uniformly distributed relative to Aβ spots (gray circle). A plot of microglial density within 500 μm × 500 μm grid squares at increasing proximity to the spot (lower left) is therefore relatively flat, with a slope near 0, indicative of little or no migratory activity at this early time point. After 9 days (right panels), microglia are clustered over and around the Aβ spot, yielding a pronounced slope to the plot, consistent with chemotaxis to the Aβ. Previous studies have referred to such slopes as "chemotactic indices" [c.f., 15].
Tests of microglial proliferation
BrdU staining kits (Zymed/Invitrogen) were applied to selected wells in order to assess whether shifts in microglial distributions over time might be due to differential proliferation of microglia relative to Aβ spots as opposed to migration of the cells. Staining with BrdU followed the manufacturer's recommended directions.
Microglial phagocytosis of Aβ spots
At Day 12 postplating, selected wells were histochemically reacted with Thioflavine S, as described earlier, and visualized at 100 × (10 × objective) with a confocal microscope. Using the ability of the confocal microscope to optically section an object at precise distances, the number of 10 μm optical slices from the well floor to the top of the remaining Aβ spot was recorded by an investigator blinded to the experimental conditions imposed in each well. The data were then assessed statistically using 2-way ANOVAs, with spot thickness as the outcome measure, antibody treatment (vehicle only, 2 μg/ml anti-A↕ IgG, or 10 μg/ml anti-A↕ IgG) as the first factor, and NSAID treatment (vehicle only, 0.1 μg/ml indomethacin, or 1.0 μg/ml indomethacin) as the second factor.
Microglial secretion of cytokines
To assess the effects of opsonization with anti-Aβ antibodies, microglial cultures were preincubated with vehicle or 10 μg/ml anti-Aβ monoclonal 6E10 followed by 4 hours exposure to 0 or 10 μM preaggregated Aβ1-42 (Bachem). Conditioned medium was then subjected to TNF-α ELISA (R&D Systems) using the manufacturer's protocols. To confirm the results with another pro-inflammatory cytokine, and to evaluate the interaction of indomethacin with antibody opsonization, microglial cultures were preincubated with vehicle or 10 μg/ml 6E10, as before, but in the presence or absence of 1 μg/ml indomethacin. After incubation for 4 hours with 0 or 10 μM Aβ1-42, the conditioned medium was subjected to IL-6 ELISA (R&D Systems) using the manufacturer's protocols.
Results
Microglial migration to Aβ spots
Overall and within each treatment condition there were shifts in microglial distributions, consistent with chemotaxis, that were both visually apparent (Figs. 2A, 2C) and statistically significant (Figs. 2B, 2D). By Day 3, the greatest concentrations of microglia were midway between the most distal and proximal points from the Aβ spots (FDistance = 40.1, P = 0.000; R = -.17, P = 0.000; m = -.016) (Fig. 2B). By Day 9, the greatest concentrations of microglia were at or adjacent to the spots (FDistance = 99.2, P = 0.000; R = -.41, P = 0.000; m = -.041) (Fig. 2D). Microglia seeded into wells without Aβ spots essentially remained randomly distributed throughout these time periods. Opsonization with anti-Aβ antibodies significantly enhanced chemotaxis-like shifts in microglial distributions, an effect that was especially prominent at Day 9 (Table 1) (Fig. 3). Indomethacin had no significant or obvious effect on changes in microglial distributions over time under any of the Aβ antibody treatment conditions. Indeed, the largest chemotactic index (slope) observed in the study occurred at the highest dose of indomethacin (1.0 μg/ml indomethacin plus 10 μg/ml anti-Aβ) (FDistance = 38.9, P = 0.000; R = 0.69, P = 0.000; m = -.073), and the second largest chemotactic index occurred at the second highest dose of indomethacin (0.1 μg/ml indomethacin plus 10 μg/ml anti-Aβ) (FDistance = 12.9, P = 0.000; R = -.53, P = 0.000; m = -.060 (Fig. 3).
Figure 2 Typical responses of cultured AD microglia to pre-aggregated Aβ1-42 spots dried down to the well floor. A) Micrograph of Aβ spot (light brown stain) and LN3 immunoreactive microglia (blue stain) 3 days postplating (vehicle control) (4 × objective). B) Graphic summary of microglial distributions at 3 days postplating (pooled data over all conditions). C) Parallel well 9 days postplating (vehicle control) (4 × objective). Wells seeded with microglia but without Aβ spots exhibited only random distributions of cells (not shown). D) Graphic summary of microglial distributions at Day 9 (pooled data over all conditions). Similar and highly significant shifts over time were observed in all treatment conditions when Aβ spots were present (see text).
Table 1 Effects of opsonization with anti-Aβ antibody 6E10 on chemotaxis-like changes in microglia distributions
ANOVA PEARSON'S SLOPE
F P R P m
Day 3
0 μg/ml anti-Aβ 3.7 0.007 -0.26 0.005 -0.022
2 μg/ml anti-Aβ 2.5 0.040 -0.14 NS -0.023
10 μg/ml anti-Aβ 5.5 0.000 -0.27 0.003 -0.027
Dose dependence* 3.6 0.008
Day 9
0 μg/ml anti-Aβ 5.6 0.000 -0.37 0.000 -0.040
2 μg/ml anti-Aβ 11.2 0.000 -0.050 0.000 -0.051
10 μg/ml anti-Aβ 16.4 0.000 -0.57 0.000 -0.056
Dose dependence* 2.3 0.050
*Dose × distance interaction term
Figure 3 Microglial distributions after 9 days incubation with Aβ spots. A) Treatment with 2 μg/ml anti-Aβ antibody plus (yellow) or minus (green) 1 μg/ml indomethacin (INDO). B) Treatment with vehicle control (red) or 10 μg/ml anti-Aβ antibody plus (yellow) or minus (green) 1 μg/ml indomethacin. C) Representative phase contrast image (4 × objective) of microglia and an Aβ spot when treated with vehicle only. D) Representative phase contrast image (4 × objective) of microglia and an Aβ spot when treated with 10 μg/ml anti-Aβ antibody plus 1 μg/ml indomethacin.
Differential proliferation versus chemotaxis
Proliferation of microglia more proximal to the Aβ spots, rather than true chemotaxis, did not explain the shifts in microglial distributions that were exhibited over time under the various treatment conditions. There was little to no BrdU staining under any condition (not shown) and, in fact, there was a slight but significant decrease in microglial numbers in all treatment conditions and overall from Day 3 (mean microglial density/0.25 mm2 grid square = 40.8 ± 0.3) to Day 9 (mean microglial density/0.25 mm2 grid square = 37.8 ± 0.4) (FOverall = 34.5, P = 0.000). Consistent with our previous experience, AD microglia stimulated with M-CSF as a positive control showed little to no evidence of proliferation. However, M-CSF-stimulated THP-1 cells (a monocyte line often used as a surrogate for microglia) that were run in parallel did show clear proliferation under the same BrdU assay conditions (data not shown).
Microglial phagocytosis of Aβ
After incubation with microglia under the various experimental conditions, visible degradation of Aβ spots was apparent (Fig. 4A), whereas Aβ spots in wells not containing microglia remained visibly intact over the same time periods (Fig. 4B). Concurrent with degradation of the Aβ spots, microglia in contact with the spots became Aβ immunoreactive (Fig. 4A), whereas they exhibited little to no Aβ immunoreactivity prior to their being seeded into the wells (Fig. 4C). Opsonization of Aβ spots with 2 μg/ml anti-Aβ antibody 6E10 (F = 28.7, P = 0.006) or 10 μg/ml anti-Aβ antibody 6E10 (F = 35.3, P = 0.004) resulted in significantly smaller (thinner) Aβ spots compared to the vehicle control condition (Fig. 4D). These effects were not significantly or materially inhibited by indomethacin even at the highest, 1.0 μg/ml indomethacin concentration (for 2 μg/ml anti-Aβ ± 1.0 μg/ml indomethacin: F = 0.3, P = 0.639) (for 10 μg/ml anti-Aβ plus ± 1.0 μg/ml indomethacin: F = 0.9, P = 0.402) (Fig. 4D).
Figure 4 Evidence for phagocytosis of Aβ by AD microglia in vitro under the various experimental conditions. A) Twelve days after plating AD microglia with Aβ spots, diminution of the spots was visually apparent and microglia concurrently had become immunoreactive for Aβ even under vehicle control conditions, as shown here (anti-Aβ antibody 4G8 immunocytochemistry). B) In the absence of microglia, the Aβ spots remained visibly intact (phase contrast). C) Likewise, prior to exposure to Aβ spots the microglia exhibited little or no immunoreactivity for Aβ (anti-Aβ antibody 4G8 immunocytochemistry). D) Summary data illustrating the effects of indomethacin and 6E10 opsonization on Aβ spot thickness. Microglia in this model system carpet the top of Aβ spots (c.f., Fig. 2C) and therefore appear to clear the Aβ from the top down, resulting in progressive thinning of the spot, as measured here. With prolonged exposure, cracks and holes in the spot appear, as shown in Fig. 4A.
Microglial secretion of cytokines
Consistent with our previous studies covering a wide range of cytokines, chemokines, and inflammatory toxins [12], exposure of microglia to Aβ significantly enhanced secretion of TNF-α (Fig. 5A) and IL-6 (Fig. 5B) compared to cultures that were not exposed to Aβ. Opsonization with 10 μg/ml anti-Aβ antibody 6E10 significantly enhanced Aβ-induced TNF-α (Fig. 5A) and IL-6 secretion (Fig. 5B). Enhancement of IL-6 expression, however, was significantly decreased by indomethacin treatment (Fig. 5B). Cytokine secretion is typically a fairly rapid response that wanes over time. Presumably, cytokine receptive cells then undergo more long-lasting responses such as enhanced chemotactic or phagocytic behaviors. Consistent with this, we observed significant changes in TNF-α and IL-6 levels 4 hours after exposure of microglia to Aβ, but not 3, 6, or 9 days after exposure to Aβ (data not shown).
Figure 5 Effects on microglial TNF-α (A) and IL-6 (B) secretion into the medium in the presence or absence of Aβ, as well as after pretreatment of Aβ with 10 μg/ml anti-Aβ antibody 6E10. Opsonization with 6E10 significantly enhanced (P < 0.05) (*) TNF-α and IL-6 levels compared to Aβ alone. IL-6 experiments also measured the effect of 1 μg/ml indomethacin on 6E10 exacerbation of cytokine secretion. Indomethacin significantly reduced this effect (P < 0.05) (#).
Discussion
The present study found that AD microglia in vitro migrate toward Aβ aggregates, attempt to phagocytose the aggregates, and increase their secretion of TNF-α and IL-6 in the process. Opsonization of Aβ aggregates with anti-Aβ antibody 6E10 significantly enhanced these processes. By contrast, the common NSAID indomethacin had no material or statistical effect on microglial migration or phagocytosis, but significantly inhibited the increased IL-6 secretion observed with anti-Aβ opsonization.
The shifts in microglial distributions relative to Aβ spots over time are most parsimoniously explained by chemotactic responses to Aβ. Proliferation of microglia more proximal to Aβ aggregates was not observed and, in fact, BrdU reactivity, a common marker for cell proliferation, was negligible at all distances from the aggregates. Chemokinesis, enhanced but undirected movement of cells, also did not appear to explain the results, since microglial migration exhibited the gradient characteristics of chemotaxis, with progressive increases in the density of microglia at distances more proximal to Aβ aggregates. In addition, microglia are now well established to express receptors that can mediate chemotactic behaviors and that appear to have Aβ as a ligand. These include the macrophage scavenger receptor [16-18], the receptor for advanced glycation endproducts (RAGE) [15], the formyl peptide receptor [19], and others [20,21]. RAGE, in particular, has been shown to help mediate microglial migration to Aβ spots in an in vitro paradigm similar to that used here, and this migration could be inhibited by anti-RAGE Fab fragments [15].
AD microglia in vitro also exhibited behaviors consistent with phagocytosis of Aβ aggregates. Entering the paradigm, the microglia showed little or no Aβ immunoreactivity. After 12 days incubation with Aβ spots, the microglia were highly immunoreactive for Aβ and the spots decreased in size. Aβ spots without microglia remained essentially intact over the same time period. Previous ultrastructural and other studies [3,22,23] have also identified Aβ filaments within microglia in the vicinity of Aβ deposits in AD cortex. Although it remains possible that the intracellular Aβ within microglia in the AD brain may have been produced by the cells [24] rather than phagocytosed from an extracellular deposit, this is clearly not the process observed in the present in vitro studies. We conclude, therefore, that AD microglia in vitro do phagocytose aggregated Aβ deposits. Given the experimental accessibility of the model, it will be of interest in future to evaluate the molecular fate of phagocytosed Aβ in cultured AD microglia.
Exposure to aggregated Aβ also induced significant increases in TNF-α and IL-6 secretion, confirming our previous experiments [12] and those of others [25-27] with TNF-α, IL-6, and a broad range of chemokines, cytokines, and inflammatory toxins such as reactive oxygen/nitrogen species. Pathways for enhancing TNF-α and IL-6 secretion have been demonstrated, including NF-kB and C/EBP transcriptional mechanisms, both of which are enhanced in pathologically-vulnerable regions of the AD brain [28,29].
Opsonization of Aβ spots with anti-Aβ antibody 6E10 significantly enhanced microglial migration to the spots, phagocytosis of the spots, and cytokine secretion. Similar effects of opsonization on microglial migration and phagocytosis have also been reported using anti-Aβ antibodies and an in vitro preparation in which cultured rodent microglia were seeded onto postmortem AD cortex sections laden with Aβ deposits [6]. Soluble Fab fragments containing the Fc region ligand for Fc receptor binding inhibited Aβ removal in this paradigm. These effects are consistent with the classic mechanisms of antibody opsonization of immune targets by antibodies specific to epitopes on the target. Scavenger cells that express receptors to the Fc region of the antibodies are then directed to or become focused at the site where the antibody-bound target resides. Fc receptor activation, in addition, activates scavenger cells, promoting attack and phagocytosis. Recently, scientists at Elan Pharmaceuticals have attempted to harness these mechanisms to enhance Aβ clearance, using immunization with Aβ to drive production of anti-Aβ antibodies for subsequent Aβ opsonization [6,30]. Although there is controversy about the exact site of action of the antibodies (e.g., brain versus peripheral circulation) [6,30,31], this approach does clearly result in significant and sometimes dramatic reductions of Aβ burden in transgenic mouse models [6], as well as the in vitro model tested here, and may also have been effective in human patients receiving Aβ immunization [30].
Unfortunately, however, inflammatory responses are often a two-edged sword. Fc receptor binding is known to enhance the activation and pro-inflammatory secretory responses of scavenger cells that bear Fc receptors, and microglia do express these receptors [6,32]. The increased TNF-α and IL-6 secretion observed in the present experiments after opsonization of Aβ aggregates with a specific anti-Aβ antibody, 6E10, is therefore not unexpected. On activation, microglial cells are, in fact, well established to secrete a wide range of inflammatory mediators that could not only cause damage to neurons and neurites locally, but also, if sufficiently activated, provide signalling to peripheral immune cells to provoke a more generalized and severe response such as that reported in several Aβ-immunized patients who experienced lethal adverse reactions [30].
The vast majority of NSAIDs in use today are based on the principle of cyclooxygenase inhibition, and cyclooxygenase inhibition, in turn, is well established to downregulate a wide range of acute phase reactants. Interestingly, however, mechanisms for chemotaxis to and phagocytosis of an inflammatory target are not necessarily cyclooxygenase dependent. In a survey, for example, of the first 100 publications retrieved from PubMed using the search phrase "indomethacin AND chemotaxis", the majority of studies found no effect of indomethacin on chemotaxis, and some of the papers actually reported enhanced chemotaxis after indomethacin exposure. Such findings have been suggested to explain why physicians commonly prescribe NSAIDs to control fever and other secondary inflammatory responses without being unduly concerned about hampering immune-mediated removal of the fever-inducing agent. Similarly, in the present experiments indomethacin had no material or statistically significant effect on microglial chemotaxis to or phagocytosis of Aβ aggregates, but did significantly inhibit the exacerbated IL-6 response under opsonized conditions. Although it is never certain that in vitro results will fully apply to the in vivo state, these results suggest that indomethacin or an NSAID like it might be a useful adjunct to Aβ immunization strategies.
Competing interests
JR is a co-inventor on an issued United States patent covering use of nonsteroidal anti-inflammatory drugs as a treatment for Alzheimer's disease. All other authors declare that they have no competing interests.
Authors' contributions
JR conceived and designed the experiments, performed all data analysis, and wrote the manuscript. RS supervised and took part in all experiments. CJK performed the chemotaxis, phagocytosis, and cytokine experiments. DM, BL, and AG prepared cultures and performed histochemistry and immunocytochemistry.
Acknowledgements
This research was directly supported by NIA AGO7367. Institutional support for Alzheimer's research was provided by the Arizona Alzheimer's Disease Core Center (P30 AG019610) (NIA) and the Arizona Alzheimer's Consortium (State of Arizona). We thank Kyle Mueller, Gita Seetharaman, and Leyla Descheny for technical assistance, and Dr. Emily Lue and Dr. Douglas Walker for technical advice.
==== Refs
Luber-Narod J Rogers J Immune system associated antigens expressed by cells of the human central nervous system Neurosci Lett 1988 94 17 22 3266526 10.1016/0304-3940(88)90263-7
Rogers J Luber-Narod J Styren SD Civin WH Expression of immune system-associated antigens by cells of the human central nervous system: relationship to the pathology of Alzheimer's disease Neurobiol Aging 1988 9 339 349 3263583
Terry RD Wisniewski HM Structural and chemical changes of the aged human brain Psychopharmacol Bull 1975 11 46 63 1197595
Weldon DT Rogers SD Ghilardi JR Finke MP Cleary JP O'Hare E Esler WP Maggio JE Mantyh PW Fibrillar beta-amyloid induces microglial phagocytosis, expression of inducible nitric oxide synthase, and loss of a select population of neurons in the rat CNS in vivo J Neurosci 1998 18 2161 2173 9482801
Ard MD Cole GM Wei J Mehrle AP Fratkin JD Scavenging of Alzheimer's amyloid beta-protein by microglia in culture J Neurosci Res 1996 43 190 202 8820967 10.1002/(SICI)1097-4547(19960115)43:2<190::AID-JNR7>3.0.CO;2-B
Bard F Cannon C Barbour R Burke RL Games D Grajeda H Guido T Hu K Huang J Johnson-Wood K Khan K Kholodenko D Lee M Lieberburg I Motter R Nguyen M Soriano F Vasquez N Weiss K Welch B Seubert P Schenk D Yednock T Peripherally administered antibodies against amyloid beta-peptide enter the central nervous system and reduce pathology in a mouse model of Alzheimer disease Nat Med 2000 6 916 919 10932230 10.1038/78682
Bard F Barbour R Cannon C Carretto R Fox M Games D Guido T Hoenow K Hu K Johnson-Wood K Khan K Kholodenko D Lee C Lee M Motter R Nguyen M Reed A Schenk D Tang P Vasquez N Seubert P Yednock T Epitope and isotype specificities of antibodies to beta – amyloid peptide for protection against Alzheimer's disease-like neuropathology Proc Natl Acad Sci U S A 2003 100 2023 2028 12566568 10.1073/pnas.0436286100
Schenk DB Yednock T The role of microglia in Alzheimer's disease: friend or foe? Neurobiol Aging 2002 23 677 679 12392771 10.1016/S0197-4580(02)00034-9
McGeer EG McGeer PL Clinically tested drugs for Alzheimer's disease Expert Opin Investig Drugs 2003 12 1143 1151 12831349
McGeer EG McGeer PL Brain inflammation in Alzheimer disease and the therapeutic implications Curr Pharm Des 1999 5 821 836 10526088
Lue LF Brachova L Walker DG Rogers J Characterization of glial cultures from rapid autopsies of Alzheimer's and control patients Neurobiol Aging 1996 17 421 429 8725904 10.1016/0197-4580(96)00006-1
Lue LF Rydel R Brigham EF Yang LB Hampel H Murphy GM JrBrachova L Yan SD Walker DG Shen Y Rogers J Inflammatory repertoire of Alzheimer's disease and nondemented elderly microglia in vitro Glia 2001 35 72 79 11424194 10.1002/glia.1072
Bruguerolle B Barbeau G Belanger PM Labrecque G Pharmacokinetics of a sustained-release product of indomethacin in the elderly Gerontology 1986 32 277 285 3569930
Lue LF Brachova L Civin WH Rogers J Inflammation, A beta deposition, and neurofibrillary tangle formation as correlates of Alzheimer's disease neurodegeneration J Neuropathol Exp Neurol 1996 55 1083 1088 8858005
Lue LF Walker DG Brachova L Beach TG Rogers J Schmidt AM Stern DM Yan SD Involvement of microglial receptor for advanced glycation endproducts (RAGE) in Alzheimer's disease: identification of a cellular activation mechanism Exp Neurol 2001 171 29 45 11520119 10.1006/exnr.2001.7732
Honda M Akiyama H Yamada Y Kondo H Kawabe Y Takeya M Takahashi K Suzuki H Doi T Sakamoto A Ookawara S Mato M Gough PJ Greaves DR Gordon S Kodama T Matsushita M Immunohistochemical evidence for a macrophage scavenger receptor in Mato cells and reactive microglia of ischemia and Alzheimer's disease Biochem Biophys Res Commun 1998 245 734 740 9588184 10.1006/bbrc.1998.8120
Christie RH Freeman M Hyman BT Expression of the macrophage scavenger receptor, a multifunctional lipoprotein receptor, in microglia associated with senile plaques in Alzheimer's disease Am J Pathol 1996 148 399 403 8579103
El Khoury J Hickman SE Thomas CA Cao L Silverstein SC Loike JD Scavenger receptor-mediated adhesion of microglia to beta-amyloid fibrils Nature 1996 382 716 719 8751442 10.1038/382716a0
Lorton D Schaller J Lala A DeNardin E Chemotactic-like receptors and Abeta peptide induced responses in Alzheimer's disease Neurobiol Aging 2000 21 463 473 10858596 10.1016/S0197-4580(00)00092-0
Koenigsknecht J Landreth G Microglial phagocytosis of fibrillar beta-amyloid through a beta1 integrin-dependent mechanism J Neurosci 2004 24 9838 9846 15525768 10.1523/JNEUROSCI.2557-04.2004
Bourin M Ripoll N Dailly E Nicotinic receptors and Alzheimer's disease Curr Med Res Opin 2003 19 169 177 12814128 10.1185/030079903125001631
Cole GM Beech W Frautschy SA Sigel J Glasgow C Ard MD Lipoprotein effects on Abeta accumulation and degradation by microglia in vitro J Neurosci Res 1999 57 504 520 10440900 10.1002/(SICI)1097-4547(19990815)57:4<504::AID-JNR10>3.0.CO;2-H
Wisniewski HM Terry RD Morphology of the aging brain, human and animal Prog Brain Res 1973 40 167 186 4371053
Banati RB Gehrmann J Czech C Monning U Jones LL Konig G Beyreuther K Kreutzberg GW Early and rapid de novo synthesis of Alzheimer beta A4-amyloid precursor protein (APP) in activated microglia Glia 1993 9 199 210 7507467 10.1002/glia.440090305
Ito S Sawada M Haneda M Fujii S Oh-Hashi K Kiuchi K Takahashi M Isobe K Amyloid-beta peptides induce cell proliferation and macrophage colony-stimulating factor expression via the P13-kinase/Akt pathway in cultured Ra2 microglial cells FEBS Lett 2005 579 1995 2000 15792809 10.1016/j.febslet.2005.02.048
Floden AM Li S Combs CK Beta-amyloid-stimulated microglia induce neuron death via synergistic stimulation of tumor necrosis factor alpha and NMDA receptors J Neurosci 2005 25 2566 2575 15758166 10.1523/JNEUROSCI.4998-04.2005
Mrak RE Griffin WS Glia and their cytokines in progression of neurodegeneration Neurobiol Aging 2005 26 349 354 15639313 10.1016/j.neurobiolaging.2004.05.010
Li R Strohmeyer R Liang Z Lue LF Rogers J CCAAT/enhancer binding protein delta (C/EBP delta) expression and elevation in Alzheimer's disease Neurobiol Aging 2004 25 991 999 15212823 10.1016/j.neurobiolaging.2003.10.016
Bales KR Du Y Holtzman D Cordell B Paul SM Neuroinflammation and Alzheimer's disease: critical roles for cytokine/Abeta-induced glial activation, NF-kappaB, and apolipoprotein E Neurobiol Aging 2000 21 427 430 10858588 10.1016/S0197-4580(00)00143-3
Masliah E Hansen L Bard F Lee C Seubert P Games D Kirby L Schenk D Aβ vaccination effects on plaque pathology in the absence of encephalitis in Alzheimer's disease Neurology 2005 64 129 131 15642916
DeMattos RB Bales KR Cummins DJ Dodart JC Paul SM Holtzman DM Peripheral anti-Abeta antibody alters CNS and plasma Abeta clearance and decreases brain Abeta burden in a mouse model of Alzheimer's disease Proc Natl Acad Sci U S A 2001 98 8850 8855 11438712 10.1073/pnas.151261398
Song X Tanaka S Cox D Lee SC Fcgamma receptor signaling in primary human microglia: differential roles of PI-3K and Ras/ERK MAPK pathways in phagocytosis and chemokine induction J Leukoc Biol 2004 75 1147 1155 14982949 10.1189/jlb.0403128
|
16111494
|
PMC1199616
|
CC BY
|
2021-01-04 16:38:21
|
no
|
J Neuroinflammation. 2005 Aug 19; 2:18
|
utf-8
|
J Neuroinflammation
| 2,005 |
10.1186/1742-2094-2-18
|
oa_comm
|
==== Front
Mol CancerMolecular Cancer1476-4598BioMed Central London 1476-4598-4-301608683510.1186/1476-4598-4-30ResearchScreening and identification of differentially expressed transcripts in circulating cells of prostate cancer patients using suppression subtractive hybridization Li Xin [email protected] Carson [email protected] Ralph [email protected] Gennady [email protected] Adam [email protected] Jarrett [email protected] C Scott [email protected] Daniel J [email protected] Bradley P [email protected] Hsueh-Kung [email protected] Department of Urology, University of Oklahoma Health Sciences Center, Oklahoma City, OK 73104, USA2005 8 8 2005 4 30 30 13 6 2005 8 8 2005 Copyright © 2005 Li et al; licensee BioMed Central Ltd.2005Li et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Tumor metastasis and changes in host immunosurveillance are important components in cancer development. Tumor cell invasion into the bloodstream is an essential step for systemic metastasis. Currently, the detection of tumor cells in the circulation is mainly dependent upon the utilization of known epithelial cell markers. However, expression of these molecules is not limited to cancer patients; healthy people also have a small number of epithelial cells in their circulation. Utilizing these markers to detect circulating tumor cells (CTCs) cannot adequately explain the mechanisms of tumor cell survival or their development of metastatic potential in peripheral blood. The immune system can also evolve along with the cancer, actually promoting or selecting the outgrowth of tumor variants. Unfortunately, both metastasis and immunosurveillance remain mysterious and are debatable because we have yet to define the molecules that participate in these processes. We are interested in identifying the existence of expressed genes, or mRNA species, that are specifically associated with circulating cells of cancer-bearing patients using prostate cancer (PCa) as a model.
Results
We established two comprehensive subtracted cDNA libraries using a molecular technique called suppression subtractive hybridization. This technique selectively amplifies transcripts that are specifically expressed in circulating cells of either PCa patients or healthy men. Following sequencing reaction, we showed that 17 out of 23 (73.9%) sequenced clones did not match any mRNAs in the GenBank database. This result suggests that genes associated with alterations in circulating cells of cancer-bearing patients are largely unknown. Semi-quantitative RT-PCR confirmed that two genes are up-regulated in circulating cells of PCa patients, whereas another two genes are down-regulated in the same patients.
Conclusion
The comprehensive gene expression analysis is capable of identifying differentially expressed genes in circulating cells of healthy men and PCa patients. We did not attempt to enrich specific cell types in this study because phenotypes of CTCs and subsets of leukocytes participating in immunosurveillance remain largely unknown. Continuous studies of these differentially expressed genes will eventually lead us to understand the mechanisms involved in tumor metastasis and immune modulation during cancer development.
circulating tumor cellssuppression subtractive hybridizationprostate cancer
==== Body
Background
Metastasis is a sequential, multi-step process in which tumor cells detach from the primary tumor, migrate through the basement membrane and extracellular matrix, and invade the lymphatic and/or blood systems [1]. This is followed by the establishment of secondary tumors at distant sites. It has been suggested that tumor cell invasion into the bloodstream can occur earlier than the time of primary diagnosis [2]. The ability to detect occult tumor cells with metastatic potential could have a substantial clinical impact on the management of cancer patients. Most, if not all, markers developed to detect occult tumor cells of epithelium origin in peripheral blood have been based on the concept that circulating tumor cells (CTCs) continue to express epithelial cell markers [3]. Based on this concept, several epithelial cell markers have been evaluated for detecting disseminated tumor cells in the blood circulation. Frequently used molecules include cytokeratins (CKs) 7, 19, and 20 [4-6], carcinoembryonic antigen (CEA) [7,8], epidermal growth factor receptor [9] including HER-2/neu [10], mucin-1 [11], β-subunit of human chorionic gonadotropin (β-hCG) [12], and α-fetoprotein [13]. In prostate cancer (PCa) patients, the expression of prostate specific antigen (PSA) [14-16], prostate-specific membrane antigen (PSMA) [17,18], and human glandular kallikrein 2 (hK2) [19] along with other epithelial cell markers has been investigated individually or in combination [20] for their ability to detect CTCs in patients with localized and metastatic PCa. This detection strategy involves the amplification of target mRNAs species by reverse transcriptase-polymerase chain reaction (RT-PCR) [21-23]. However, the use of these markers to detect CTCs fails to explain mechanisms that regulate tumor cell survival in the circulation and the development of their metastatic capability.
Recent reports have also emphasized that the immune system actively participates in cancer formation and development. Although this concept of immune response was formulated more than half a century ago [24], the existence of cancer "immunosurveillance" is still largely unknown and debatable because we know very little about the molecules participating in this event. If cancer "immunoediting" is present under the concept of cancer immunosurveillance, we hypothesized that genes expressed in immune cells participating in this event are significantly different from their counterparts in healthy persons.
We also hypothesized that both CTCs and immune cells need to evolve through their gene expression at stages of cancer formation and progression. The identified mRNA species associated with circulating cells of cancer-bearing patients will serve as independent markers for future tumor staging and help us understand metastasis and immunosurveillance. In this study, using PCa as a model, we applied the suppression suppressive hybridization (SSH) technique [25] to establish two libraries consisting of mRNA species that are either present or absent in circulating cells of PCa patients. We sequenced a small number of clones present in these libraries, and identified that the majority, 17 out of 23 (73.9%), of the sequenced clones did not match previously identified mRNA species. From the sequenced clones, we confirmed that four genes are differentially expressed in circulating cells of healthy men and PCa patients using semi-quantitative RT-PCR. Two mRNA species were identified to be significantly elevated in PCa patients, and two mRNA species were identified to be significantly suppressed in PCa patients.
Results and Discussion
Current protocols for detecting CTCs mainly utilize known epithelial cell markers or other tissue-specific molecules [26]. However, the presence of these markers in CTCs does not correlate with their survival in the circulation and their metastatic capability. Furthermore, molecules that participate in the process of "immunosurveillance" remain poorly understood. In this report, we used a PCR-based, genome-wide gene expression analysis named SSH to establish comprehensive, subtracted cDNA libraries to catalogue mRNA species either present or absent in circulating cells of PCa patients.
The SSH libraries were constructed from two age- and race-matched, pooled sample populations, healthy men and PCa patients. Each pooled sample consisted of 25 individual double-stranded cDNA libraries derived from circulating cell poly(A)+ RNA of 25 men. We used a PCR-based method to evaluate the hybridization efficiency. After two rounds of hybridization, β-actin was amplified from a subtracted population using a pair of gene-specific primers located within the very 3' end of Rsa I digested β-actin. No DNA product was detectable after 40 cycles of amplification (Figure 1A), whereas the corresponding un-subtracted library showed the presence of abundant β-actin. This result demonstrated that β-actin, and possibly the majority of commonly expressed genes between the two sample populations, formed heterohybrids, and could not be amplified using the suppression PCR technique [25]. To demonstrate that the subtracted cDNA libraries contain potential differently expressed mRNA species, the PCR products were electrophoresed on an agarose gel following the second round of PCR. A series of DNA fragments, ranging from 300 to 1,000 bp in size, representing mRNA species specifically expressed in PCa patients were detected. The various PCR products represent the Rsa I digested cDNA fragments are shown in Figure 1B, lane 1. These results indicated that there are differentially expressed genes present only in the circulating cells of PCa patients but not healthy men. PCR products representing genes expressed only in the circulating cells of healthy men are shown in Figure 1B, lane 2.
Figure 1 Evaluation of subtraction efficiency and the presence of potential differentially expressed genes in the subtracted libraries. To determine the subtraction efficiency, β-actin was PCR amplified using a primer set located within the very 3'-end of Rsa I digested β-actin fragment following the second round of hybridization. PCR products were electrophoresed on an agarose gel. No β-actin product was detected after 40 cycles of PCR amplification in a subtracted library, whereas β-actin was detected after 25 cycles of amplification in corresponding un-subtracted library (A). To amplify differentially expressed genes in circulating cells of healthy men and PCa patients, two rounds of PCR amplification was performed following hybridization steps described in Materials and Methods. To demonstrate the presence of potential differentially expressed genes in the subtracted libraries, the final PCR products were analyzed on a 1.5% agarose gel followed by ethidium bromide staining. We detected a series of distinct bands ranging from 300 to 1,000 bp. These DNA fragments represented genes that are either present (B, lane 1) or absent (B, lane 2) in circulating cells of PCa patients.
To reveal the identities of the cDNA clones, the second round PCR products from both subtracted libraries were subcloned into the pCRII TA cloning vector. A total of 23 clones from both subtracted libraries were randomly selected for sequencing using M13 reverse primer. Identities of these clones are listed in Table 1. A majority of the sequenced clones, 17 out of 23 (73.9%), matched to genomic DNA fragments in the GenBank database, but not previously identified mRNA species. These results might reflect that some of these clones were amplified from rare CTCs in PCa patients and mRNA species reflecting these cells' biology or pathology have not been identified using traditional cDNA construction and sequencing. In addition, it is possible that genes identified from both subtracted libraries may represent previously un-identified molecules participating in tumor-immune system interactions [27-30].
Table 1 Identities of selected cDNA clones present in subtracted libraries
Clone I.D. GenBank Accession No. Gene Description
PCa-001* AC026205 Homo sapiens chromosome 3 clone RP11-61I9 map 3p, complete sequence
PCa-002 AC019106 Homo sapiens BAC clone RP11-479L11 from 2, complete sequence
PCa-004 AY341247.1 Homo sapiens integral membrane protein 2B (ITM2B) gene, complete cds
PCa-005 AC104771.4 Homo sapiens BCA clone RP11-1E1 from 4, complete sequence
PCa-006 AC132068 Homo sapiens chromosome 16 clone CTD-2326c4, complete sequence
PCa-007 AK091994 Home sapiens cDNA FLJ34675 fis, clone Liver2001608
PCa-008 AC092910.9 Homo sapiens 3 BAC Rp11-767L7 (Roswell Park Cancer institute human BCA Library)
PCa-009 AC004690.2 Homo sapiens PAC clone RP-630c24 from 7, complete sequence
PCa-010 AC097461 Homo sapiens bCA clone RP11-6P6 from 2, complete sequence
PCa-011 BC047553 Homo sapiens calmodulin 2 mRNA (phosphorylase kinase δ)
PCa-012 AL031274 Homo sapiens chromosome 1q24 (clone RP4-798A17) contains the 3' part of the FMO1 gene and the FMO4 gene
PCa-013 AC010369 Homo sapiens chromosome 5 (clone CTC-2048F20)
PCa-014 NG_002397 Homo sapiens major histocompatibility complex, class I, BC (HLA-BC)
PCa-015 BC016320 Homo sapiens cathepsin D (Lysosomal aspartyl protease) mRNA
PCa-016 AC021701 Homo sapiens chromosome 18 (clone RP11-704G7)
Nrml-001** AC004914.1 Homo sapiens PCA clone RP5-88608 from 7, complete sequence
Nrml-002 AK095899.1 Homo sapiens cDNA FLJ38580 fis, clone HCHON2008582, highly similar to ferritin heavy chain
Nrml-003 AC006083 Homo sapiens chromosome 17, clone hRPK.1053_B_8, complete sequence
Nrml-004 AL109759.4 Human chromosome 14 DNA sequence BAC R-898B23 of library RPCI-11 from chromosome 14 of Homo sapiens (Human), complete sequence
Nrml-005 AK026823.1 Homo sapiens cDNA: FLJ23170 fis, clone LNG09984
Nrml-006 AC019335.5 Homo sapiens chromosome 8, clone RP11-453N18, complete sequence
Nrml-007 AL162252.17 Human DNA sequence from clone RP11-55J24 on chromosome 9, complete sequence
Nrml-008 AC016644.9 Homo sapiens chromosome 5 clone RP11-52M14, complete sequence
* PCa clones were selected from the subtracted cDNA library that represents mRNA species only present in circulating cells of PCa patients.
** Nrml clones were selected from the subtracted cDNA library that represents mRNA species absent in circulating cells of PCa patients but present in their counterparts in cancer-free healthy men.
Since a portion of the subtracted cDNAs may be false positive clones [25], we needed to confirm the identified clones as truly differentially expressed genes in our two sample populations. We used semi-quantitative RT-PCR to confirm that the cloned cDNAs are associated with peripheral blood circulating cells of cancer-bearing patients. Samples were collected from 12 PCa patients and 8 age- and race-matched healthy men for this analysis. RT-PCR was performed on individual samples. A total of 20 samples, 8 healthy men and 12 PCa patients, total RNA was analyzed for selected genes. Four target genes were selected to be confirmed by RT-PCR. Two genes, PCa-001 and PCa-002, were selected from the library that consists of mRNA species only present in circulating cells of PCa patients. PCR primers were designed according to the sequencing results (Table 2) and the designed primers were subjected to a BLAST search to ensure that these sequences do not match any identified mRNA with high homology. As expected, PCa-001 and PCa-002 were expressed at significantly higher levels in circulating cells obtained from PCa patients than in healthy men (Figure 2). The detection of low levels of PCa-001 and PCa-002 may due to high sensitivity of RT-PCR-based detection. Martin et al. also reported low levels of Mdm-2 and Gro-alpha expression in peripheral blood mononuclear cells of healthy samples using RT-PCR, whereas array-based analysis did not show a detectible signal for these two genes [31]. It is also possible that a very small number of epithelial cells can be present in the peripheral blood of healthy men [32] and PCa-001 and PCa-002 are epithelial cells markers. Another two genes, Nrml-001 and Nrml-002, were selected as absent in PCa patients' circulating cells. These two genes were expressed, as expected, at higher levels in healthy men than in PCa patients (Figure 2). β-actin has been shown to be constantly expressed in leukocytes of different individuals [33]. We also demonstrated that level β-actin expression is similar in all our samples (Figure 2). The relative levels of PCa-001, PCa-002, Nrml-001, and Nrml-002 expression were then normalized to β-actin and illustrated in Figure 3. Levels of these genes' expression were statistically different between healthy men and PCa patients.
Table 2 PCR primers and conditions for detecting levels of mRNA expression in circulating cells of healthy men and patients with PCa
Clone I.D. GenBank Accession No. PCR primers cycles
PCa-001 AC132068 5'-AGG AAT AAG TCA CAC CGT GGA-3' 5'-ACC TGT TGG GAC TAG ACG CAT-3' 20
nested 5'-TGG TCT GTA ACC CTT AGG AGA-3' 5'-TCT GCC CTT TGA GTC CAA GT-3' 25
Pca-002 AC019106 5'-AGG TCA GCA GAG ATG TCT GT-3' 5'-TAG TCC CCG AGA AAG AAT TA-3' 32
Nrml-001 AC004690 5'-TGA GCA GTT TCT TCA GCC TCA-3' 5'-TGA TAA GTC CAA CCC AAA GGC T-3' 20
nested 5'-TAT CTG GGT GAC ACT GGG AAA-3' 5'-AGA GAC CAG CGT AAT ATC CCT-3' 30
Nrml-002 AK095899 5'-AGG TAA AGG AAA CCC CAA CAT GCA-3' 5'-AAC CAA CGA GGT GGC CGA ATC TT-3' 35
Figure 2 Confirmation of differential gene expression in circulating cells of healthy men and PCa patients using semi-quantitative RT-PCR. RT-PCR was performed on individual samples from 8 healthy controls and 12 PCa patients to confirm the SSH results. After sequencing reaction to reveal the identities of a total of 23 clones present in the subtracted libraries, PCR primers were designed (Table 2). β-actin was also amplified from the same samples using a β-actin primer set (BD Bioscinces Clontech) to serve as an internal control for standardizing the quantity of the RNA applied in each reaction. After PCR amplification, aliquots (10 μl) of these PCR products were electrophoresed into 2% agarose gels followed by ethidium bromide staining.
Figure 3 Relative levels of target genes expression in peripheral blood circulating cells of healthy men and PCa patients. Images obtained from Figure 2 were captured and analyzed using the Quantity One® software. For each target gene, levels of gene expression were normalized to the level of β-actin expression for each individual sample. * indicates statistical significance between healthy men and PCa patients at p < 0.05.
In order to ensure that the PCR products were not resulted from genomic DNA contamination, we performed two pre-cautionary procedures. First, all total RNA samples were subjected to RNase-free, DNase I digestion to remove residual genomic DNA. Second, we noticed that, in reactions with reverse transcription, β-actin signal was detected after 26 cycles of PCR amplification (Figure 2), whereas β-actin signal was absent in reactions without reverse transcription even after 35 cycles of amplification using amplimers within the same exon (data not shown). However, before we can confirm our statement, full-length cDNAs corresponding to our identified sequences need to be cloned.
The genome-wide screening and identification of mRNA species associated with circulating cells of tumor-bearing patients has been described for various cancers. Twine et al. reported the use of the microarray approach to differentiate gene expression patterns of mononuclear cells in patients with advanced renal cell carcinoma [34]. However, due to the low number of tumor cells present in the circulation compared to the large number of normal leukocytes, hybridization array (cDNA array and oligonucleotide array) may not be an ideal tool for identifying rare molecular events occurring in small number of CTCs. To overcome this problem, Smirnov et al. used magnetic separation of epithelial cell adhesion molecule (EpCAM) expressing cells from peripheral blood circulating cells and compared gene expression patterns between EpCAM-enriched and EpCAM-depleted cells using microarray analysis in cancer-bearing patients [35]. Another comprehensive gene expression method, named mRNA differential display, has been conducted to compare genes that are expressed in peripheral blood mononuclear cells of tumor-free individuals with those from lung, breast, and colon cancer patients [36]. This study found a total of 21 mRNA species expressed in tumor patients' blood samples but not in samples from tumor free volunteers. In addition, Martin et al. reported the use of the differential display to first identify transcripts differentially expressed in breast cancer cells and normal breast epithelial cells followed by an array analysis of these transcripts in the circulating cells of breast cancer patients [31]. Results from these experiments demonstrate that the detection of disseminated cancer cells in peripheral blood is attainable.
It has been suggested that selective enrichment of the tumor cell population from both bone marrow and blood before analysis can increase the sensitivity for detecting CTCs [37-40]. Various cell separation techniques have been devised to enrich the CTC population from whole blood. However, these methods may also introduce artifacts into the sample preparation steps. For example, the addition of anticoagulants to blood samples affect leukocyte gene expression ex vivo [41-43]. Efforts were made to avoid RNA degradation and alteration in gene expression during in vitro processing of blood cells and to avoid under- and over-estimation of in vivo mRNA expression. We used direct isolation of poly(A)+ RNA and total RNA from whole blood circumventing a prior cell separation step. Our intention was also to prevent the loss of rare, un-identified target cells from blood samples during enrichment procedures since we do not know much about tumor cells' genotypes/phenotypes or any type(s) of immune cells' participation in cancer development.
Although it has been suggested that cancers are composed of a heterogeneous collection of cells with different degrees of tumor marker expressions [44,45], CTCs of all types might need to develop a "common" mechanism(s) to survive in the circulation and acquire metastatic capability. We speculate that universal "tumor-specific" markers can be identified in occult tumor cells from different cancers. Moreover, we expect to identify "tissue-specific" molecules in CTCs if tumor cells continue to express their tissue-specific markers. The identification of tissue-specific markers will help to identify the origins of CTCs. Emerging evidence also demonstrates that detection of tumor cells disseminated in peripheral blood can provide clinically important data for tumor staging, prognostication, and identification of surrogate markers for early assessment of the effectiveness of adjuvant therapy. Furthermore, by comparing gene expression profiling of all circulating cells, we expect to identify genes that might play a role in "immunosurveillance". Our future objectives include the identification of cell types that expressed differentially regulated mRNA species. We also intend to study the functional activities of these molecules in circulating cells during cancer development and establish an association between these genes' expression and cancer stages.
Conclusion
Using the PCR-based SSH technique, we established two comprehensive subtracted cDNA libraries consisting of potentially differentially regulated genes in circulating cells of PCa patients. We further confirmed that both elevated and suppressed transcripts can be detected in circulating cells of PCa patients. This is an initial attempt to perform genome-wide gene expression analysis in peripheral blood circulating cells and demonstrate the presence of previously un-identified mRNA species in circulating cells of cancer-bearing patients. This is the first step toward understanding tumor metastasis and tumor-induced immune reactions in the development of cancer. We will continue to investigate these molecules' physiological/pathological function and their use in cancer detection.
Methods
Patient selection
PCa patients were enrolled at the time of diagnosis of elevated PSA and positive biopsy. Healthy men's samples were collected from volunteers with similar age and race distribution without evidence of diseases or use of any medications. Attending physicians provided all participants with informed consent forms for collecting samples used in this study. Sample collection was also HIPAA compliant. Blood was drawn before scheduled surgery from PCa patients. There was no evidence of systemic metastases for all PCa patients when the primary tumor was resected through surgical prostatectomy. For initial construction of SSH libraries, we collected 50 samples, 25 healthy men and 25 patients with PCa. We collected an additional 20 blood samples, 8 healthy men and 12 patients with PCa, for RT-PCR analysis.
Blood collection and RNA isolation
For SSH, whole blood (5 ml) drawn from each individual was immediately mixed with 10x volume of RNA stabilization reagents for blood/bone marrow (Roche). The cells were then lysed. Poly(A)+ RNA was immediately isolated by a two-step procedure through magnetic separation using the mRNA isolation kit for blood/bone marrow (Roche). The poly(A)+ enriched samples were finally eluted from magnetic beads with H2O. Purified poly(A)+ RNA was quantitated spectrophotometrically and stored in liquid nitrogen until use.
For RT-PCR, blood (2.5 ml) from each individual was colleted into a PAXgene™ Blood RNA tube (QIAGEN) following the manufacturer's protocol. Whole blood was thoroughly mixed with PAXgene stabilization reagent and stored at room temperature for 6 hours prior to RNA extraction. Total RNA was then extracted using the PAXgene™ Blood RNA kit according to the manufacturer's directions (QIAGEN). As the resulting RNA was usually contaminated with genomic DNA [46], total RNA samples absorbed to the PAXgene™ Blood RNA System spin column were incubated with DNase I (QIAGEN) at 25°C for 20 min to remove genomic DNA. Total RNA was eluted, quantitated, and stored in liquid nitrogen.
Suppression subtractive hybridization (SSH) procedures
SSH was performed according to procedures described by Diatchenko et al. [25]. All reagents are now commercially available from BD Biosciences Clontech. Briefly, reverse transcription was performed with 2 μg poly(A)+ RNA from an individual patient sample in the presence of a mixture of three 3' anchored primers (5'-TTTGCATGCTCGAG-(T)25-A/G/C-3') at 42°C for 2 hours. Second strand cDNA was then synthesized with the addition of E. coli DNA polymerase I (250 μU/μl; Invitrogen), E. coli RNase H (8.5 μU/μl; Invitrogen), and E. coli DNA ligase (30 μU/μl; Invitrogen) at 16°C for an additional 2 hours. The double-stranded cDNA libraries were then pooled into healthy and PCa groups. The pooled samples were subjected to Rsa I digestion. To identify mRNA species expressed only in patients with PCa, the Rsa I digested pooled cDNAs derived from PCa were ligated to specially designed adapters A and B (BD Biosciences Clontech) in two different reactions [25].
To form heterohybrids between two sample populations, the adapter A and adaptor B ligated cDNAs (20 ng) were combined with excess Rsa I digested cDNAs (400 ng) from healthy men in two separate reactions, heat-denatured, and hybridized at 68°C for 10 hours. In a second hybridization step, the two separate samples from adapters A and B containing reactions were combined. A fresh aliquot of 150 ng heat-denatured Rsa I digested cDNAs derived from healthy men was added to the combined reaction. Hybridization was continued for another 10 hours at 68°C. Commonly expressed sequences between controls and PCa patients formed hybrids in these two sequential hybridization steps. The heterohybrids are less likely to be amplified in the following PCR step due to the design of SSH adaptors [25].
Genes specifically expressed in PCa patients' circulating cells were amplified by two consecutive rounds of PCR according to the procedures reported by Diatchenko et al. [25]. The PCR-amplified products were then ligated to the pCRII vector (Invitrogen) followed by transformation. The bacteria were plated on agar plates containing ampicillin and overlaid with X-gal and IPTG. After overnight incubation, white colonies were picked and used for subsequent sequencing reaction. Sequencing results were used to design PCR primer sets to determine the genes' expression levels in healthy controls and PCa patients.
To detect sequences present in circulating cells of healthy men but absent in circulating cells of PCa patients, the initial adaptors ligation reaction was reversed. Aliquots of Rsa I digested pooled cDNAs derived from healthy men were ligated to adapters A or B followed by hybridization and PCR amplification as described above.
RT-PCR confirmation
First strand cDNAs were reverse transcribed from 2.5 μg of the total RNA in the presence of oligo d(T) primer (Invitrogen), 20 μM each of dNTPs, and 200 units of M-MLV reverse transcriptase (Invitrogen). This was done in a total of 50 μl at 42°C for 2 hours. PCR reactions were performed by mixing 1 μl of first-strand cDNAs, 0.2 μM gene-specific 5' and 3' primers (Table 2), and 5 units Taq DNA polymerase (Invitrogen) in a total of 50 μl. Reactions were performed by heat activation at 94°C for 2 min, followed by cycling through 94°C for 30 sec, 50–55°C for 1 min, and 72°C for 1 min. The minimal numbers of PCR cycles required for detecting these gene products were first determined and is indicated in Table 2. β-actin (NM_001101) was also amplified and used as an internal control for comparing relative levels of target gene expression. We also included RNA samples without reverse transcription for β-actin amplification to determine levels of genomic DNA contamination. Following gel electrophoresis, images were captured using a Bio-Rad Gel Doc system; and band intensities were analyzed by the Quantity One® software (Bio-Rad).
Statistical Analysis
Levels of target gene expression were expressed as mean ± standard deviation (SD) following normalization to β-actin. A Student's t test was used to compare means of these genes expressions between the healthy controls and PCa patients. A probability value of p < 0.05 was considered significant.
List of abbreviations
CTC, circulating tumor cell
SSH, suppression suppressive hybridization
PCa, prostate cancer.
Authors' contributions
XL and HKL conducted sample preparations, subtracted library construction, transcript abundance analysis, and data analysis. CW, RM, GS, AM, JK, and CSM participated in patient selection, patient enrollment, and sample collection. DJC, BPK, and HKL participated in study design, data interpretation, and manuscript preparation.
Acknowledgements
This work was supported by the Oklahoma Center for the Advancement of Science and Technology (OCAST) grant HR02-081R to HKL. We also thank Dr. Ying Zhang for her statistical consultation.
==== Refs
Gilbey AM Burnett D Coleman RE Holen I The detection of circulating breast cancer cells in blood J Clin Pathol 2004 57 903 911 15333648 10.1136/jcp.2003.013755
Pantel K Riethmuller G Micrometastasis detection and treatment with monoclonal antibodies Curr Top Microbiol Immunol 1996 213 1 18 8814999
Dearnaley DP Sloane JP Ormerod MG Steele K Coombes RC Clink HM Powles TJ Ford HT Gazet JC Neville AM Increased detection of mammary carcinoma cells in marrow smears using antisera to epithelial membrane antigen Br J Cancer 1981 44 85 90 7020739
Felton T Harris GC Pinder SE Snead DR Carter GI Bell JA Haines A Kollias J Robertson JF Elston CW Ellis IO Identification of carcinoma cells in peripheral blood samples of patients with advanced breast carcinoma using RT-PCR amplification of CK7 and MUC1 Breast 2004 13 35 41 14759714 10.1016/S0960-9776(03)00126-7
Fujii Y Kageyama Y Kawakami S Kihara K Oshima H Detection of disseminated urothelial cancer cells in peripheral venous blood by a cytokeratin 20-specific nested reverse transcriptase-polymerase chain reaction Jpn J Cancer Res 1999 90 753 757 10470288
Soeth E Vogel I Roder C Juhl H Marxsen J Kruger U Henne-Bruns D Kremer B Kalthoff H Comparative analysis of bone marrow and venous blood isolates from gastrointestinal cancer patients for the detection of disseminated tumor cells using reverse transcription PCR Cancer Res 1997 57 3106 3110 9242433
Jotsuka T Okumura Y Nakano S Nitta H Sato T Miyachi M Suzumura K Yamashita J Persistent evidence of circulating tumor cells detected by means of RT-PCR for CEA mRNA predicts early relapse: a prospective study in node-negative breast cancer Surgery 2004 135 419 426 15041966 10.1016/j.surg.2003.08.014
Gerhard M Juhl H Kalthoff H Schreiber HW Wagener C Neumaier M Specific detection of carcinoembryonic antigen-expressing tumor cells in bone marrow aspirates by polymerase chain reaction J Clin Oncol 1994 12 725 729 7512130
Hildebrandt M Mapara MY Korner IJ Bargou RC Moldenhauer G Dorken B Reverse transcriptase-polymerase chain reaction (RT-PCR)-controlled immunomagnetic purging of breast cancer cells using the magnetic cell separation (MACS) system: a sensitive method for monitoring purging efficiency Exp Hematol 1997 25 57 65 8989908
Ady N Morat L Fizazi K Soria JC Mathieu MC Prapotnich D Sabatier L Chauveinc L Detection of HER-2/neu-positive circulating epithelial cells in prostate cancer patients Br J Cancer 2004 90 443 448 14735191 10.1038/sj.bjc.6601532
Noguchi S Aihara T Motomura K Inaji H Imaoka S Koyama H Detection of breast cancer micrometastases in axillary lymph nodes by means of reverse transcriptase-polymerase chain reaction. Comparison between MUC1 mRNA and keratin 19 mRNA amplification Am J Pathol 1996 148 649 656 8579127
Hoon DS Sarantou T Doi F Chi DD Kuo C Conrad AJ Schmid P Turner R Guiliano A Detection of metastatic breast cancer by b-hCG polymerase chain reaction Int J Cancer 1996 69 369 374 8900369 10.1002/(SICI)1097-0215(19961021)69:5<369::AID-IJC3>3.0.CO;2-3
Matsumura M Niwa Y Kato N Komatsu Y Shiina S Kawabe T Kawase T Toyoshima H Ihori M Shiratori Y Detection of a-fetoprotein mRNA, an indicator of hematogenous spreading hepatocellular carcinoma, in the circulation: a possible predictor of metastatic hepatocellular carcinoma Hepatology 1994 20 1418 1425 7527002
Shariat SF Kattan MW Song W Bernard D Gottenger E Wheeler TM Slawin KM Early postoperative peripheral blood reverse transcription PCR assay for prostate-specific antigen is associated with prostate cancer progression in patients undergoing radical prostatectomy Cancer Res 2003 63 5874 5878 14522912
Ghossein RA Scher HI Gerald WL Kelly WK Curley T Amsterdam A Zhang ZF Rosai J Detection of circulating tumor cells in patients with localized and metastatic prostatic carcinoma: clinical implications J Clin Oncol 1995 13 1195 1200 7537803
Seiden MV Kantoff PW Krithivas K Propert K Bryant M Haltom E Gaynes L Kaplan I Bubley G DeWolf W Sklar J Detection of circulating tumor cells in men with localized prostate cancer J Clin Oncol 1994 12 2634 2639 7527455
Schmidt B Anastasiadis AG Seifert HH Franke KH Oya M Ackermann R Detection of circulating prostate cells during radical prostatectomy by standardized PSMA RT-PCR: association with positive lymph nodes and high malignant grade Anticancer Res 2003 23 3991 3999 14666708
Wang ZP Eisenberger MA Carducci MA Partin AW Scher HI Ts'o PO Identification and characterization of circulating prostate carcinoma cells Cancer 2000 88 2787 2795 10870062 10.1002/1097-0142(20000615)88:12<2787::AID-CNCR18>3.0.CO;2-2
Ylikoski A Pettersson K Nurmi J Irjala K Karp M Lilja H Lovgren T Nurmi M Simultaneous quantification of prostate-specific antigen and human glandular kallikrein 2 mRNA in blood samples from patients with prostate cancer and benign disease Clin Chem 2002 48 1265 1271 12142383
O'Hara SM Moreno JG Zweitzig DR Gross S Gomella LG Terstappen LW Multigene reverse transcription-PCR profiling of circulating tumor cells in hormone-refractory prostate cancer Clin Chem 2004 50 826 835 14988224 10.1373/clinchem.2003.028563
Raj GV Moreno JG Gomella LG Utilization of polymerase chain reaction technology in the detection of solid tumors Cancer 1998 82 1419 1442 9554517 10.1002/(SICI)1097-0142(19980415)82:8<1419::AID-CNCR1>3.0.CO;2-4
Ghossein RA Bhattacharya S Rosai J Molecular detection of micrometastases and circulating tumor cells in solid tumors Clin Cancer Res 1999 5 1950 1960 10473071
Johnson PW Burchill SA Selby PJ The molecular detection of circulating tumour cells Br J Cancer 1995 72 268 276 7640205
Dunn GP Bruce AT Ikeda H Old LJ Schreiber RD Cancer immunoediting: from immunosurveillance to tumor escape Nat Immunol 2002 3 991 998 12407406 10.1038/ni1102-991
Diatchenko L Lau YF Campbell AP Chenchik A Moqadam E Huang B Lukyanov S Lukyanov K Gurskaya N Severdlov ED Siebert P Suppression subtractive hybridization: a method for generating differentially regulated or tissue-specific cDNA probes and libraries Proc Natl Acad Sci USA 1996 93 6025 6030 8650213 10.1073/pnas.93.12.6025
Pantel K Cote RJ Fodstad Detection and clinical importance of micrometastatic disease J Natl Cancer Inst 1999 91 1113 1124 10393719 10.1093/jnci/91.13.1113
Nakano O Sato M Naito Y Suzuki K Orikasa S Aizawa M Suzuki Y Shintaku I Nagura H Ohtani H Proliferative activity of intratumoral CD8(+) T-lymphocytes as a prognostic factor in human renal cell carcinoma: clinicopathologic demonstration of antitumor immunity Cancer Res 2001 61 5132 5136 11431351
Girardi M Oppenheim DE Steele CR Lewis JM Glusac E Filler R Hobby P Sutton B Tigelaar RE Hayday AC Regulation of cutaneous malignancy by gd T cells Science 2001 294 605 609 11567106 10.1126/science.1063916
Kelly JM Darcy PK Markby JL Godfrey DI Takeda K Yagita H Smyth MJ Induction of tumor-specific T cell memory by NK cell-mediated tumor rejection Nat Immunol 2002 3 83 90 11743585 10.1038/ni746
Diefenbach A Hsia JK Hsiung MY Raulet DH A novel ligand for the NKG2D receptor activates NK cells and macrophages and induces tumor immunity Eur J Immunol 2003 33 381 391 12645935 10.1002/immu.200310012
Martin KJ Graner E Li Y Price LM Kritzman BM Fournier MV Rhei E Pardee AB High-sensitivity array analysis of gene expression for the early detection of disseminated breast tumor cells in peripheral blood Proc Natl Acad Sci USA 2001 98 2646 2651 11226293 10.1073/pnas.041622398
Garber K Trial results boost circulating tumor cell field J Natl Cancer Inst 2004 96 1055 1057 15265962
Vandesompele J De Preter K Pattyn F Poppe B Van Roy N De Paepe A Speleman F Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes Genome Biol 2002 3 RESEARCH0034 12184808 10.1186/gb-2002-3-7-research0034
Twine NC Stover JA Marshall B Dukart G Hidalgo M Stadler W Logan T Dutcher J Hudes G Dorner AJ Slonim DK Trepicchio WL Burczynski ME Disease-associated expression profiles in peripheral blood mononuclear cells from patients with advanced renal cell carcinoma Cancer Res 2003 63 6069 6075 14522937
Smirnov DA Zweitzig DR Foulk BW Miller MC Doyle GV Pienta KJ Meropol NJ Weiner LM Cohen SJ Moreno JG Connelly MC Terstappen LW O'Hara SM Global gene expression profiling of circulating tumor cells Cancer Res 2005 65 4993 4997 15958538
Fournier MV Carvalho MG Pardee AB A strategy to identify genes associated with circulating solid tumor cell survival in peripheral blood Mol Med 1999 5 313 319 10390547
Racila E Euhus D Weiss AJ Rao C McConnell J Terstappen LW Uhr JW Detection and characterization of carcinoma cells in the blood Proc Natl Acad Sci USA 2004 95 4589 4594 9539782 10.1073/pnas.95.8.4589
Gertler R Rosenberg R Fuehrer K Dahm M Nekarda H Siewert JR Detection of circulating tumor cells in blood using an optimized density gradient centrifugation Recent Restuls Cancer Res 2003 162 149 155
Rosenberg R Gertler R Friederichs J Fuehrer K Dahm M Phelps R Thorban S Nekarda H Siewert JR Comparison of two density gradient centrifugation systems for the enrichment of disseminated tumor cells in blood Cytometry 2002 49 150 158 12454978 10.1002/cyto.10161
Naume B Borgen E Nesland JM Beiske K Gilen E Renolen A Ravnas G Qvist H Karesen R Kvalheim G Increased sensitivity for detection of micrometastases in bone-marrow/peripheral-blood stem-cell products from breast-cancer patients by negative immunomagnetic separation Int J Cancer 1998 78 556 560 9808522 10.1002/(SICI)1097-0215(19981123)78:5<556::AID-IJC5>3.0.CO;2-G
Riches P Gooding R Millar BC Rowbottom AW Influence of collection and separation of blood samples on plasma IL-1, IL-6 and TNF-a concentrations J Immunol Methods 1992 153 125 131 1517582 10.1016/0022-1759(92)90314-J
Freeman R Wheeler J Robertson H Paes ML Laidler J In-vitro production of TNF-a in blood samples Lancet 1990 336 312 313 1973999 10.1016/0140-6736(90)91848-5
Rainen L Oelmueller U Jurgensen S Wyrich R Ballas C Schram J Herdman C Bankaitis-Davis D Nicholls N Trollinger D Tryon V Stabilization of mRNA expression in whole blood samples Clin Chem 2002 48 1883 1890 12406972
Baker M Gillanders WE Mikhitarian K Mitas M Cole DJ The molecular detection of micrometastatic breast cancer Am J Surg 2003 186 351 358 14553849 10.1016/S0002-9610(03)00262-9
Braun S Hepp F Sommer HL Pantel K Tumor-antigen heterogeneity of disseminated breast cancer cells: implications for immunotherapy of minimal residual disease Int J Cancer 1999 84 1 5 9988223 10.1002/(SICI)1097-0215(19990219)84:1<1::AID-IJC1>3.0.CO;2-A
Breit S Nees M Schaefer U Pfoersich M Hagemeier C Muckenthaler M Kulozik AE Impact of pre-analytical handling on bone marrow mRNA gene expression Br J Haematol 2004 126 231 243 15238145 10.1111/j.1365-2141.2004.05017.x
|
16086835
|
PMC1199617
|
CC BY
|
2021-01-04 16:36:34
|
no
|
Mol Cancer. 2005 Aug 8; 4:30
|
utf-8
|
Mol Cancer
| 2,005 |
10.1186/1476-4598-4-30
|
oa_comm
|
==== Front
RetrovirologyRetrovirology1742-4690BioMed Central London 1742-4690-2-461604278710.1186/1742-4690-2-46Short ReportPDZ domain-binding motif of human T-cell leukemia virus type 1 Tax oncoprotein is essential for the interleukin 2 independent growth induction of a T-cell line Tsubata Chikako [email protected] Masaya [email protected] Masahiko [email protected] Masayasu [email protected] Yuetsu [email protected] Fumitake [email protected] Masahiro [email protected] Division of Virology, Niigata University Graduate School of Medical and Dental Sciences, 1-757 Asahimachi-Dori, Niigata 951-8510, Japan2 Division of Clinical Nephrology and Rheumatology, Niigata University Graduate School of Medical and Dental Sciences, 1-757 Asahimachi-Dori, Niigata 951-8510, Japan3 Department of Infectious Disease and Immunology, Okinawa-Asia Research Center of Medical Science, Faculty of Medicine, University of the Ryukyus, Okinawa, Japan2005 23 7 2005 2 46 46 20 7 2005 23 7 2005 Copyright © 2005 Tsubata et al; licensee BioMed Central Ltd.2005Tsubata et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Human T-cell leukemia virus type 1 (HTLV-1) is the etiologic agent of adult T-cell leukemia (ATL), whereas HTLV type 2 (HTLV-2), is not associated with ATL or any other leukemia. HTLV-1 encodes the transforming gene tax1, whose expression in an interleukin (IL)-2-dependent T-cell line (CTLL-2) induces IL-2-independent growth.
Results
In this study, we demonstrated that IL-2-independent growth induction by Tax1 was abrogated by mutations of the PDZ domain-binding motif (PBM) at the Tax1 C-terminus. HTLV-2 Tax2, which shares 75% amino acid identity with Tax1 but does not have a PBM, was not able to induce IL-2-independent growth of CTLL-2.
Conclusion
Our results suggest that Tax1, through interaction with PDZ domain protein(s) induces IL-2-independent growth, which may be a factor in multi-step leukemogenesis caused by HTLV-1.
==== Body
Findings
Adult T-cell leukemia (ATL) is an extremely aggressive T-cell leukemia [1,2], and it is characterized by malignant expansion of CD4 positive T-cells infected with human T-cell leukemia virus type 1 (HTLV-1). HTLV-1 is an onco-retrovirus, which immortalizes human CD4 T-cells in vitro [3,4]. Such an immortalization event is, however, not sufficient for ATL development, since a minority of HTLV-1-infected individuals (~5%) suffer ATL 60 years on average after the infection [2,5,6]. Accumulating evidence suggests that genetic and epigenetic changes in HTLV-1-infected T-cells and deterioration of host immune activities are prerequisites for ATL development [2]. HTLV type 2 (HTLV-2) is molecularly and biologically similar to HTLV-1 [7,8]. HTLV-2 also immortalizes primary human T-cells with equivalent efficiency to HTLV-1, although HTLV-2 preferentially immortalizes CD8 T-cells [9]. Regardless of such similarities, HTLV-2 is not associated with ATL or any other leukemia [10]. Thus, HTLV-2 can not promote multi-step leukemogenesis. However, the underlying mechanism by which HTLV-1 promotes multi-step leukemogenesis has not yet been elucidated.
HTLV-1 and HTLV-2 encode functionally and structurally similar proteins, Tax1 and Tax2, respectively [7,11,12], and they are candidate factors responsible for distinct pathogenic activities of the two viruses. Tax1 and Tax2 were originally identified as transcriptional activators of their own gene expression [11,12]. Later they were shown to play crucial roles in the immortalization of T-cells [13,14]. Tax1 by itself immortalizes primary human T-cells in an interleukin (IL)-2-dependent manner [15,16]. Tax1 inhibits several modes of apoptosis [17], and stimulates the cell cycle progression in primary T-cells as well as in T-cell lines [18,19]. In addition, in transgenic animals Tax1 induces various malignancies such as fibrosarcoma and natural killer cell leukemia [20,21]. Consistent with the above activities, recombinant HTLV-1 and HTLV-2 carrying inactive tax1 and tax2 genes, respectively, cannot transform primary human T-cells [13,14]. Evidence suggests that the activation of cellular genes by Tax1 is essential for T-cell immortalization [22]. For instance, Tax1 activates the expression of genes encoding cytokines, cytokine receptors, chemokines, cell cycle regulators and anti-apoptotic factors [22-31]. Tax1 and Tax2 generally activate the same sets of cellular genes with equivalent efficiency, although some differences have been reported.
We previously found that Tax1 and Tax2 transform a rat fibroblast cell line (Rat-1) to induce colonies in soft agar (CFSA, colony formation in soft agar), and the activity of Tax1 is greater than that of Tax2 [32]. The experiments using their chimeric proteins indicated that the PDZ domain-binding motif (PBM) located at the Tax1 C-terminus, S/TXV (S/T, serine or threonine; X, any amino acid; V, valine), is responsible for the high CFSA activity relative to Tax2 [33]. Through this motif, Tax1 but not Tax2 was found to bind to PDZ domain-containing proteins, including Dlg, a mammalian homologue of Drosophila discs large tumor suppressor [33-36]. These results present an attractive hypothesis that PBM is a factor responsible for the distinct pathogenic activities of HTLV-1 and HTLV-2. Since HTLV-1 is a T-cell-tropic virus, in this study, we examined the activity of Tax1 PBM in T-cells. To do this, we used several mutant genes that had been previously characterized (Figure 1 and Table 1) [33]. The TaxΔC gene contains a C-terminal four-amino acid deletion abrogating PBM in Tax1. Tax351A and Tax353A are substitution mutants of Tax1 PBM, at amino acids 351 and 353 in Tax1, respectively. These three PBM mutants did not interact with PDZ domain-containing proteins such as Dlg and MAGI-3 [33,36]. Tax2B+C is a chimeric Tax2B gene with a wild type Tax1 PBM peptide, and Tax2B+C but not Tax2B interacts with Dlg. These genes in pβA-IRES-puro plasmid (pβAIP) were transfected into an IL-2 dependent mouse T-cell line (CTLL-2), and the cells were then selected by puromycin in the presence of IL-2. Western blotting analysis using anti-Tax1 antibody showed that two independent TaxΔC clones (TaxΔC-7, TaxΔC-21) and two independent Tax1 clones (Tax1-12, Tax1-24) expressed TaxΔC and Tax1 protein, respectively. The amounts of mutant Tax1 protein relative to wild type protein were generally equivalent, and TaxΔC-21 cells expressed the Tax protein higher than Tax1-24 cells (Figure 2). These characterized cells were then cultured in the absence of IL-2 for 3–5 days. CTLL-2 cells transfected with a control plasmid (pβAIP) did not grow in the absence of IL-2, and most of the cells died approximately 3 days after IL-2 withdrawal, whereas two CTLL-2 clones transfected with wild-type tax1 plasmid continued to grow for 5 days. This was consistent with previous findings that stable Tax1 expression in CTLL-2 conferred a permanent IL-2-independent growth phenotype [37]. On the other hand, two CTLL-2 clones transfected with taxΔC did not grow in the absence of IL-2, and were close to cell death approximately 2 days after IL-2 withdrawal. In addition, two PBM mutant clones, CTLL-2/Tax351A and CTLL-2/Tax353A, did not grow in the absence of IL-2. These results show that Tax1 PBM is essential for IL-2-independent growth induction of CTLL-2 cells. Unexpectedly, in spite of three independent trials, we could not establish CTLL-2 cells expressing Tax2B or Tax2B+C even in the presence of IL-2. Although anti-Tax1 and anti-Tax2 antibodies did not detect Tax2B and Tax1, respectively, previous study using a chimeric Tax1 with Tax2B showed that the amounts of Tax1 and Tax2B proteins were expressed equivalently in the cells [32]. Thus, these results suggested that Tax2B has a toxic effect in CTLL-2 cells.
Figure 1 Structure of Tax1, Tax2B and their mutant proteins. The amino acid sequence of PBM and its mutants are indicated. The tax1, tax2B genes and their mutant genes inserted in the pHβ Pr-1-neo expression plasmid have been described previously [33]. To convert the expression plasmid from pHβPr-1-neo to pβA-IRESpuro plasmid (pβAIP), an EcoRI-BamHI fragment containing the β-actin promoter of pHβPr-1-neo was inserted into the NruI-BamHI site of pIRESpuro3 (BD Biosciences) by a blunt-end ligation. Then, wild type tax or tax mutant cDNAs were inserted into the BamHI site of pβAIP.
Table 1 Characterization of Tax1, Tax2B, and their mutants
Tax Tax△C Tax351A Tax353A Tax2B Tax2B+C
IL-2 independent proliferation of CTLL-2 + - - - - -
% Outgrowth 30–40 3–5 2–5 1–4 0 0
Dlg binding# ++ - - - - ++
CFSA of Rat-1# +++ + + + + +++
#The results from reference 33.
Figure 2 PBM is essential for IL-2-independent growth of CTLL-2 cells induced by Tax1. (A) CTLL-2 is a mouse T-cell line. This cell line was cultured in RPMI1640 medium supplemented with 10% heat-inactivated fetal bovine serum (RPMI-FBS), antibiotics and 0.5 nM recombinant human IL-2. To establish CTLL-2 cell lines expressing Tax or Tax mutant proteins, CTLL-2 cells (1 × 107) were suspended in 400 μl Opti MEM1 (Gibco BRL, Gaithersburg, MD), mixed with 20 μg of the vector plasmid (pβAIP) or with expression plasmids encoding Tax1 or Tax mutants, and then pulsed at 200 V and 975 F. The cells were seeded in 96 well plates 24 h after electroporation and cultured in RPMI-FBS containing 0.5 nM IL-2 and 2 μg/ml puromycin for 4 to 6 weeks. Puromycin-resistant cells were screened for the expression of Tax protein by Western blot analysis using mouse anti-Tax1 monoclonal antibody (TAXY-7) [42] as described previously [33]. (B) CTLL-2/Vector, and two of each CTLL-2 clone expressing Tax1 or Tax mutant proteins were washed twice with phosphate-buffered saline (PBS) and cultured in IL-2-free medium for 3–5 days. Cell growth was measured by the trypan blue staining method using light microscopy.
To confirm the role of PBM in IL-2 independent growth of CTLL-2, we transfected tax1, tax2B or their mutant plasmids into CTLL-2 cells, and the cells were seeded onto 96-well plates at a density of 1 × 104 cells/well, and cultured without IL-2 for three weeks. CTLL-2 cells transfected with the tax1 plasmid induced visible cell colonies in about 40% of the wells. This was due to the expression of Tax1 protein, since such cell growth was not observed in any wells containing CTLL-2 transfected with the empty vector plasmid. CTLL-2 cells transfected with three Tax1 PBM mutants also induced IL-2-independent cell growth, but the number of positive wells and the number of cells in each well were much lower than those of CTLL-2/Tax1 (Figure 3B, data not shown). Moreover, while CTLL-2/Tax1 cells continued to grow in the absence of IL-2 for at least two months, none of CTLL-2/TaxΔC cells grew any further (data not shown). This weakened activity of Tax1 PBM mutants was not due to their expression level in CTLL-2 cells, since Western blot analysis using anti-Tax1 antibodies detected equivalent expression relative to Tax1 after transfection (Figure 3A). These results indicate that TaxΔC still has IL-2-independent growth inducing activity in CTLL-2 cells, but the activity is much less than that of Tax1. Both Tax2B and Tax2B+C are completely devoid of such activity, further indicating that Tax2B does not have IL-2-independent growth inducing activity in CTLL-2 cells.
Figure 3 PBM is essential for outgrowth of CTLL-2/Tax cells in the absence of IL-2. (A, B) CTLL-2 cells (107) were transfected either with the vector plasmid (pHβPr-1-neo) or with expression plasmids encoding Tax1, Tax2B or their mutants by electroporation. The cells were divided into two groups 24 h after transfection. From the first group, living cells were collected using Ficoll-Paque Plus (Amersham Biosciences) and used for Western blot analysis (A) using anti-Tax1 antibody (TAXY7) or anti-Tax2B polyclonal antibody [43]. The second group (B) was seeded into 96 well plates and cultured in RPMI-FBS without IL-2 for 3 weeks, and the number of IL-2-independent colonies was counted under light microscopy. The percentage of positive wells indicates the proportion of the wells containing outgrowth of CTLL-2 cells. The data relates to two independent experiments with each duplicated transfection.
HTLV-1 Tax1 oncoprotein changes the cell growth of CTLL-2 from being IL-2-dependent to being IL-2-independent [37]. In this study, we showed that the PBM of Tax1 is essential for this activity in CTLL-2. Unlike Tax1, HTLV-2 Tax2 did not induce IL-2-independent growth, consistent with the absence of PBM in Tax2. Taken together with the strict conservation of PBM only in HTLV-1 Tax1 [33], these results suggest that HTLV-1 and HTLV-2 infection have distinct activity to growth of infected T-cells, and such a difference may be a factor responsible for ATL development.
Tax1, but not Tax2, interacts with the PDZ domain containing proteins Dlg and MAGI-3 [33,36]. Cotransfection and immunoprecipitation experiments showed that the three Tax1 mutants used here are severely defective in interaction with both Dlg and MAGI-3 proteins (Table 1). Dlg is highly expressed in T-cells including HTLV-1-infected T-cell lines [33,35], whereas MAGI-3 was detected only by reverse-transcription polymerase chain reaction analysis [36]. Since Dlg is a tumor suppressor gene product in Drosophila, it is an attractive candidate to play a role in IL-2-independent growth induction in CTLL-2 cells. It should be noted that there are many PDZ domain-containing genes in human. Thus, it is important to consider such proteins as candidates to mediate Tax1 activity in HTLV-1-infected T-cells.
We recently showed that Tax2, through the activation of transcription factor NFAT, constitutively induces the expression of IL-2, and the induced IL-2 promotes the cell growth of HTLV-2-infected T-cell lines, whereas such autocrine growth stimulation was not detected in HTLV-1-infected T-cell lines [38]. Tax2, however, did not induce IL-2-independent growth of CTLL-2 cells. In addition to Tax2B, Tax2B+C also failed to induce IL-2-independent growth of CTLL-2 cells. Tax2B+C, but not Tax2, transforms Rat-1 cells (CFSA) to the same extent as Tax1, and interacts with Dlg and MAGI-3 (Table 1) [33,36]. Thus, the binding of Tax2B+C to PDZ domain-containing proteins is not sufficient to induce IL-2-independent growth. Although it is unclear why we could not detect the activity of Tax2 to induce IL-2-independent growth, one possibility is that NFAT activation by Tax2 may induce the expression of pro-apoptotic genes such as Fas ligand, which may induce apoptosis of CTLL-2 cells, thereby masking the growth-promoting effect of IL-2.
Tax1 PBM plays crucial roles in the growth promoting activities in two different cell backgrounds; IL-2-independent growth induction of a T-cell line and transformation (CFSA) of a Rat-1 fibroblast cell line (Table 1), but it is unclear whether these two activities utilize the same mechanism. Both the number and size of the transformed colonies of Rat-1/Tax1 cells were greater than those of Rat-1/Tax2B cells, but the presence of Tax1 PBM was only correlated with the number but not the colony size [33]. These results suggest that the Tax1 PBM may have a selective role in the initiation of anchorage-independent growth of Rat-1 cells in soft agar but not the subsequent growth speed. Similarly, Tax1 PBM might be required for the initial cell growth of CTLL-2 deprived from IL-2, but not the subsequent rate of growth. Further analysis is required to solve this interesting question.
Several tumor viruses have both high-risk and low-risk subtypes. High-risk viruses induce malignancies such as cancers or leukemia in the host, whereas low-risk viruses induce benign tumors or lymphoproliferative diseases. Human papilloma virus (HPV) is such a virus, and only high-risk subtypes are associated with cervical cancers. Interestingly, an E6 oncoprotein of high-risk HPVs also contains PBM, and the motif is associated with high level of transforming activities measured by CFSA or focus formation of fibroblast cell lines in vitro [39]. Moreover, while E6 induces tumors in transgenic mice, deletion of the E6 PBM abrogates such activity [40]. Thus, PBM is a common determinant for high-risk oncoviruses, thereby being a useful tool for elucidating the molecular mechanism of malignant conversion of virus-infected cells.
Several inhibitors of transcription factor NF-κB induced apoptosis in HTLV-1-infected T-cell lines [41]. In addition, activation of NF-κB by Tax1 was well correlated with the induction of IL-2-independent growth of CTLL-2 [37]. However, NF-κB does not account for cell death of CTLL-2/TaxΔC cells in the absence of IL-2, since TaxΔC has equivalent NF-κB activity to Tax1 [36]. Taken together, the present results suggest that Tax1 PBM cooperates with NF-κB to induce IL-2-independent growth of HTLV-1-infected cells.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
CT, MH and MT carried out the establishing the cell lines and the functional analysis of the cell lines. MO, YT, FG and MF participated in the experimental design, data interpretation, and writing of the manuscript.
Acknowledgements
We thank William W Hall for donating the Tax2B plasmid and the antibody against Tax2B. We thank the Takeda pharmaceutical company for providing recombinant human IL-2. We also thank Sayoko Takizawa and Chika Yamamoto for the excellent technical assistance. This work was supported in part by a Grant-in-Aid for Scientific Research on Priority Areas and a Grant-in-Aid for Scientific Research (C) of Japan and a Grant for Promotion of Niigata University Research Projects.
==== Refs
Uchiyama T Yodoi J Sagawa K Takatsuki K Uchino H Adult T-cell leukemia: clinical and hematologic features of 16 cases Blood 1977 50 481 492 301762
Matsuoka M Human T-cell leukemia virus type I and adult T-cell leukemia Oncogene 2003 22 5131 5140 12910250 10.1038/sj.onc.1206551
Miyoshi I Kubonishi I Yoshimoto S Akagi T Ohtsuki Y Shiraishi Y Nagata K Hinuma Y Type C virus particles in a cord T-cell line derived by co-cultivating normal human cord leukocytes and human leukaemic T cells Nature 1981 294 770 771 6275274 10.1038/294770a0
Yamamoto N Okada M Koyanagi Y Kannagi M Hinuma Y Transformation of human leukocytes by cocultivation with an adult T cell leukemia virus producer cell line Science 1982 217 737 739 6980467
Poiesz BJ Ruscetti FW Gazdar AF Bunn PA Minna JD Gallo RC Detection and isolation of type C retrovirus particles from fresh and cultured lymphocytes of a patient with cutaneous T-cell lymphoma Proc Natl Acad Sci U S A 1980 77 7415 7419 6261256
Hinuma Y Nagata K Hanaoka M Nakai M Matsumoto T Kinoshita KI Shirakawa S Miyoshi I Adult T-cell leukemia: antigen in an ATL cell line and detection of antibodies to the antigen in human sera Proc Natl Acad Sci U S A 1981 78 6476 6480 7031654
Seiki M Hattori S Hirayama Y Yoshida M Human adult T-cell leukemia virus: Complete nucleotide sequence of the provirus genome integrated in leukemia cell DNA Proc Natl Acad Sci U S A 1983 80 3618 3622 6304725
Chen IS McLaughlin J Gasson JC Clark SC Golde DW Molecular characterization of genome of a novel human T-cell leukaemia virus Nature 1983 305 502 505 6312323 10.1038/305502a0
Chen IS Quan SG Golde DW Human T-cell leukemia virus type II transforms normal human lymphocytes Proc Natl Acad Sci U S A 1983 80 7006 7009 6316341
Sugamura K Hinuma Y Levy JA Human retroviruses:HTLV-I and HTLV-II The Retroviridae 1993 2 New York: Plenum Press 399 435
Chen IS Slamon DJ Rosenblatt JD Shah NP Quan SG Wachsman W The x gene is essential for HTLV replication Science 1985 229 54 58 2990037
Seiki M Hattori S Hirayama Y Yoshida M Direct evidence that p40x of human T-cell leukemia virus type I is a trans- acting transcriptional activator EMBO J 1986 5 561 565 3011413
Ross TM Pettiford SM Green PL The tax gene of human T-cell leukemia virus type 2 is essential for transformation of human T lymphocytes J Virol 1996 70 5194 5202 8764028
Robek MD Ratner L Immortalization of CD4(+) and CD8(+) T lymphocytes by human T-cell leukemia virus type 1 Tax mutants expressed in a functional molecular clone J Virol 1999 73 4856 4865 10233947
Akagi T Shimotohno K Proliferative response of Tax1-transduced primary human T cells to anti-CD3 antibody stimulation by an interleukin-2-independent pathway J Virol 1993 67 1211 1217 8437212
Grassmann R Berchtold S Radant I Alt M Fleckenstein B Sodroski JG Haseltine WA Ramstedt U Role of human T-cell leukemia virus type 1 X region proteins in immortalization of primary human lymphocytes in culture J Virol 1992 66 4570 4575 1351105
Copeland KF Haaksma AG Goudsmit J Krammer PH Heeney JL Inhibition of apoptosis in T cells expressing human T cell leukemia virus type I Tax AIDS Res Hum Retroviruses 1994 10 1259 1268 7531462
Low KG Dorner LF Fernando DB Grossman J Jeang KT Comb MJ Human T-cell leukemia virus type 1 Tax releases cell cycle arrest induced by p16INK4a J Virol 1997 71 1956 1962 9032327
Ohtani K Iwanaga R Arai M Huang Y Matsumura Y Nakamura M Cell type-specific E2F activation and cell cycle progression induced by the oncogene product Tax of human T-cell leukemia virus type I J Biol Chem 2000 275 11154 11163 10753922 10.1074/jbc.275.15.11154
Nerenberg M Hinrichs SH Reynolds RK Khoury G Jay G The tat gene of human T-lymphotropic virus type 1 induces mesenchymal tumors in transgenic mice Science 1987 237 1324 1329 2888190
Grossman WJ Kimata JT Wong FH Zutter M Ley TJ Ratner L Development of leukemia in mice transgenic for the tax gene of human T-cell leukemia virus type I Proc Natl Acad Sci U S A 1995 92 1057 1061 7862633
Azran I Schavinsky-Khrapunsky Y Aboud M Role of Tax protein in human T-cell leukemia virus type-I leukemogenicity Retrovirology 2004 1 20 15310405 10.1186/1742-4690-1-20
Inoue J Seiki M Taniguchi T Tsuru S Yoshida M Induction of interleukin 2 receptor gene expression by p40x encoded by human T-cell leukemia virus type I EMBO J 1986 5 2883 2888 3024966
Ballard DW Bohnlein E Lowenthal JW Wano Y Franza BR Greene WC HTLV-I tax induces cellular proteins that activate the kappa B element in the IL-2 receptor alpha gene Science 1988 241 1652 1655 2843985
Cross SL Feinberg MB Wolf JB Holbrook NJ Wong-Staal F Leonard WJ Regulation of the human interleukin-2 receptor alpha chain promoter: activation of a nonfunctional promoter by the transactivator gene of HTLV-I Cell 1987 49 47 56 3030566 10.1016/0092-8674(87)90754-9
Maruyama M Shibuya H Harada H Hatakeyama M Seiki M Fujita T Inoue J Yoshida M Taniguchi T Evidence for aberrant activation of the interleukin-2 autocrine loop by HTLV-I-encoded p40x and T3/Ti complex triggering Cell 1987 48 343 350 3026643 10.1016/0092-8674(87)90437-5
Tsukahara T Kannagi M Ohashi T Kato H Arai M Nunez G Iwanaga Y Yamamoto N Ohtani K Nakamura M Fujii M Induction of Bcl-x(L) expression by human T-cell leukemia virus type 1 Tax through NF-kappaB in apoptosis-resistant T-cell transfectants with Tax J Virol 1999 73 7981 7987 10482545
Akagi T Ono H Shimotohno K Expression of cell-cycle regulatory genes in HTLV-I infected T-cell lines: possible involvement of Tax1 in the altered expression of cyclin D2, p18Ink4 and p21Waf1/Cip1/Sdi1 Oncogene 1996 12 1645 1652 8622884
Mori N Fujii M Hinz M Nakayama K Yamada Y Ikeda S Yamasaki Y Kashanchi F Tanaka Y Tomonaga M Yamamoto N Activation of cyclin D1 and D2 promoters by human T-cell leukemia virus type I tax protein is associated with IL-2-independent growth of T cells Int J Cancer 2002 99 378 385 11992406 10.1002/ijc.10388
Iwanaga R Ohtani K Hayashi T Nakamura M Molecular mechanism of cell cycle progression induced by the oncogene product Tax of human T-cell leukemia virus type I Oncogene 2001 20 2055 2067 11360190 10.1038/sj.onc.1204304
Fujii M Niki T Mori T Matsuda T Matsui M Nomura N Seiki M HTLV-1 Tax induces expression of various immediate early serum responsive genes Oncogene 1991 6 1023 1029 1906155
Endo K Hirata A Iwai K Sakurai M Fukushi M Oie M Higuchi M Hall WW Gejyo F Fujii M Human T-cell leukemia virus type 2 (HTLV-2) Tax protein transforms a rat fibroblast cell line but less efficiently than HTLV-1 Tax J Virol 2002 76 2648 2653 11861831 10.1128/JVI.76.6.2648-2653.2002
Hirata A Higuchi M Niinuma A Ohashi M Fukushi M Oie M Akiyama M Tanaka Y Gejyo F Fujii M PDZ domain-binding motif of human T-cell leukemia virus type 1 Tax oncoprotein augments the transforming activity in a rat fibroblast cell line Virology 2004 318 327 336 14972558 10.1016/j.virol.2003.10.006
Rousset R Fabre S Desbois C Bantignies F Jalinot P The C-terminus of the HTLV-1 Tax oncoprotein mediates interaction with the PDZ domain of cellular proteins Oncogene 1998 16 643 654 9482110 10.1038/sj.onc.1201567
Suzuki T Ohsugi Y Uchida-Toita M Akiyama T Yoshida M Tax oncoprotein of HTLV-1 binds to the human homologue of Drosophila discs large tumor suppressor protein, hDLG, and perturbs its function in cell growth control Oncogene 1999 18 5967 5972 10557085 10.1038/sj.onc.1203008
Ohashi M Sakurai M Higuchi M Mori N Fukushi M Oie M Coffey RJ Yoshiura K Tanaka Y Uchiyama M Hatanaka M Fujii M Human T-cell leukemia virus type 1 Tax oncoprotein induces and interacts with a multi-PDZ domain protein, MAGI-3 Virology 2004 320 52 62 15003862 10.1016/j.virol.2003.11.014
Iwanaga Y Tsukahara T Ohashi T Tanaka Y Arai1 M Nakamura M Ohtani K Koya Y Kannagi M Yamamoto N Fujii M Human T-cell leukemia virus type 1 Tax protein abrogates interleukin 2 dependence in a mouse T-cell line J Virol 1999 73 1271 1277 9882331
Niinuma A Higuchi M Takahashi M Oie M Tanaka Y Gejyo F Tanaka N Sugamura K Xie L Green PL Fujii M Aberrant Activation of the Interleukin 2 Autocrine Loop through NFAT by Non-leukemogenic Human T-cell Leukemia Virus Type 2 but not Leukemogenic Type 1 Virus J Virol
Kiyono T Hiraiwa A Fujita M Hayashi Y Akiyama T Ishibashi M Binding of high-risk human papillomavirus E6 oncoproteins to the human homologue of the Drosophila discs large tumor suppressor protein Proc Natl Acad Sci U S A 1997 94 11612 11616 9326658 10.1073/pnas.94.21.11612
Nguyen ML Nguyen MM Lee D Griep AE Lambert PF The PDZ ligand domain of the human papillomavirus type 16 E6 protein is required for E6's induction of epithelial hyperplasia in vivo J Virol 2003 77 6957 6964 12768014 10.1128/JVI.77.12.6957-6964.2003
Mori N Yamada Y Ikeda S Yamasaki Y Tsukasaki K Tanaka Y Tomonaga M Yamamoto N Fujii M Bay 11–7082 inhibits transcription factor NF-kappaB and induces apoptosis of HTLV-I-infected T-cell lines and primary adult T-cell leukemia cells Blood 2002 100 1828 1834 12176906 10.1182/blood-2002-01-0151
Tanaka Y Yoshida A Tozawa H Shida H Nyunoya H Shimotohno K Production of a recombinant human T-cell leukemia virus type-I trans-activator (tax1) antigen and its utilization for generation of monoclonal antibodies against various epitopes on the tax1 antigen Int J Cancer 1991 48 623 630 1710610
Lewis MJ Novoa P Ishak R Ishak M Salemi M Vandamme AM Kaplan MH Hall WW Isolation, cloning, and complete nucleotide sequence of a phenotypically distinct Brazilian isolate of human T-lymphotropic virus type II (HTLV-II) Virology 2000 271 142 154 10814579 10.1006/viro.2000.0284
|
16042787
|
PMC1199618
|
CC BY
|
2021-01-04 16:36:39
|
no
|
Retrovirology. 2005 Jul 23; 2:46
|
utf-8
|
Retrovirology
| 2,005 |
10.1186/1742-4690-2-46
|
oa_comm
|
==== Front
RetrovirologyRetrovirology1742-4690BioMed Central London 1742-4690-2-511609297010.1186/1742-4690-2-51ResearchEfficient trapping of HIV-1 envelope protein by hetero-oligomerization with an N-helix chimera Ou Wu [email protected] Jonathan [email protected] Laboratory of Molecular Microbiology, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Building 4, Room 336, Bethesda, MD 20892, USA2005 10 8 2005 2 51 51 22 6 2005 10 8 2005 Copyright © 2005 Ou and Silver; licensee BioMed Central Ltd.2005Ou and Silver; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The N-heptad repeat region of the HIV-1 Transmembrane Envelope protein is a trimerization domain that forms part of a "six helix bundle" crucial to Envelope-mediated membrane fusion. N-heptad repeat peptides have been used as extracellular reagents to inhibit virus fusion.
Results
When expressed intracellularly with wild-type HIV-1 Envelope protein, the N-heptad repeat domain efficiently hetero-oligomerized with Envelope and trapped it in the endoplasmic reticulum or early Golgi, as indicated by lack of transport to the cell surface, absent proteolytic processing, and aberrant glycosylation.
Conclusion
Post-translational processing of HIV Envelope is very sensitive to an agent that binds to the N-heptad repeat during synthesis, suggesting that it might be possible to modify drugs that bind to this region to have transport-blocking properties.
==== Body
Background
Retroviral envelope proteins (Env) are synthesized as precursor proteins in the secretory pathway. After co-translational transfer to the endoplasmic reticulum (ER), the envelope precursor trimerizes and becomes extensively glycosylated. On passage through the medial- and trans-Golgi, sugar residues are trimmed and modified, and Env is proteolytically cleaved by a furin-like enzyme into Surface (SU) and Transmembrane (TM) moieties [1-6]. Trimerization is largely determined by a ~ 30 amino acid alpha-helical domain near the amino-terminus of TM designated the N-heptad repeat or N-helix, residues on one side of which associate hydrophobically to form a trimeric "coiled coil" [7-10]. In the case of HIV and related lentiviruses, about 50 amino acids downstream of the N-heptad repeat is another domain that forms an alpha-helix during rearrangements associated with receptor-binding and membrane fusion. This C-helix region of each Env monomer folds back and binds in an anti-parallel orientation in grooves between N-helix monomers to form a thermodynamically stable, "6-helix bundle" whose structure has been determined [8-10]. Formation of the 6-helix bundle is thought to drive fusion by pulling virus and target cell membranes together [11-16]. Subtle interactions between helix residues that do not affect 6-helix bundle thermal stability also impact fusion [17].
Because of their structural and mechanistic importance for fusion, the N and C-helix regions are targets for therapeutic peptides and drugs. C-helix peptides inhibit fusion at nanomolar concentration [18-20]. Extensive structural and mutagenesis studies have shown that they work, at least in part, by competing with the C-helix for binding to the N-helix trimer [21-25]. Three bulky hydrophobic side chains at one end of the C-helix fit into a deep hydrophobic pocket in the N-helix trimer that has been proposed as a target for small molecule drugs[22]. N-helix peptides are less potent fusion inhibitors, requiring micromolar concentration[26]. Two mechanisms have been proposed for their action: forming homotrimers that bind viral C-helices, and forming heterotrimers with viral N-helix monomers[27]. When N-helix peptides are added extracellularly, forming heterotrimers requires peptide exchange with monomers in pre-formed virus trimer, which may be inefficient.
We previously reported that when Moloney-murine leukemia virus (Mo-MLV) N-helix was expressed intracellularly as a chimeric protein, it formed heterotrimers with co-expressed wild-type Mo-MLV Env, which blocked transport to the cell surface[28]. The heterotrimers were apparently trapped in the ER since Env in the heterotrimer had an immature glycosylation pattern and was not cleaved into SU and TM, although it could be cleaved by furin in vitro[28]. We now show that similar trapping of HIV-1 Env occurs in cells expressing an HIV-1 N-helix-YFP chimeric protein. The trapping is remarkably efficient as no proteolytically cleaved, heterotrimeric molecules were detectable by Western blot, implying that heterotrimeric molecules do not reach the late Golgi. The strength of the trapping suggests that small molecule drugs that bind N-helix in the ER might be engineered to block subsequent trafficking and thereby inhibit assembly of infectious particles.
Results and Discussion
The amino acid sequence of the HIV N-helix is remarkably conserved among isolates, especially in the helical wheel "a" and "d" positions that mediate trimer association (Figure 1A). We chose a consensus sequence for the N-helix and inserted it in frame between a signal sequence and the yellow fluorescent protein (YFP) gene in a CMV promoter-driven, cell-surface expression vector with a glycosylphosphatidylinositol (gpi) membrane linkage sequence (pYFPgpi)[29] to make pNH-YFPgpi (Figure 1B). We expected that the signal sequence in this construct would direct the nascent N-helix to the secretory pathway where it could interact with co-expressed HIV Envelope, and the YPF provided a convenient tag for visualization and immunoprecipitation (see below). HEK293 cells transfected with this plasmid expressed YFP mainly on the cell surface in a pattern indistinguishable from that induced by pYFPgpi[28] (data not shown). Western blot analysis using anti-GFP antibody showed that HEK293 cells transfected with the N-helix expression plasmid contained the expected 40 kD YFP fusion protein, versus a 36 kD YFP product in cells transfected with the parent vector lacking the N-helix insertion (Figure 1C). As noted previously[28], the parent vector, pYFPgpi, also generated a higher molecular weight YFP species possibly due to aberrant glycosylation (* in figure 1C, lane 2).
Figure 1 A. Comparison of consensus amino acid sequences of N-helix regions from various HIV-1 clades, and the consensus sequence that was used to make pNH-YFPgpi. The letters a and d under the consensus sequence indicate the position of corresponding amino acids on a helical wheel. B. Schematic diagram of the coding sequence regions in expression plasmids pYFPgpi and pNH-YFPgpi. SP, signal peptide; YFP, yellow fluorescent protein; NH, N-helix; GPI, gpi attachment signal. C. Western blot with anti-GFP antiserum of HeLa cells transfected with pNH-YFPgpi (lane 1), pYFPgpi (lane 2), or untransfected HeLa cells (lane 3). *, aberrant YFPgpi product. One of three independent experiments with similar results is shown.
To see if the N-helix-YFP fusion protein affected synthesis or trafficking of wild-type HIV-1 Env, we co-transfected HeLa cells with an expression vector for HIV-1 Env strain AD8 (pAD8) plus either pNH-YFPgpi or pYFPgpi as a control. Western blot analysis of whole cell lysates using polyclonal anti-gp120 (SU) antiserum showed that the N-helix fusion protein partially inhibited processing the gp160 Env precursor to gp120 (Figure 2A, lane 2 versus lane 3). The total amount of Env protein was not affected. Western blot with anti-actin antibody showed that equal amounts of protein were loaded in all samples (Figure 2A, left lower panel).
Figure 2 A. Western blot analysis of HeLa cells untransfected (lanes 1, 4), or transfected with an expression vector for HIV-1 Env and Tat plus either pNH-YFPgpi (lanes 2, 5) or pYFPgpi as control (lanes 3, 6). Left side, cell lysates analyzed with rabbit anti-gp120 antiserum (upper panel), or anti-actin as loading control (lower panel). Right side, cell surface proteins labeled with NHS-biotin, precipitated with avidin-agarose, and analyzed with rabbit anti-gp120 antiserum (upper panel), or anti-integrin α5 as loading control (lower panel). One of two independent experiments with similar results is shown. B. Cell fusion assay. Indicator HeLa-TZM cells (CD4+, CXCR4+, CCR5+, containing an HIV LTR-luciferase reporter) were cultured overnight with HEK293 cells untransfected (left bar) or transfected with an expression vector for pNL4-3 strain HIV-1 Env and Tat, plus either pNH-YFPgpi (middle bar) or pYFPgpi (right bar) and analyzed for luciferase activity. RLU, relative light units. One of four independent experiments with similar results is shown.
The partial inhibition of Env processing was associated with a more striking inhibition of transport to the cell surface, evaluated by biotinylating cell-surface proteins with biotin-NHS, precipitating biotinylated proteins with avidin-agarose, and analyzing the precipitated proteins by Western blot using anti-gp120 antiserum. Co-expressed N-helix fusion protein markedly reduced cell surface gp-120 (Figure 2A, lane 5 versus lane 6). Western blot using antibody to integrin alpha5 showed that equal amounts of biotinylated cell surface proteins were loaded in all lanes (Figure 2A, right lower panel). The absence of a biotinylated form of gp160 shows that the biotin label did not attach to intracellular proteins.
The reduction in cell surface gp120 was associated with a comparable reduction in cell fusion activity, measured using a standard assay in which HeLa cells or HEK293 cells transfected with plasmids which express HIV-1 Tat as well as Env were mixed with indicator HeLa-TZM cells that express HIV receptor (CD4) and co-receptors (CXCR4 and CCR5), and contain a luciferase reporter driven by the HIV-1 LTR. Cell fusion induced by a CXCR4-tropic Env (derived from pNL4-3) was reduced 8- to 10-fold by co-expression of the N-helix fusion protein, compared to co-expression of the control YFPgpi (Figure 2B, lower panel). Cell fusion induced by a CCR5-tropic Env (derived from pAD8) was reduced 2–5 fold in 3 comparable experiments. Lower inhibition in the case of the CCR5-tropic Env may be due to greater expression of Env by the pAD8Env vector compared to the pNL4-3Env vector (unpublished observations), and/or to greater expression of CCR5 than CXCR4 by the TZM indicator cells, which were engineered to overexpress CCR5.
To see if the N-helix-YFP fusion protein physically associated with HIV-1 Env, we immuno-precipitated cell lysates with anti-GFP antibody and analyzed the immunoprecipitates by Western blot using anti-gp120 antiserum. In cells co-transfected with pNH-YFPgpi plus the HIV-1 Env expression vector, anti-GFP antibody co-immunoprecipitated the Env precursor gp160 but not gp-120, even though gp120 was present in the cell lysate (Figure 3, lane 2 versus lane 1). This shows that, to the limit of sensitivity of Western blot, all of the HIV Env that hetero-oligomerized with the N-helix fusion protein was prevented from being processed to gp120. A similar result was obtained in the case of MLV: the Env that co-immunoprecipitated with chimeric N-helix was not detectably proteolytically processed[28]. The small amount of Env that was processed to SU in the current experiments (Figure 3, lane 1) presumably came from wild-type Env molecules that homotrimerized rather than forming hetero-oligomers with N-helix-YFP. In control cells co-transfected with pYFPgpi instead of pNH-YFPgpi, the Env precursor was more efficiently processed to gp120, as expected (Figure 3, lane 4 versus lane 1), and the anti-GFP antibody did not co-immunoprecipitate HIV Env (lanes 4 and 5); the latter shows that the interaction between N-helix-YFP and Env was not due to non-specific stickiness of YFP.
Figure 3 Western blot analysis of HeLa cells transfected with an expression vector for HIV-1 Env plus either pNH-YFPgpi (lanes 1–3) or pYFPgpi as control (lanes 4–6). Total cell lysates (lanes 1, 4) or anti-GFP immunoprecipitates (lanes 2, 3, 5, 6) were treated with furin (lanes 3, 6) or mock treated (lanes 1, 2, 4, 5) and analyzed with rabbit anti-gp120 antiserum.
To see if the processing defect of the Env precursor complexed with N-helix-YFP was due to resistance of this form of Env to furin, we incubated the immunoprecipitates with furin enzyme. Exogenous furin cleaved the co-immunoprecipitated Env precursor to a species that migrated slightly faster than native gp120 (Figure 3, lane 3 versus lane 1). Altered mobility of the furin cleavage product is likely due to aberrant glycosylation. In similar experiments with MLV[28], the in vitro cleavage product of hetero-oligomerized Env treated with furin also migrated slightly faster than normal SU, but co-migrated with SU from cells treated with brefeldin A, a drug that disrupts the Golgi and blocks Golgi-associated sugar modifications[30]. Since the HIV Env precursor complexed with N-helix-YFP was cleavable in vitro but was not cleaved in vivo, the simplest interpretation of the data is that hetero-oligomerization of HIV Env gp160 with N-helix-YFP leads to arrest of this species in the ER or cis Golgi, preventing maturation of sugars and proteolytic cleavage that normally occur in the medial and trans Golgi. It is also possible that the hetero-oligomerized Env is misrouted to some other furin-negative compartment.
In comparable experiments with Mo-MLV we showed that blocking the ability of the MLV N-helix to trimerize by substituting proline for leucine in the center of the trimerization domain abolished its ability to trap Env in the ER, providing additional evidence that oligomerization was responsible for the trapping[28]. Further, the YFPgpi portion of the chimeric N-helix did not contribute to inhibition, since the MLV N-helix linked to a 9 amino acid HA epitope instead of YFPgpi was equally potent in trapping MLV Env in the ER[28]. Since neither YPF nor the HA epitope inhibit trafficking when attached to other proteins, we surmise that inclusion of N-helix by itself in a heterotrimer with Env causes misfolding.
Given the strong conservation of amino acids that direct N-helix trimerization, it is likely that intracellular expression of an N-helix chimera would inhibit processing of all strains of HIV. From a practical point of view, however, the dominant negative effect of N-helix constructs is limited by their level of expression in the ER compared to that of wild-type Env. Both the HIV and MLV N-helix-YFP fusion proteins are efficiently transported to the cell surface when expressed alone, based on the pattern of fluorescence in confocal microscopy, which is mainly restricted to the plasma membrane as previously shown[28]. In cells co-expressing Env, there was a slight increase in intracellular fluorescence but most of the fluorescence remained on the plasma membrane, suggesting that most N-helix-YFP molecules leave the ER before having a chance to hetero-oligomerize with Env. To attempt to block "premature" egress, which might reduce its ability to form a heterotrimer, we replaced the gpi attachment peptide signal with a "KDEL" ER retention signal to make pNH-YFP-KDEL. The KDEL construct was efficiently retained in the ER as judged by a reticular, cytoplasmic fluorescence pattern; however, it was not more inhibitory than the unmodified fusion protein when co-transfected with HIV Env in a cell fusion assay (data not shown). We also explored the effect of shortening the N-helix by deleting 7 or 14 amino acids (two or four alpha-helical turns) from either end, since short peptides can sometimes be induced to cross cell membranes by attaching a basic membrane transport domain[31]. However, the shorter N-helix versions were less inhibitory in the cell fusion assay than the full N-helix.
How do the N-helix chimeric proteins interact with HIV Env expressed in the secretory pathway? Like extracellular N-helix peptides, they could form heterotrimers with N-helix regions in Env molecules[32], or homotrimerize and then interact with C-helix regions in Env[33]. These possibilities might be distinguished by seeing how mutations in Env C-helix residues versus N-helix residues affect hetero-oligomerization with N-helix constructs. Extracellular N-helix peptides preferentially bind receptor-activated Env[33], presumably because the interacting N- or C-helix regions are poorly exposed in the mature, unactivated Env. Our observations imply that surfaces in Env that interact with N-helix chimeras are exposed in nascent Env. Our results do not exclude the possibility that N-helix-YFPgpi also inhibits fusion by interacting with receptor-activated Env on the cell surface.
An unexpected observation made in the course of these studies was that the control vector pYFPgpi inhibited fusion about ten-fold when transfected with HIV Env expression vectors. Therefore, to evaluate the effect of the N-helix we compared transfections with pNH-YFPgpi to transfections with pYFPgpi. The reason for inhibition by pYFPgpi is currently under investigation.
Conclusion
The remarkable efficacy of trapping by hetero-oligomerization suggests a drug strategy of trying to engineer small molecules that bind the Env N-helix in the ER in a fashion that blocks trafficking. Small molecules that bind to the hydrophobic pocket at one end of the N-helix trimer are under development[22,34-37]. Coupling them to an ER retention signal like KDEL might inhibit Env trafficking. Macrocycle drugs such as cyclosporinA act as bivalent ligands that bring together two proteins, one of which can function as an ER chaperone (e.g., cyclophilinB)[38]. Structures of several of these macrocycle-chaperone complexes are known, and they show that only one side of the macrocycle is necessary for tight (nanomolar) binding to the chaperone[39,40]. Based on these results, it might be possible to engineer a bi-dentate drug, one portion of which binds in grooves of the HIV Env N-helix trimer while another portion binds an ER chaperone, promoting ER retention. A natural example related to this strategy was recently described: a small molecule intermediate in the cholesterol synthesis pathway (farnesol) that binds an ER-associated enzyme in this pathway (HMG-CoA reductase), resulting in accelerated degradation of the enzyme[41]. The idea we propose is the "flip side" of a hunt for small molecules that inhibit protein misfolding[42,43]. HIV Env may provide a propitious target for drug-induced trapping since it is naturally inefficiently processed[4] and HIV virions from several strains bear very few Env trimers on their surface[44,45].
Materials and methods
Constructs
We aligned N-helix amino acid sequences of HIV-1 envelopes in the Los Alamos database and generated a consensus sequence for each clade (A, B, C, D, F, G, H and O), then generated the consensus sequence for all the clades (Fig. 1A), which is the N-helix sequence used in this paper. Oligonucleotides encoding this HIV-1 N-helix with Sal I restriction enzyme overhanging sequences were synthesized, annealed and ligated into plasmid pYFP-gpi[29] at the Sal I site, to generate plasmid pNH-YFPgpi (Figure 1B). The oligonucleotide sequences used were: 5' tcgacttctggtatagtgcagcagcagaacaatttgctgagggctattgaggcgcaacagcatctgtt-gcaactcacagtctggggcatcaaacagctccaggcaagagtcctggcg 3', and 5' tcgacgccaggactcttgcct-ggagctgtttgatgccccagactgtgagttgcaacagatgctgttgcgcctcaatagccctcagcaaattgttctgctgctgcactataccagaag 3'. For expression of T-tropic (CXCR4-using) and M-tropic (CCR5-using) HIV-1 Env, we used plasmids pdl1443 and pAD8Env, respectively, which were derived from molecular clones pNL4-3 and pAD8 by deleting 3.1 kb of gag sequences between SphI and MscI sites [46]. These plasmids express HIV-1 Tat as well as Env.
Transfection, surface protein labeling and cell fusion
HEK 293 or HeLa cells were co-transfected with Env-expressing constructs pdl1443 or pAD8Env, plus pNH-YFPgpi or pYFP-gpi as control, using Lipofectamine2000 (Invitrogen, Carlsbad, CA). Twenty four to 48 hours later, the cells were rinsed with phosphate buffered saline (PBS) and labeled on ice with 1 mg/ml Sulfo-NHS-LC-LC biotin (Pierce, Rockford, IL) in PBS for one hour. After labeling, the biotinylation reagent was quenched with 100 mM glycine in PBS buffer. Following PBS wash, some of the cells were lysed with RIPA lysis buffer (150 mM NaCl, 1% Triton X-100, 0.1% SDS, 0.5% sodium deoxycholate) for immunoprecipitation or direct western blot, and the remainder of the cells were co-cultivated with TZM-bl cells[47,48] overnight and then assayed for luciferase activity (Promega, Madison, WI) as described[28].
Immunoprecipitation, furin cleavage in vitro and western blot
To immunoprecipitate cell surface biotin-labeled HIV-1 Env protein, avidin beads were directly added to the labeled cell lysate. After binding for 2 hours with agitation, beads were washed with lysis buffer and PBS, and bound proteins eluted by boiling for 3 min in 1X SDS-PAGE sample buffer (Invitrogen). The eluate was run on a 4–12% SDS-PAGE, transferred to PVDF membrane, and blotted with polyclonal anti-gp120 serum (a gift from Klaus Strebel, LMM/NIAID, made by immunizing rabbits with purified gp120 of HIV-1 strain IIIB), or with anti-integrin α5 (BD Transduction Lab, San Diego, CA) as a control. To cross-immunoprecipitate intracellular HIV-1 Env protein, cell lysates were pre-cleared with normal mouse serum and protein G Sepharose beads (Amersham, Piscataway, NJ) for 4 hour at 4°C with agitation. The supernatant was collected and immunoprecipitated with monoclonal anti-GFP antibody (Clontech, Palo Alto, CA) overnight and protein G beads for an additional 2 hours. Beads were washed 3 times with lysis buffer and 2 times with PBS buffer. Protein was eluted from one half of the beads by boiling in 1X SDS-PAGE sample buffer, and the remaining beads were re-suspended in furin reaction buffer (0.5% triton X-100, 1 mM CaCl2, 100 mM HEPES, 1 mM β-mercaptoethanol) and treated with 0.578 mg/ml furin (R&D systems, Minneapolis, MN) at 37°C for 16 hours as descibed[28]. The reaction was stopped and protein eluted by boiling in 1× SDS-PAGE sample buffer. The eluted protein was analyzed by western blot using rabbit anti-gp120 serum.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
WO carried out the studies, participated in the design and conception of the project, and helped draft the manuscript. JS participated in the design and conception of the project and drafted the manuscript. Both authors read and approved the final manuscript.
Acknowledgements
We thank Dr. P. Keller for the pYFP-gpi plasmid. TZM-bl cells were obtained from Drs. J. C. Kappes and X. Wu through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH.
==== Refs
Dewar RL Natarajan V Vasudevachari MB Salzman NP Synthesis and processing of human immunodeficiency virus type 1 envelope proteins encoded by a recombinant human adenovirus J Virol 1989 63 129 136 2535720
Doms RW Earl PL Moss B The assembly of the HIV-1 env glycoprotein into dimers and tetramers Adv Exp Med Biol 1991 300 203 219 1781345
Earl PL Moss B Doms RW Folding, interaction with GRP78-BiP, assembly, and transport of the human immunodeficiency virus type 1 envelope protein J Virol 1991 65 2047 2055 1900540
Willey RL Bonifacino JS Potts BJ Martin MA Klausner RD Biosynthesis, cleavage, and degradation of the human immunodeficiency virus 1 envelope glycoprotein gp160 Proc Natl Acad Sci U S A 1988 85 9580 9584 2849111
Hallenberger S Moulard M Sordel M Klenk HD Garten W The role of eukaryotic subtilisin-like endoproteases for the activation of human immunodeficiency virus glycoproteins in natural host cells J Virol 1997 71 1036 1045 8995623
Merkle RK Helland DE Welles JL Shilatifard A Haseltine WA Cummings RD gp160 of HIV-I synthesized by persistently infected Molt-3 cells is terminally glycosylated: evidence that cleavage of gp160 occurs subsequent to oligosaccharide processing Arch Biochem Biophys 1991 290 248 257 1898096 10.1016/0003-9861(91)90616-Q
Center RJ Lebowitz J Leapman RD Moss B Promoting trimerization of soluble human immunodeficiency virus type 1 (HIV-1) Env through the use of HIV-1/simian immunodeficiency virus chimeras J Virol 2004 78 2265 2276 14963123 10.1128/JVI.78.5.2265-2276.2004
Chan DC Fass D Berger JM Kim PS Core structure of gp41 from the HIV envelope glycoprotein Cell 1997 89 263 273 9108481 10.1016/S0092-8674(00)80205-6
Weissenhorn W Dessen A Harrison SC Skehel JJ Wiley DC Atomic structure of the ectodomain from HIV-1 gp41 Nature 1997 387 426 430 9163431 10.1038/387426a0
Tan K Liu J Wang J Shen S Lu M Atomic structure of a thermostable subdomain of HIV-1 gp41 Proc Natl Acad Sci U S A 1997 94 12303 12308 9356444 10.1073/pnas.94.23.12303
Lu M Stoller MO Wang S Liu J Fagan MB Nunberg JH Structural and functional analysis of interhelical interactions in the human immunodeficiency virus type 1 gp41 envelope glycoprotein by alanine-scanning mutagenesis J Virol 2001 75 11146 11156 11602754 10.1128/JVI.75.22.11146-11156.2001
Weissenhorn W Dessen A Calder LJ Harrison SC Skehel JJ Wiley DC Structural basis for membrane fusion by enveloped viruses Mol Membr Biol 1999 16 3 9 10332732 10.1080/096876899294706
Melikyan GB Markosyan RM Hemmati H Delmedico MK Lambert DM Cohen FS Evidence that the transition of HIV-1 gp41 into a six-helix bundle, not the bundle configuration, induces membrane fusion J Cell Biol 2000 151 413 423 11038187 10.1083/jcb.151.2.413
Cao J Bergeron L Helseth E Thali M Repke H Sodroski J Effects of amino acid changes in the extracellular domain of the human immunodeficiency virus type 1 gp41 envelope glycoprotein J Virol 1993 67 2747 2755 8474172
Ji H Shu W Burling FT Jiang S Lu M Inhibition of human immunodeficiency virus type 1 infectivity by the gp41 core: role of a conserved hydrophobic cavity in membrane fusion J Virol 1999 73 8578 8586 10482611
Mo H Konstantinidis AK Stewart KD Dekhtyar T Ng T Swift K Matayoshi ED Kati W Kohlbrenner W Molla A Conserved residues in the coiled-coil pocket of human immunodeficiency virus type 1 gp41 are essential for viral replication and interhelical interaction Virology 2004 329 319 327 15518811
Suntoke TR Chan DC The fusion activity of HIV-1 gp41 depends on interhelical interactions J Biol Chem 2005 280 19852 19857 15772068 10.1074/jbc.M502196200
Jiang S Lin K Strick N Neurath AR HIV-1 inhibition by a peptide Nature 1993 365 113 8371754 10.1038/365113a0
Eckert DM Kim PS Mechanisms of viral membrane fusion and its inhibition Annu Rev Biochem 2001 70 777 810 11395423 10.1146/annurev.biochem.70.1.777
Wild C Greenwell T Matthews T A synthetic peptide from HIV-1 gp41 is a potent inhibitor of virus-mediated cell-cell fusion AIDS Res Hum Retroviruses 1993 9 1051 1053 8312047
Lu M Blacklow SC Kim PS A trimeric structural domain of the HIV-1 transmembrane glycoprotein Nat Struct Biol 1995 2 1075 1082 8846219 10.1038/nsb1295-1075
Chan DC Chutkowski CT Kim PS Evidence that a prominent cavity in the coiled coil of HIV type 1 gp41 is an attractive drug target Proc Natl Acad Sci U S A 1998 95 15613 15617 9861018 10.1073/pnas.95.26.15613
Wild C Greenwell T Shugars D Rimsky-Clarke L Matthews T The inhibitory activity of an HIV type 1 peptide correlates with its ability to interact with a leucine zipper structure AIDS Res Hum Retroviruses 1995 11 323 325 7786578
Rimsky LT Shugars DC Matthews TJ Determinants of human immunodeficiency virus type 1 resistance to gp41-derived inhibitory peptides J Virol 1998 72 986 993 9444991
Furuta RA Wild CT Weng Y Weiss CD Capture of an early fusion-active conformation of HIV-1 gp41 Nat Struct Biol 1998 5 276 279 9546217 10.1038/nsb0498-276
Wild C Oas T McDanal C Bolognesi D Matthews T A synthetic peptide inhibitor of human immunodeficiency virus replication: correlation between solution structure and viral inhibition Proc Natl Acad Sci U S A 1992 89 10537 10541 1438243
Bewley CA Louis JM Ghirlando R Clore GM Design of a novel peptide inhibitor of HIV fusion that disrupts the internal trimeric coiled-coil of gp41 J Biol Chem 2002 277 14238 14245 11859089 10.1074/jbc.M201453200
Ou W Silver J Inhibition of murine leukemia virus envelope protein (env) processing by intracellular expression of the env N-terminal heptad repeat region J Virol 2005 79 4782 4792 15795264 10.1128/JVI.79.8.4782-4792.2005
Keller P Toomre D Diaz E White J Simons K Multicolour imaging of post-Golgi sorting and trafficking in live cells Nat Cell Biol 2001 3 140 149 11175746 10.1038/35055042
Misumi Y Misumi Y Miki K Takatsuki A Tamura G Ikehara Y Novel blockade by brefeldin A of intracellular transport of secretory proteins in cultured rat hepatocytes J Biol Chem 1986 261 11398 11403 2426273
Wadia JS Dowdy SF Transmembrane delivery of protein and peptide drugs by TAT-mediated transduction in the treatment of cancer Adv Drug Deliv Rev 2005 57 579 596 15722165 10.1016/j.addr.2004.10.005
Caffrey M Kaufman J Stahl S Wingfield P Gronenborn AM Clore GM Monomer-trimer equilibrium of the ectodomain of SIV gp41: insight into the mechanism of peptide inhibition of HIV infection Protein Sci 1999 8 1904 1907 10493592
He Y Vassell R Zaitseva M Nguyen N Yang Z Weng Y Weiss CD Peptides trap the human immunodeficiency virus type 1 envelope glycoprotein fusion intermediate at two sites J Virol 2003 77 1666 1671 12525600 10.1128/JVI.77.3.1666-1671.2003
Eckert DM Malashkevich VN Hong LH Carr PA Kim PS Inhibiting HIV-1 entry: discovery of D-peptide inhibitors that target the gp41 coiled-coil pocket Cell 1999 99 103 115 10520998 10.1016/S0092-8674(00)80066-5
Ferrer M Kapoor TM Strassmaier T Weissenhorn W Skehel JJ Oprian D Schreiber SL Wiley DC Harrison SC Selection of gp41-mediated HIV-1 cell entry inhibitors from biased combinatorial libraries of non-natural binding elements Nat Struct Biol 1999 6 953 960 10504731 10.1038/13324
Zhou G Ferrer M Chopra R Kapoor TM Strassmaier T Weissenhorn W Skehel JJ Oprian D Schreiber SL Harrison SC Wiley DC The structure of an HIV-1 specific cell entry inhibitor in complex with the HIV-1 gp41 trimeric core Bioorg Med Chem 2000 8 2219 2227 11026535 10.1016/S0968-0896(00)00155-3
Debnath AK Radigan L Jiang S Structure-based identification of small molecule antiviral compounds targeted to the gp41 core structure of the human immunodeficiency virus type 1 J Med Chem 1999 42 3203 3209 10464007 10.1021/jm990154t
Mikol V Kallen J Walkinshaw MD X-ray structure of a cyclophilin B/cyclosporin complex: comparison with cyclophilin A and delineation of its calcineurin-binding domain Proc Natl Acad Sci U S A 1994 91 5183 5186 8197205
Sedrani R Kallen J Martin Cabrejas LM Papageorgiou CD Senia F Rohrbach S Wagner D Thai B Jutzi Eme AM France J Oberer L Rihs G Zenke G Wagner J Sanglifehrin-cyclophilin interaction: degradation work, synthetic macrocyclic analogues, X-ray crystal structure, and binding data J Am Chem Soc 2003 125 3849 3859 12656618 10.1021/ja021327y
Kallen J Sedrani R Zenke G Wagner J Structure of human cyclophilin a in complex with the novel immunosuppressant sanglifehrin a at 1.6 a resolution J Biol Chem 2005 280 21965 21971 15772070 10.1074/jbc.M501623200
Shearer AG Hampton RY Lipid-mediated, reversible misfolding of a sterol-sensing domain protein EMBO J 2005 24 149 159 15635451 10.1038/sj.emboj.7600498
Gestwicki JE Crabtree GR Graef IA Harnessing chaperones to generate small-molecule inhibitors of amyloid beta aggregation Science 2004 306 865 869 15514157 10.1126/science.1101262
Tanaka M Machida Y Nukina N A novel therapeutic strategy for polyglutamine diseases by stabilizing aggregation-prone proteins with small molecules J Mol Med 2005 83 343 352 15759103 10.1007/s00109-004-0632-2
Chertova E Bess JJWJ Crise BJ Sowder II RC Schaden TM Hilburn JM Hoxie JA Benveniste RE Lifson JD Henderson LE Arthur LO Envelope glycoprotein incorporation, not shedding of surface envelope glycoprotein (gp120/SU), Is the primary determinant of SU content of purified human immunodeficiency virus type 1 and simian immunodeficiency virus J Virol 2002 76 5315 5325 11991960 10.1128/JVI.76.11.5315-5325.2002
Zhu P Chertova E Bess JJ Lifson JD Arthur LO Liu J Taylor KA Roux KH Electron tomography analysis of envelope glycoprotein trimers on HIV and simian immunodeficiency virus virions Proc Natl Acad Sci U S A 2003 100 15812 15817 14668432 10.1073/pnas.2634931100
Felser JM Klimkait T Silver J A syncytia assay for human immunodeficiency virus type I (HIV-I) envelope protein and its use in studying HIV-I mutations Virology 1989 170 566 570 2543131 10.1016/0042-6822(89)90448-0
Derdeyn CA Decker JM Sfakianos JN Wu X O'Brien WA Ratner L Kappes JC Shaw GM Hunter E Sensitivity of human immunodeficiency virus type 1 to the fusion inhibitor T-20 is modulated by coreceptor specificity defined by the V3 loop of gp120 J Virol 2000 74 8358 8367 10954535 10.1128/JVI.74.18.8358-8367.2000
Wei X Decker JM Liu H Zhang Z Arani RB Kilby JM Saag MS Wu X Shaw GM Kappes JC Emergence of resistant human immunodeficiency virus type 1 in patients receiving fusion inhibitor (T-20) monotherapy Antimicrob Agents Chemother 2002 46 1896 1905 12019106 10.1128/AAC.46.6.1896-1905.2002
|
16092970
|
PMC1199619
|
CC BY
|
2021-01-04 16:36:39
|
no
|
Retrovirology. 2005 Aug 10; 2:51
|
utf-8
|
Retrovirology
| 2,005 |
10.1186/1742-4690-2-51
|
oa_comm
|
==== Front
RetrovirologyRetrovirology1742-4690BioMed Central London 1742-4690-2-531610917210.1186/1742-4690-2-53ResearchCXCR4 and CCR5 shRNA transgenic CD34+ cell derived macrophages are functionally normal and resist HIV-1 infection Anderson Joseph [email protected] Ramesh [email protected] Dept. Microbiology, Immunology and Pathology, Colorado State University, Fort Collins, Colorado 80523, USA2005 18 8 2005 2 53 53 20 7 2005 18 8 2005 Copyright © 2005 Anderson and Akkina; licensee BioMed Central Ltd.2005Anderson and Akkina; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Stable simultaneous knock down of the HIV-1 coreceptors CCR5 and CXCR4 is a promising strategy to protect cells from both R5 macrophage tropic and X4 T cell tropic as well as dual tropic viral infections. The potency of shRNAs in targeted gene silencing qualifies them as powerful tools for long term HIV gene therapy. Our previous work with a bispecific lentiviral vector containing CXCR4 and CCR5 shRNAs showed efficacy in down regulating both coreceptors and conferring viral resistance to both X4 and R5-tropic strains of HIV-1 in cultured cell lines. To extend these results to a stem cell gene therapy setting, here we show transduction of primary CD34+ hematopoietic progenitor cells to derive normal end stage cells that are resistant to HIV-1 infection.
Results
The bispecific XHR lentiviral vector harboring CXCR4 and CCR5 shRNA expression cassettes was efficient in transducing CD34+ cells. The transduced cells gave rise to morphologically normal transgenic macrophages when cultured in cytokine media. There was a marked down regulation of both coreceptors in the stably transduced macrophages which showed resistance to both R5 and X4 HIV-1 strains upon in vitro challenge. Since off target effects by some shRNAs may have adverse effects on transgenic cells, the stably transduced macrophages were further analyzed to determine if they are phenotypically and functionally normal. FACS evaluation showed normal levels of the characteristic surface markers CD14, CD4, MHC class II, and B7.1. Phagocytic functions were also normal. The transgenic macrophages demonstrated normal abilities in up-regulating the costimulatory molecule B7.1 upon LPS stimulation. Furthermore, IL-1 and TNFα cytokine secretion in response to LPS stimulation was also normal. Thus, the transgenic macrophages appear to be phenotypically and functionally normal.
Conclusion
These studies have demonstrated for the first time that a bispecific lentiviral vector could be used to stably deliver shRNAs targeted to both CCR5 and CXCR4 coreceptors into CD34+ hematopoietic progenitor cells and derive transgenic macrophages. Transgenic macrophages with down regulated coreceptors were resistant to both R5 and X4 tropic HIV-1 infections. The differentiated cells were also phenotypically and functionally normal indicating no adverse effects of shRNAs on lineage specific differentiation of stem cells. It is now possible to construct gene therapeutic lentiviral vectors incorporating multiple shRNAs targeted to cellular molecules that aid in HIV-1 infection. Use of these vectors in a stem cell setting shows great promise for sustained HIV/AIDS gene therapy.
==== Body
Background
Gene therapy approaches using the strategy of intracellular immunization hold considerable promise towards controlling HIV infection. Previous attempts with anti-HIV molecules that employed RNA decoys, transdominant proteins, and ribozymes were promising towards developing novel therapies [1-12]. With the recent discovery of RNA interference (RNAi), a new and more powerful tool has become available to add to the growing anti-HIV arsenal. The phenomenon of RNA interference has proven to be highly potent in post-transcriptional gene silencing [13-15]. Mediated by sequence specific small-interfering RNAs (siRNAs), RNAi can effectively down regulate the expression of either viral or cellular RNA targets by selective degradation of homologous mRNAs [16]. The mechanism of mRNA degradation involves an endonuclease present in the RNA-induced silencing complex (RISC) which is guided by the antisense component of the siRNA for target recognition [13,14]. A number of reports have shown that delivery of siRNAs by transfection of presynthesized siRNAs or plasmids encoding siRNAs into cultured cells can effectively inhibit HIV-1 infections [17-26]. However, due to the transient nature of transfected nucleic acid, the antiviral effects are only temporary. For HIV gene therapy strategies to succeed long range, it is necessary that siRNA coding transgenes be maintained and expressed long term in a virus susceptible target cell. In this regard, lentiviral vectors have proven to be highly effective in high efficiency gene transduction and sustained gene expression [27-32].
A number of studies using siRNAs have targeted HIV genes as well as the cellular molecules critical for HIV entry, namely CD4, CXCR4 and CCR5 [18,19,21,23,24,33-37]. SiRNAs targeting HIV genes alone will not be sufficient to ward off chronic infection due to the high possibility of generating escape mutants [38,39]. Therefore by targeting host cellular genes critical for viral entry and/or replication, a more sustained efficacy of antiviral effects may be obtained. As a critical player in immunological function, CD4 is physiologically indispensable. The chemokine receptors CXCR4 and CCR5 also play critical roles as coreceptors for viral entry during infection with T cell tropic X4 and macrophage tropic R5 HIV-1 viral strains respectively [40,41]. Their sustained knock down may prove to be more efficacious for long range siRNA therapy.
Since both R5 and X4-tropic viral strains are involved in disease pathogenesis, it is important to consider both coreceptors when developing effective therapeutics. In a segment of the human population, a naturally occurring 32-bp deletion in the CCR5 gene results in the loss of coreceptor function thus conferring significant resistance to HIV infection [42-44]. Homozygous or heterozygous individuals with this mutation remain physiologically normal. With regard to the CXCR4 coreceptor, it was found to be dispensable for T cell development and maturation in murine studies [45].
Based on this rationale, recent work with synthetic siRNAs demonstrated that down regulating either CXCR4 or CCR5 will protect cells from X4 or R5 HIV-1 strains, respectively, at the level of viral entry [18,19,21,23,24,33-37]. Stable expression of an anti-CCR5 siRNA was also achieved using a lentiviral vector. However, down regulating CCR5 alone in the face of an HIV-1 infection is insufficient [34]. Therefore, we recently demonstrated that synthetic bispecific combinatorial constructs as well as a bispecific lentiviral vector targeting both CXCR4 and CCR5 showed efficacy in inhibiting HIV-1 infections in cell culture lines [24,37]. In translating these findings into a stem cell gene therapy setting, this bispecific lentiviral vector was used in the present studies to generate shRNA expressing transgenic macrophages.
Macrophages, along with T cells, are major cell targets of HIV infections. Programming these cells to express shRNAs targeted to the essential coreceptors, CXCR4 and CCR5, could confer resistance to HIV infection. Macrophages also have a significant role in immune system functions as antigen presenting cells and as major effector cells in inflammation. Therefore, protecting macrophages from HIV infection is important in maintaining immune system homeostasis. Since shRNAs can have possible off target effects thus dysregulating cellular physiology, transgenic macrophages also need to be assessed for proper functionality [46]. Here we show that CD34+ hematopoietic progenitor cell derived macrophages expressing shRNAs targeting CXCR4 and CCR5 are functionally normal and resist infection to both X4 and R5-tropic strains of HIV-1.
Results
Lentiviral vector transduction of CD34+ cells with CXCR4 and CCR5 shRNAs and derivation of mature macrophages
A bispecific lentiviral vector XHR, coding for an shRNA targeting CXCR4 driven by a U6 promoter and a CCR5 shRNA under the control of an H1 promoter was designed as previously described (Fig. 1) [37]. This vector also contains an EGFP reporter gene downstream from the shRNA cassettes. CD34+ hematopoietic progenitor cells were transduced with either control GFP or XHR vectors. Cells were then sorted for EGFP and driven towards a myeloid lineage in semi-solid methyl cellulose cytokine media to generate transgenic macrophages. No significant differences were found in the levels of macrophages obtained when compared between the control GFP vector and XHR vector transduced cells or control non-transduced CD34+ cells. The morphology of the transgenic macrophages also appeared normal (data not shown).
Figure 1 Bispecific lentiviral vector (XHR) encoding anti-CXCR4 and CCR5 shRNAs: A) Control transfer vector pHIV-7-GFP encoding a CMV promoter driven EGFP reporter gene. B) To derive the bispecific vector pHIV-XHR-GFP, a U6 promoter driven short hairpin CXCR4 shRNA cassette was cloned into the BamHI site upstream of the CMV-EGFP cassette. The H1-CCR5 shRNA cassette was inserted into an MluI site downstream to the U6-CXCR4 shRNA cassette.
Down regulation of HIV-1 coreceptors CXCR4 and CCR5 in transgenic macrophages
CD34+ derived macrophages normally express both major HIV-1 coreceptors, CXCR4 and CCR5, albeit a lower level of CXCR4. In XHR transduced cells FACS analysis showed an 82% decrease in CXCR4 expression. GFP-alone control vector transduced cells and non-transduced cells displayed normal levels of CXCR4 expression (94%) (Fig. 2A). Similar analysis for CCR5 expression showed a 73% decrease in XHR transduced macrophages with normal levels seen in GFP-alone vector transduced cells similar to non-transduced cells (98%) (Fig. 2B). Thus, stably transduced macrophages exhibited significant down regulation of both the coreceptors CXCR4 and CCR5 due to shRNA targeting.
Figure 2 Down regulation of the coreceptors CXCR4 and CCR5 in XHR transgenic macrophages: GFP-alone and XHR transduced CD34+ derived macrophages were labeled with PE-CY5 conjugated antibodies specific for CXCR4 (A) and CCR5 (B) and analyzed by FACS. Control, nontransduced macrophages are shown superimposed as unshaded areas.
XHR transgenic macrophages resist HIV-1 challenge
To determine if down regulation of CXCR4 and CCR5 coreceptors conferred viral resistance, transduced macrophages were challenged with X4-tropic (NL4-3) and R5-tropic (BaL-1) strains of HIV-1. Antigen ELISAs to detect viral p24 in culture supernatants were performed on various days post-infection. Over a 2-log reduction in viral yield was seen in XHR transduced macrophages challenged with X4-tropic HIV-1 as compared to control cells (Fig. 3A). In BaL-1 challenge experiments, there was over a 1-log reduction in viral titer in XHR transduced macrophages compared to control cells (Fig. 3B). Thus stable coreceptor down regulation by siRNAs resulted in marked protection of transgenic macrophages against viral challenge.
Figure 3 HIV-1 resistance of XHR transgenic macrophages: Control nontransduced (◆), GFP-alone (■), and XHR (▲) transduced CD34+ derived macrophages were challenged with (A) X4-tropic NL4-3 and (B) R5-tropic BaL-1 strains of HIV-1. p24 ELISAs were performed on culture supernatants taken at various time points post-infection. Experiments were performed in triplicate.
Transgenic macrophages display characteristic phenotypic cell surface markers
Macrophages are critical players in the immune system and also participate in the inflammatory response. Recent work demonstrated possible off target effects of some siRNAs [46]. Such effects may disrupt the phenotypic properties of macrophages or alternatively, may interfere with their normal function. Therefore, transgenic macrophages were subjected to phenotypic analyses to assess their characteristic cell surface markers by FACS. Levels of the monocyte/macrophage marker CD14 in XHR macrophages were found to be similar to GFP-alone transduced or nontransduced cells (98% and 97% respectively) (Fig. 4A). Similarly the levels of CD4, a primary HIV-1 receptor, were found at comparable levels for XHR and GFP-alone transduced macrophages at 95% and 93% respectively, coinciding with levels in nontransduced cells (Fig. 4B). The antigen presenting cell surface specific marker, HLA-DR (MHC II) present on macrophages is critical for presenting antigen to CD4+ T cells. A second co-stimulatory molecule B7.1 needed to activate T cells is present at low levels on normal macrophages. Its expression is elevated upon activation with certain stimuli such as LPS. Our evaluation showed that XHR transgenic macrophages displayed similar levels of HLA-DR (92%) when compared to GFP-alone (89%) or with non transduced macrophages (Fig. 4C). The levels of the costimulatory molecule B7.1 were found to be normal at ~15% without stimulation. The transgenic macrophages also displayed capacity to upregulate B7.1 (65%) after LPS stimulation similar to that seen with vector alone and non-transduced control cells (Fig. 4D).
Figure 4 Transgenic macrophages display normal cell surface markers: GFP-alone and XHR transduced CD34+ derived macrophages were labeled with antibodies specific for (A) CD14, (B) CD4, and (C) HLA-DR and analyzed by FACS. Control, nontransduced macrophages are shown superimposed as unshaded areas. (D) B7.1 upregulation of transgenic macrophages stimulated with LPS. Twenty-four hours post-stimulation, macrophages were labeled with a PE-CY5 conjugated anti-B7.1 antibody and analyzed by FACS. B7.1 upregulation data are representative of triplicate experiments.
Transgenic macrophages are functionally normal
As stable expression of some shRNAs could have possible off-target global effects leading to disruption of normal cellular functions, we performed functional assays on transgenic macrophages to evaluate this possibilty. A typical function of macrophages is phagocytosis of foreign material and presentation of antigenic peptides. To determine if XHR transgenic macrophages retained the phagocytic function, they were presented with fluorescently labeled E. coli (Bioparticles®). Foreign cell uptake was measured by FACS. In comparing non-transduced, GFP-alone transduced, and XHR transduced macrophages, no significant differences in the phagocytic capacity were found between the transgenic macrophages and the vector alone transduced or non-transduced cells. Based on fluorecscence levels, XHR macrophage phagocytosis was quantified at 68.2% (Fig. 5E) compared to non transduced and GFP-alone cells at 63.5% and 61.5%, respectively (Fig. 5C and 5D). Transduced Magi-CXCR4 cells, serving as non-phagocytic cell controls did not display any phagocytic activity (Fig. 5B).
Figure 5 Phagocytosis of fluorescently labeled E.coli by CD34+ derived macrophages: E. coli Bioparticles® were added directly to the cultured macrophages along with 5 μg/ml LPS. Twenty four hours post-stimulation, cells were analyzed by FACS. (A) Control macrophages without Bioparticles®. Panels B-E show plots of cells incubated with Bioparticles® (B) Transduced Magi-CXCR4 (non-phagocytic cell culture), (C) nontransduced, (D) GFP-alone, and (E) XHR macrophages. These data are representative of triplicate experiments.
Due to their role in immunity and inflammatory response, macrophages secrete and respond to a number of important cytokines that include IL-1 and TNF-α. To determine if siRNA transgenic macrophages retained their functional capacity to secrete these cytokines at normal levels, they were stimulated with LPS. Levels of released cytokines were measured by ELISA. No significant differences were seen in levels of IL-1 and TNF-α cytokine secretion among the transgenic and control cell types (Fig. 6A and 6B). Basal levels of cytokine production were also detected without LPS stimulation with no differences seen between cell types (data not shown). Collectively the above data showed that coreceptor siRNA transgenic macrophages were phenotypically and functionally normal.
Figure 6 XHR transgenic macrophages secrete normal levels of the cytokines IL-1 and TNFα: Control nontransduced, GFP-alone, and XHR macrophages were stimulated with 5 μg/ml LPS. On days 1, 2, and 3 post-stimulation, supernatants were collected and assayed by ELISA for cytokine secretion of (A) IL-1 and (B) TNFα. Experiments were done in triplicate.
Discussion
Down regulation of the major HIV-1 coreceptors CXCR4 and CCR5 in virus susceptible cells is a promising approach to prevent viral entry and establishment of productive infection. As noted above, targeting both coreceptors simultaneously will have the added advantage of protecting cells from both X4 and R5 tropic viruses as well as dual tropic strains. In the present studies we have shown that a bispecific lentiviral vector was effective in transducing the respective siRNAs targeted to these coreceptors into primary CD34+ hematopoietic progenitor cells which can give rise to all the blood cell lineages including macrophages, T cells, and dendritic cells.
Since siRNAs are new tools being used for genetic manipulation, it is necessary that they be systematically evaluated in a stem cell setting for their long range utility in protecting end stage differentiated cells such as macrophages. Recent studies have demonstrated that some siRNA constructs may have off target effects [46]. This may adversely affect cell differentiation pathways. Our results have demonstrated that mature macrophages could be derived from lentivirally transduced shRNAs targeting both CXCR4 and CCR5. No significant differences were found in the yields of macrophages from control non-transduced, control GFP-alone vector, and the bispecific shRNA vector transduced CD34+ cells when cultured in cytokine media permitting cell differentiation. This suggests that the respective shRNAs did not interfere with the lineage specific differentiation of gene transduced CD34+ cells into macrophages.
The transgenic macrophages showed significant down regulation of the respective targeted coreceptors CXCR4 and CCR5. Thus, differentiated cells retained functional shRNAs that were effective against their respective target mRNAs. When challenged with HIV-1 in vitro they showed marked resistance to infection with both X4 and R5 tropic viral strains. Most primary infections with HIV-1 are believed to be caused by R5 tropic HIV-1 as it is transmitted with relative ease with macrophages as the initial in vivo target. During disease progression, X4 tropic viruses are believed to emerge. However recent studies showed that primary X4 HIV-1 isolates could also infect macrophages obtained from human tissue establishing that initial infection of these cells in vivo is not confined to R5 strains [51]. Therefore, protecting macrophages against both R5 and X4 tropic viruses is essential to prevent initial viral infection. Thus, the bispecific lentiviral vector harboring both CXCR4 and CCR5 shRNAs, described here, would be ideal in preventing HIV-1 infection at the cell entry stage.
A requirement for successful HIV-1 gene therapy is for transgenic virus resistant cells to be phenotypically and functionally normal to maintain and restore the body's immunological function. Accordingly, transgenic macrophages were evaluated to determine if they met these criteria. Although the levels of coreceptor expression diminished substantially as a result of shRNA targeting, phenotypic analyses of shRNA transgenic macrophages showed that they were otherwise phenotypically normal. This was shown by the comparable levels of CD14 and CD4 cell surface markers for both control cells and shRNA transgenic macrophages. Levels of the MHC class II molecule HLA-DR were also found to be normal. Upregulation of the costimulatory molecule B7.1 in response to LPS stimulation was comparable between shRNA transgenic and control vector containing cells. Furthermore, phagocytic functions were also found to be normal. To analyze the critical function of macrophages in secreting cytokines during the inflammatory response, the levels of IL-1 and TNF-α secretion were analyzed. Our results demonstrated that the expression of CXCR4 and CCR5 shRNAs and the subsequent downregulation of these chemokine receptors had no apparent effect on IL-1 or TNF-α secretion. These data collectively suggest that phenotypically and functionally normal macrophages could be obtained from CD34+ cells lentivirally transduced with CXCR4 and CCR5 shRNA constructs. These results establish for the first time that simultaneous knock down of both the chemokine receptors CXCR4 and CCR5 have no apparent adverse effects on macrophage differentiation, phenotype or function.
The above data showed the efficacy of this bispecific shRNA construct in deriving HIV-1 resistant macrophages in vitro in a stem cell setting. Further preclinical testing of this construct is needed in vivo to determine its suitability for use in the human. The SCID-hu mouse model that harbors a functional human thymus permits evaluation of vector transduced CD34+ cells to determine their capacity to give rise to mature T cells. The transgenic T lymphocytes so derived could be assessed for their functionality and viral resistance as we have shown previously [29]. Adverse effects are not expected by the stable knock down of CCR5 in vivo as it was previously documented in many studies that individuals harboring a 32 bp deletion in the CCR5 gene do not exhibit any immunological abnormalities [42-44]. However, stable CXCR4 knock down may have possible side effects in a stem cell setting due to its role in cell homing [52]. Therefore, a systematic evaluation of the CCR5 and CXCR4 bispecific construct in vivo in the SCID-hu mouse model is necessary to determine its efficacy and possible toxicity in differentiated T cells prior to its evaluation in human subjects. Such studies are currently underway.
Conclusion
Stable simultaneous knock down of both the coreceptors CCR5 and CXCR4 is necessary to prevent HIV-1 infection at the entry level by both R5 and X4, as well as dual tropic viral strains. Our present studies have demonstrated for the first time that a bispecific lentiviral vector could be used to stably deliver shRNAs targeted to both CCR5 and CXCR4 coreceptors into CD34+ hematopoietic progenitor cells and derive transgenic macrophages. Stable down regulation of both the coreceptors was achieved in transgenic macrophages which displayed marked resistance to HIV-1 challenge in vitro. The siRNA expressing macrophages were also found to be phenotypically and functionally normal. It is now possible to construct gene therapeutic lentiviral vectors incorporating multiple siRNAs targeted to cellular molecules that aid in HIV-1 infection. Use of these vectors in a stem cell setting shows great promise for sustained HIV/AIDS gene therapy.
Methods
Generation of CXCR4 and CCR5 bispecific siRNA lentiviral vector XHR
A third-generation lentiviral vector system was used to produce the bispecific shRNA-expressing lentiviral vector [47]. The transfer vector pHIV-7-GFP was designed to contain an anti-CXCR4 shRNA cassette under the control of the Pol-III U6 promoter and an anti-CCR5 shRNA cassette under the control of the Pol-III H1 promoter, as previously described [37]. The anti-CXCR4 shRNA targets the CXCR4 transcript at nucleotides 3–23 and the anti-CCR5 shRNA targets the CCR5 transcript at nucleotides 13–31. A depiction of this bispecific lentiviral vector along with two important cis-acting elements is shown (Fig. 1). The two cis-acting elements, namely, the central DNA flap consisting of the cPPT and CTS (to facilitate the nuclear import of the viral preintegration complex) and the WPRE (to promote nuclear export of transcripts and/or increase the efficiency of polyadenylation of transcripts), are used to enhance the performance of the vector [47,48]. To generate lentiviral vectors, 293T cells, maintained in complete DMEM containing 10% FBS, were transfected with the plasmids pCHGP-2, pCMV-Rev, pCMV-VSVG, and the appropriate transfer vector, GFP-alone or XHR, using a calcium phosphate transfection kit (Sigma-Aldrich, St. Louis, MO). Cell culture supernatants were collected at 24, 36, 48, and 60 hours post-transfection, pooled, and concentrated by ultracentrifugation. Vector titers were then analyzed on 293T cells by FACS for EGFP expression. Concentrated vector titers ranged from 8.0 × 107 to 1.5 × 108 for XHR and GFP-alone vectors, respectively.
Transduction of CD34+ hematopoietic stem cells and derivation of macrophages
CD34+ hematopoietic progenitor cells were purified from human fetal liver by selection with monoclonal antibody-conjugated immunomagnetic beads (Miltenyi Biotech, Auburn, CA)[8]. The purity of CD34+ cells was determined by FACS using a PE conjugated CD34+ antibody. The purity of cells was routinely >93% (data not shown). CD34+ cells were maintained in Iscove's modified Dulbecco's growth medium containing IL-3, IL-6, and stem cell factor (SCF) each at 10 ng/ml (R&D Systems, Minneapolis, MN) supplemented with 10% FBS. Lentiviral vector transductions were performed on 2 consecutive days at an m.o.i. of 30 in the presence of polybrene (4 ug/ml). Transduced cells were then sorted by FACS for EGFP expression and subsequently placed in semi-solid methylcellulose Methocult media (Stem Cell Technologies, Vancouver, BC, Canada) for 10–12 days to derive myeloid colonies. Total myeloid colonies were then pooled and cultured in vitro in DMEM supplemented with the cytokines M-CSF (25 ng/ml) and GM-CSF (25 ng/ml) (R&D Systems, Minneapolis, MN) for 4 days to derive mature macrophages.
Phenotypic and functional analysis of transgenic macrophages
To determine if stem cell derived anti-coreceptor shRNA transgenic macrophages were otherwise phenotypically normal, analysis of macrophage cell surface markers was performed by FACS with respective conjugated antibodies, PE-CD14 (Caltag, Burlingame, CA), PE-HLA-DR, PE-CY5-CD4, PE-CY5-CXCR4, and PE-CY5-CCR5 (BD Biosciences, San Jose, CA).
Activated macrophages up-regulate the expression of B7.1 co-stimulatory molecules upon stimulation with various stimuli. Accordingly, control non-transduced, GFP-alone, and XHR vector transduced macrophages were stimulated with LPS (5 μg/ml) (Sigma-Aldrich, St. Louis, MO). Twenty-four hours post-stimulation, macrophages were stained with PE-CY5 conjugated anti-B7.1 antibody (BD Biosciences, San Jose, CA) and analyzed by FACS. FACS analyses were performed on the Beckman Coulter Epics XL using ADC software for analysis.
Macrophages play an important role in the immune system as phagocytes. To determine if XHR transgenic macrophages retained the ability to phagocytose foreign material, a phagocytosis assay utilizing tetramethylrhodamine fluorescently labeled E. coli Bioparticles® (Invitrogen, Carlsbad, CA) were used. To the cell culture media, 5 ug/ml of LPS and 5 ug/ml of E. coli particles were added. Twenty-four hours post-addition, cells were analyzed by FACS. Transduced Magi-CXCR4, maintained as previously described [49,50], were used as a non-phagocytic cell control. Bioparticles® were detected in the PE (FL2) channel for FACS analysis.
Transgenic macrophages were also analyzed for the secretion of two major cytokines, IL-1 and TNF-a. Macrophages were stimulated with 5 ug/ml of LPS. On days 1, 2, and 3 post-stimulation, cell culture supernatant samples were collected and analyzed by a Quantikine® ELISA kit (R&D Systems, Minneapolis, MN). Non-stimulated supernatants were also analyzed for basal levels of cytokine secretion.
HIV-1 Challenge of CXCR4 and CCR5 siRNA Transgenic Macrophages
To determine if the stable down regulation of CXCR4 and CCR5 conferred resistance to HIV-1 infection in CD34+ derived macrophages, cells were challenged with X4 (NL4-3) or R5 (BaL-1) tropic strains of HIV-1. Both NL4-3 and BaL-1 challenge experiments were carried out at an m.o.i. of 0.01 for 2 hours in the presence of polybrene (4 ug/ml). Viral supernatants were collected on various days post-infection for p24 antigen ELISAs. To quantify viral p24 levels, a Coulter-p24 kit (Beckman Coulter, Fullerton, CA) was used.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
JA performed all experiments. RA was responsible for the overall experimental design and implementation of the project.
Acknowledgements
Work reported here was supported by NIH grants AI50492 and AI057066 to R.A. This work has also been facilitated by the infrastructure and resources provided by the Colorado Center for AIDS Research Grant P30 AI054907. We thank Leila Remling for CD34 cell purifications, Karen Helms for help with FACS sorting and Mayur Tamhane for critically reading the manuscript. We thank NIH AIDS Research and Reference Reagents Program for providing many reagents and cell lines used in this work.
==== Refs
Malim MH Freimuth WW Liu J Boyle TJ Lyerly HK Cullen BR Nabel GJ Stable expression of transdominant rev protein in human T cells inhibits Human Immunodeficiency Virus replication J Exp Med 1992 176 1197 1201 1402661 10.1084/jem.176.4.1197
Bonyhadi ML Moss K Voytovich A Auten J Kalfoglou C Plavec I Forestell S Su L Bohnlein E Kaneshima H RevM10-expressing T cells derived in vivo from transduced human hematopoietic stem-progenitor cells inhibit human immunodeficiency virus replication J Virol 1997 71 4707 4716 9151864
Ding SF Lombardi R Nazari R Joshi S A combination anti-HIV-1 gene therapy approach using a single transcription unit that expresses antisense, decoy, and sense RNAs, and transdominant negative mutant Gag and Env proteins Front Biosci 2002 7 a15 28 11815282
Michienzi A Li S Zaia JA Rossi J A nucleolar TAR decoy inhibitor of HIV-1 replication Proc Natl Acad Sci USA 2002 99 14047 14052 12376617 10.1073/pnas.212229599
Akkina R Banerjea A bai J Anderson J Li MJ Rossi J siRNAs, ribozymes, and RNA decoys in modeling stem cell-based gene therapy for HIV/AIDS Anticancer Res 2003 23 1997 2006 12894572
Bahner I Kearns K Hao QL Smogorzewska EM Kohn DB Transduction of human CD34+ hematopoietic progenitor cells by a retroviral vector expressing an RRE decoy inhibits human immunodeficiency virus type 1 replication in myelomonocytic cells produced in long-term culture J Virol 1996 70 4352 4360 8676458
Lisziewicz J Sun D Smythie J Lusso P Lori F Louie A Markham P Rossi J Reitz M Gallo RC Inhibition of human immunodeficiency virus type 1 replication by regulated expression of a polymeric tat activation response RNA decoy as a strategy for gene therapy in AIDS Proc Natl Acad Sci USA 1993 90 8000 8004 8367455
Bai J Gorantla S Banda N Cagnon L Rossi J Akkina R Characterization of anti-CCR5 ribozyme-transduced CD34+ hematopoietic progenitor cells in vitro and in a SCID-hu mouse model in vivo Mol Therapy 2000 1 244 254 10.1006/mthe.2000.0038
Bai J Rossi J Akkina R Multivalent anti-CCR5 ribozymes for stem cell-based HIV type 1 gene therapy AIDS Res Hum Retroviruses 2001 17 385 399 11282007 10.1089/088922201750102427
Bai J Banda N Lee NS Rossi J Akkina R RNA-based anti-HIV-1 gene therapeutic constructs in SCID-hu mouse model Mol Therapy 2002 6 770 782 10.1006/mthe.2002.0800
Cagnon L Rossi J Down regulation of the CCR5 beta-chemokine receptor and inhibition of HIV-1 infection by stable VA1-ribozyme chimeric transcripts Anti Nucl Acid Drug Dev 2000 10 251 261 10.1089/108729000421439
Feng Y Leavitt M Tritz R Duarte E Kang D Mamounas M Gilles P Wong-Staal F Kennedy S Merson J Yu M Barber JR Inhibition of CCR5-dependent HIV-1 infection by hairpin ribozyme gene therapy against CC-chemokine receptor 5 Virol 2000 276 271 278 10.1006/viro.2000.0536
Fire A xu S Montgomery MK Kostas SA Driver SE Mello CC Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans Nature 1998 391 806 811 9486653 10.1038/35888
Hannon GJ RNA Interference Nature 2002 418 244 251 12110901 10.1038/418244a
Sharp P RNA interference-2001 Genes Dev 2001 15 485 490 11238371 10.1101/gad.880001
Elbashir SM Harborth J Lendeckel W Yalcin A Weber K Tuschl T Duplexes of 21-nucleotide RNAs mediate RNA interference in cultured mammalian cells Nature 2001 411 494 498 11373684 10.1038/35078107
Lee NS Dohjima T Bauer G Li H Li M Ehsani A Salvaterra P Rossi J Expression of small interfering RNAs targeted against HIV-1 rev transcripts in human cells Nat Biotechnol 2002 20 500 505 11981565
Song E Lee S Dykxhoorn DM Novina C Zhang D Crawford K Cerny J Sharp PA Leiberman J Manjunath N Shankar P Sustained small interfering RNA- mediated human immunodeficiency virus type 1 inhibition in primary macrophages J Virol 2003 77 7174 7181 12805416 10.1128/JVI.77.13.7174-7181.2003
Novina CD Murray MF Dykxhoorn DM Beresford PJ Riess J Lee S Collman RG Lieberman J Shankar P Sharp PA siRNA-directed inhibition of HIV-1 infection Nat Med 2002 8 681 686 12042777
Jacque J Triques K Stevenson M Modulation of HIV-1 replication by RNA interference Nature 2002 418 435 438 12087358 10.1038/nature00896
Martinez MA Gutierrez A Armand-Ugon M Blanco J Parera M Gomez J Clotet B Este JA Suppression of chemokine receptor expression by RNA interference allows for inhibition of HIV-1 replication AIDS 2002 16 2385 2390 12461411 10.1097/00002030-200212060-00002
Coburn GA Cullen BR Potent and specific inhibition of human immunodeficiency virus type-1 replication by RNA interference J Virol 2002 76 9225 9231 12186906 10.1128/JVI.76.18.9225-9231.2002
Anderson J Banerjea A Planelles V Akkina R Potent suppression of HIV type 1 infection by a short hairpin anti-CXCR4 siRNA AIDS Res and Hum Retroviruses 2003 19 699 706 13678472 10.1089/088922203322280928
Anderson J Banerjea A Akkina R Bispecific short hairpin siRNA constructs targeted to CD4, CXCR4, and CCR5 confer HIV-1 resistance Oligonucleotides 2003 13 303 312 15000821 10.1089/154545703322616989
Capodici J Kariko K Weissman D Inhibition of HIV-1 infection by small interfering RNA-mediated RNA interference J Immunol 2002 169 5196 5201 12391237
Haasnoot PCJ Cupac D Berkhout B Inhibition of virus replication by RNA interference J Biomed Sci 2003 10 607 616 14576463 10.1159/000073526
Lee MM Coburn G McClure MO Cullen BR Inhibition of human immunodeficiency virus type 1 replication in primary macrophages by using tat- or CCR5-specific small interfering RNAs expressed from a lentivirus vector J Virol 2003 77 11964 11972 14581533 10.1128/JVI.77.22.11964-11972.2003
Li M Bauer G Michienzi A Yee J Lee NS Kim J Li S Castanotto D Zaia J Rossi J Inhibition of HIV-1 infection by lentiviral vectors expressing Pol III- promoted anti-HIV RNAs Mol Therapy 2003 8 196 206 10.1016/S1525-0016(03)00165-5
Banerjea A Li M Bauer G Remling L Lee NS Rossi J Akkina R Inhibition of HIV-1 by lentiviral vector-transduced siRNAs in lymphocytes diferentiated in SCID-hu mice and CD34+ progenitor cell-derived macrophages Mol Therapy 2003 8 62 71 10.1016/S1525-0016(03)00140-0
Ketteler R Glaser S Sandra O Martens UM Klingmuller U Enhancedtrnsgene expression in primitive hematopoietic progenitor cells and embryonic stem cells efficiently trnsduced by optimized retroviral hybrid vectors Gene Therapy 2002 9 477 487 11948372 10.1038/sj.gt.3301653
An DS Koyanagi Y Zhao J Akkina R Bristol G Yamamoto N Zack JA Chen ISY High-efficiency transduction of huma lymphoid progenitor cells and expression in differentiated T cells J Virol 1997 71 1397 1404 8995665
Mautino MR Morgan RA Gene therapy of HIV-1 infection using lentiviral vectors expressing anti-HIV-1 genes AIDS Patient Care and STDs 2002 16 11 26 11839215 10.1089/108729102753429361
Butticaz C Ciuffi A Munoz M Thomas J Bridge A Pebernard S Iggo R Meylan P Telenti A Protection from HIV-1 infection of primary CD4 T cells by CCR5 silencing is effective for the full spetrum of CCR5 expression Antiviral Therapy 2003 8 373 377 14640383
Qin X An DS Chen ISY Baltimore D Inhibiting HIV-1 infection in human T cells by lentiviral-mediated delivery of small interfering RNA against CCR5 Proc Natl Acad Sci USA 2003 100 183 188 12518064 10.1073/pnas.232688199
Cordelier P Morse B Strayer DS Targeting CCR5 with siRNAs: Using Recombinant SV40-Derived Vectors to Protect Macrophages and Microglia from R5-Tropic HIV Oligonucletides 2003 13 281 294 10.1089/154545703322616961
Zhou N Fang J Mukhtar M Acheampong E Pomerantz RJ Inhibition of HIV-1 fusion with small interfering RNAstargeting the chemokine coreceptor CXCR4 Gene Therapy 2004 11 1703 1712 15306840 10.1038/sj.gt.3302339
Anderson J Akkina R HIV-1 resistance conferred by siRNA cosuppression of CXCR4 and CCR5 coreceptors by a bispecific lentiviral vector AIDS Res Therapy 2005 2 1 12 10.1186/1742-6405-2-1
Boden D Pusch O Lee F Tucker L Ramratnam B Human immunodeficiency virus type 1 escape from RNA interference J Virol 2003 77 11531 11535 14557638 10.1128/JVI.77.21.11531-11535.2003
Das AT Brummelkamp TR Westerhout EM Vink M Madiredjo M Bernards R Berkhout B Human Immunodeficiency virus type 1 escapes from RNA interference-mediated inhibition J Virol 2004 78 2601 2605 14963165 10.1128/JVI.78.5.2601-2605.2004
Bieniasz PD Cullen BR Chemokine receptors and Human Immunodeficiency Virus infection Front in Biosci 1998 3 44 58
Berger EA Murphy PM Farber JM Chemokine receptors as HIV-1 coreceptors: Roles in viral entry, tropism, and disease Annu Rev Immunol 1999 17 657 700 10358771 10.1146/annurev.immunol.17.1.657
Liu R Paxton W Choe S Ceradini D Martin S Horuk R MacDonald M Stuhlman H Koup R Landau N Homozygous defect in HIV-1 coreceptor accounts for resistance of some multiply exposed individuals to HIV-1 infection Cell 1996 86 267 377
Huang Y Paxton WA Wolinsky SM Neumann AU Zhang L He T Kang S Ceradini D Jin Z Yazdanbakhsh K Kunstman K Erickson D Dragon E Landau NR Phair J Ho DD Koup RA The role of a mutant CCR5 allele in HIV-1 transmission and disease progression Nat Med 1996 2 1240 1243 8898752 10.1038/nm1196-1240
Naif HM Cunningham AL Alali M Li S Nasr N Buhler MM Schols D Clercq E Stewart G A human immunodeficiency virus type 1 isolate froman infected person homozygous for CCR5Δ32 exhibits dual tropism by infecting macrophages and MT2 cells via CXCR4 J Virol 2002 76 3114 3124 11884536 10.1128/JVI.76.7.3114-3124.2002
Ma Q Jones D Borghesani PR Segal RA Nagasawa T Kishimoto T Bronson RT Springer TA Impaired B-lymphopoiesis, myelopoiesis, and derailed cerebellar neuron migration in CXCR4- and SDF-1-deficient mice Proc Natl Acad Sci USA 95 9448 9453 9689100 10.1073/pnas.95.16.9448
Scacheri PC Rozenblatt-Rosen O Caplen NJ Wolfsberg TG Umayam L Lee JC Hughes CM Shanmugam KS Bhattacharjee A Meyerson M Collins FS Short interfering RNAs can induce unexpected and divergent changes in the levels of untargeted proteins in mammalian cells Proc Natl Acad Sci USA 2004 101 1892 1897 14769924 10.1073/pnas.0308698100
Yam P Li S Wu J Hu J Zaia J Yee J Design of HIV-1 vectors for efficient gene delivery into human hematopoietic cells Mol Therapy 2002 6 770 782 10.1006/mthe.2002.0800
Ailles LE Naldini L Trono D HIV-1 Derived Lentiviral Vectors Lentiviral Vectors 2002 Berlin: Springer-Verlag 31 48
Kimpton J Emerman M Detection of replication-competent and pseudotyped human immunodeficiency virus with a sensitive cell line on the basis of activation of an integrated β-galactosidase gene J Virol 1992 66 2232 2239 1548759
Vodicka MA Goh WC Wu LI Rogel ME Bartz SR Schweickart VL Raport CJ Emerman M Indicator cell lines for detection of primary strains ofhuman and simian immunodeficiency viruses Virol 1997 233 193 198 10.1006/viro.1997.8606
Jayakumar P Berger I Autschbach F Weinstein M Funke B Verdin E Goldsmith MA Keppler OT Tissue-Resident Macrophages Are Productively Infected Ex Vivo by Primary X4 Isolates of Human Immunodeficiency Virus Type 1 J Virol 2005 79 5220 5226 15795306 10.1128/JVI.79.8.5220-5226.2005
Molyneaux KA Zinszner H Kunwar PS Schaible K Stebler J Sunshine MJ O'Brien W Raz E Littman D Wylie C Lehmann R The chemokine SDF1/CXCL12 and its receptor CXCR4 regulate mouse germ cell migration and survival Development 2002 130 4279 4286 12900445 10.1242/dev.00640
|
16109172
|
PMC1199620
|
CC BY
|
2021-01-04 16:36:39
|
no
|
Retrovirology. 2005 Aug 18; 2:53
|
utf-8
|
Retrovirology
| 2,005 |
10.1186/1742-4690-2-53
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-771603365510.1186/1465-9921-6-77ResearchVitamin A deficiency alters the pulmonary parenchymal elastic modulus and elastic fiber concentration in rats McGowan Stephen E [email protected] Erika J [email protected] Amey J [email protected] Department of Veterans Affairs Research Service and Department of Internal Medicine, Roy A. and Lucille J Carver College of Medicine, University of Iowa, Iowa City, IA, USA2005 20 7 2005 6 1 77 77 1 2 2005 20 7 2005 Copyright © 2005 McGowan et al; licensee BioMed Central Ltd.2005McGowan et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Bronchial hyperreactivity is influenced by properties of the conducting airways and the surrounding pulmonary parenchyma, which is tethered to the conducting airways. Vitamin A deficiency (VAD) is associated with an increase in airway hyperreactivity in rats and a decrease in the volume density of alveoli and alveolar ducts. To better define the effects of VAD on the mechanical properties of the pulmonary parenchyma, we have studied the elastic modulus, elastic fibers and elastin gene-expression in rats with VAD, which were supplemented with retinoic acid (RA) or remained unsupplemented.
Methods
Parenchymal mechanics were assessed before and after the administration of carbamylcholine (CCh) by determining the bulk and shear moduli of lungs that that had been removed from rats which were vitamin A deficient or received a control diet. Elastin mRNA and insoluble elastin were quantified and elastic fibers were enumerated using morphometric methods. Additional morphometric studies were performed to assess airway contraction and alveolar distortion.
Results
VAD produced an approximately 2-fold augmentation in the CCh-mediated increase of the bulk modulus and a significant dampening of the increase in shear modulus after CCh, compared to vitamin A sufficient (VAS) rats. RA-supplementation for up to 21 days did not reverse the effects of VAD on the elastic modulus. VAD was also associated with a decrease in the concentration of parenchymal elastic fibers, which was restored and was accompanied by an increase in tropoelastin mRNA after 12 days of RA-treatment. Lung elastin, which was resistant to 0.1 N NaOH at 98°, decreased in VAD and was not restored after 21 days of RA-treatment.
Conclusion
Alterations in parenchymal mechanics and structure contribute to bronchial hyperreactivity in VAD but they are not reversed by RA-treatment, in contrast to the VAD-related alterations in the airways.
Elastinretinoic acidemphysemabronchial hyperreactivitycholinergic
==== Body
Background
Previous studies have shown that vitamin A deficiency (VAD) in rats is associated with a decrease in gas-exchange surface area, a decrease in the bronchial elastic fiber density, and with an increase in airway responsiveness to cholinergic agents [1,2]. Although VAD is uncommon in economically developed countries, it remains an important public health problem in the developing world particularly in children during the first seven years of life, when pulmonary alveolarization occurs [3]. Vitamin A and its active metabolite retinoic acid influence alveolar development and restoration, however the mechanisms responsible for these effects remain poorly understood [4,5]. In our experimental model of VAD, rats do not become deficient until after the period of maximal alveolar formation, which is completed by 3 weeks of age [2,6]. During these first 3 weeks of postnatal life there is an increase in the mRNA for tropoelastin, the soluble precursor of cross-linked elastin, which is an important determinant of the mechanical properties of the lung parenchyma and airways [7]. Once it is cross-linked, elastin normally undergoes very little turnover, although this does occur in pathological conditions such as emphysema [6,8].
In order to better identify the mechanisms that are responsible for airway hyperreactivity in VAD rats, with respect to morphological and biochemical characteristics of the pulmonary elastic fiber network, we evaluated the mechanical properties of the lung parenchyma that are most involved in regulating small airway diameter. Airway responsiveness to cholinergic agents is influenced by airway-parenchymal interactions [9]. The elastic fibers in the walls of alveoli and alveolar ducts, which form a continuous network with elastic fibers in the small and larger airways, are an important structural determinant of these interactions [10,11]. The elastic fibers within the airway connect the epithelial basement membrane to the smooth muscle layer [11]. Fibers in the adventitia that surrounds the airway smooth muscle are connected to parenchymal elastic fibers located in the surrounding alveoli and alveolar ducts. The contractile cells in the alveolar ducts may also influence airway smooth muscle contraction because contractile cells in the two locations are connected through the intervening elastic fiber network [11]. Physiological measurements of the elastic modulus of the lung are sensitive to alterations in both the airways and the parenchyma [12]. For an isotropic material, the ability to resist volume and shape distortion, respectively, is described by the bulk modulus (k, which is proportional to the ability to resist uniform expansion) and the shear modulus (μ, which is proportional to the ability to resist a small isovolume shape distortion). The lung is more constrained in volume expansion than in shape distortion, and k increases exponentially with volume whereas μ increases arithmetically [13]. There are three mechanisms whereby the lung resists deformation: (a) altering the spacing between microstructural elements, (b) altering the orientation of the microstructural elements, and (c) stretching of the microstructural elements [12]. Any or all of these three factors may be affected if there are abnormalities of the elastic fiber network. In pulmonary emphysema there are changes in all three mechanisms. Dilated alveoli and alveolar ducts increase the spacing between elastic fibers, elastic fibers are disarrayed and are abnormally connected, and the remaining alveolar walls and ducts are stretched by dilation. The elastic modulus of the lung parenchyma may also be altered in VAD rats, which have fewer and dilated gas exchange units compared to the lungs of VAS rats [1]. Because the inhalation of aerosolized cholinergic agents distorts the lung parenchyma producing inter-dispersed regions of localized hyperinflation and atelectasis, one would predict that alterations in the elastic modulus would be accentuated after cholinergic administration [14]. We hypothesized that because of parenchymal distortion and localized hyperinflation, cholinergic administration would produce a larger increase in the bulk modulus of VAD compared to vitamin A sufficient (VAS) rat lungs. To address this hypothesis we have characterized the effects of VAD on parenchymal mechanics and elastic fiber architecture. We have studied elastic fiber length per unit volume of lung, elastin production, and measured the elastic modulus of the lung parenchyma in VAS and VAD rats before and after the administration of CCh. We further hypothesized that if the elastic fiber network was a major determinant of the bulk and shear moduli, then restoration of the elastic fiber network may restore the elastic moduli to values that are similar to those in VAS rats. Therefore, we administered retinoic acid (RA) to determine whether reversing the tissue effects of VAD would coordinately reverse abnormalities in the elastic fibers and in the bulk and shear moduli. The elastic fiber length per unit volume was decreased in VAD rat lungs and may have contributed to the observed differences in shear modulus. However, other architectural modifications accounted for the observed differences in the bulk modulus in VAD compared to VAS rats.
Methods
Production of Vitamin A Deficiency
Specific pathogen-free female Lewis rats were obtained from Harlan-Sprague Dawley (Madison, WI). All animals were maintained in HEPA-filtered cages and sentinel animals were used to establish that the colony remained specific-pathogen free. The protocol was approved by the animal use committees at the Veterans Affairs Medical Center and the University of Iowa. The rats were weaned at postnatal day 21 and placed on a VAD diet-modified (catalog number 96022, ICN Corp., Aurora, OH), for 7 to 10 weeks to achieve vitamin A deficiency [15]. Vitamin A sufficient rats were littermates of the VAD animals or age-matched females were purchased from Harlan-Sprague Dawley. The general health of the VAD rats was monitored and the VAD animals were used prior to the onset of weight loss or keratitis. We have previously shown that this protocol consistently produces vitamin A deficiency [1]. The onset of VAD was identified by the cessation of weight gain which occurred earlier than in females who were fed the control diet. When the VAD rats stopped gaining weight, they received 25 μg of retinyl acetate that was administered orally at weekly intervals to prevent weight loss and a generalized nutritional deficiency. Twenty-five micrograms of retinoic acid (RA), in safflower oil, were administered orally daily for 12 or 21 days to some rats to determine whether this reversed the effects of VAD. Supplementation of VAD rats with RA for 12 days is sufficient to completely restore the expression of retinaldehyde dehydrogenase, a retinoid responsive gene [16].
Analysis of the elastic modulus of the distal lung
Rats were anesthetized, the trachea was cannulated with a 14 gauge catheter, and the animals inhaled 100% oxygen for 6 minutes. The tracheal cannula was plugged, a medium sternotomy and laparotomy were performed, and the heart was allowed to pump for 5 minutes to induce total pulmonary atelectasis. After exposing the heart, the lungs were perfused with 15 ml of 137 mM NaCl, 8 mM Na2HPO4, 2.7 mM KCl, 1.5 mM KH2PO4, pH 7.4 (PBS) to clear the pulmonary circulation. The trachea, mediastinum, heart, and diaphragm were excised en bloc and the preparation was immersed in PBS. The lungs were inflated with 0.2 ml of air every 5 s over approximately 3 minutes to a constant pressure of 25 cm and then allowed to collapse to 0 cm H2O pressure, and the inflation and deflation were repeated once. The deflation volume-pressure curve was assessed (method described subsequently) both before and after the intratracheal administration of 16 mg/ml of carbamylcholine (CCh) during ventilation of the lungs with a tidal volume of 0.3 ml for 90 s using a DeVilbiss AeroSonic ultrasonic nebulizer [1]. In each case, the lungs were inflated once to 25 cm H2O pressure and deflated to 0 cm prior to the inflation phase of the volume-pressure analysis.
Studies were performed to evaluate the elastance of the distal lung by ventilating the excised lungs at a small tidal volume (0.3 ml) with a volume-cycled rodent respirator (Inspra, Harvard Apparatus, Holliston, MA). Flow was measured by a pneumotachograph attached between the mechanical ventilator and the endotracheal tube, and volume was calculated by integrating the flow. Tracheal pressure was measured continuously and data were acquired and sampled at 50 Hz using a RSS 100HR Research Pneumotach system (Hans Rudolph, Kansas City, MO). Ventilating at 0.3 ml minimized minimized air-trapping. The resistance (R) and elastance (E) were calculated from the equation PL = RLQ + EV + K where K is a parameter reflecting the end-expiratory pressure, Q = flow, and V = volume [17].
We followed the methods that have been described by Salerno and Ludwig for evaluating the bulk modulus (k) and the shear modulus (μ) of rats [18]. Bulk modulus (k) is expressed by the equation k = V·dP/dV and changes with the absolute volume of the lung. The k was calculated from the incremental changes in P and V, over the 0.44 s that were required for the ventilator to deliver 0.3 ml, and expressed as the mean of 5 inflations. The shear modulus was calculated from the equation G/2wD = μ/[1-(3k-2μ)/2(3k+μ)] where G = the lung's resistive force against the displacement, w = the displacement of the punch, D = diameter of the punch, k = bulk modulus and μ = shear modulus [19]. The end tidal volumes of the preparations were controlled by adjusting the positive end expiratory pressure (PEEP) to 3 cm or 8 cm. The shear modulus was measured by the punch-indentation test (using a punch with a diameter of 0.45 cm and advancing it by 0.5 mm increments) at the same inflation volumes by adjusting the airway pressure to 3 or 8 cm using a biased flow of air, an adjustable valve and a pressure transducer [18]. The volume at atmospheric pressure was assessed by volume displacement [20]. The absolute volume of the lung at 3 cm H2O was calculated by adding the volume that yielded this pressure during the volume-pressure maneuver to the residual volume. The absolute volume at 3 cm H2O pressure did not change with the administration of CCh. The bulk modulus and shear modulus were analyzed at 3 cm and 8 cm H2O prior to administration of CCh and at 3 cm after CCh administration. All of the measurements were completed within 90 minutes after euthanizing the animals.
Analysis of elastin
The right lungs of the rats that were used for the analyses of elastic moduli were frozen in liquid nitrogen, without separation of the bronchovascular bundles from the parenchyma. A portion of the lung was extracted with chloroform and methanol, dried under vacuum and weighed (the dry-defatted weight) [21]. The dried lung tissues were used to isolate elastin by extracting with 0.1 M NaOH at 98°[4]. The washed, alkali-resistant insoluble elastin residue was hydrolyzed for 20 hours in 6 N HCl under vacuum and the HCl was removed by evaporation under a stream of nitrogen. The amino acid composition of the hydrolysate was analyzed using reverse-phase HPLC following a procedure that has been described previously [4]. The elastin contents were normalized to the dry-defatted weight of the lungs.
Analysis of the pressure-volume characteristics of VAS and VAD lungs
A deflation volume-pressure curve was generated for the excised lungs before and immediately after exposure to carbamylcholine. The lung was inflated to 25 cm H2O pressure over 90 s and deflated in 0.5 ml increments using a Harvard PHD 2000 programmable syringe pump, pausing for 12 seconds at each volume before recording the pressure. Pressure was measured using a Validyne Model DP45-28 (Validyne, Northridge, CA) pressure transducer. The signal was conditioned by a Validyne carrier-demodulator and sent to a strip-chart recorder. The transducer was calibrated using a water manometer. The volume-pressure data that were obtained at volumes from 80% to 30% total lung capacity were subjected to a double logarithmic transformation. Linear regression analysis was applied to the normalized data to calculate the slope of the deflation volume-pressure curve. [22].
The effects cholinergic administration on hysteresis in VAS and VAD rats was analyzed in a separate set of experiments. The thoracic cavity was entered by a median sternotomy and the chest wall was widely retracted. The abdominal contents were deflected with a retractor, the rats were euthanized by exsanguination, and the lungs were perfused with heparinized PBS. The lungs were inflated with 10 ml of air and allowed to return to residual volume, and the inflation and deflation were repeated once. Then the lungs were inflated in one ml increments up to 10 ml and then deflated in one ml increments, pausing for 12 seconds at the end or each increment prior to the pressure measurement. Methacholine (16 mg/ml) was delivered as an aerosol for 90 sec and the lungs were inflated once with 10 ml of air and allowed to return to residual volume. Then the incremental inflation-deflation maneuver was repeated to assess the effects of methacholine. The tracheal pressure was plotted at each increment and the hysteresis ratio was calculated using Microsoft Excel and a specially designed macro (Huvard Research and Consulting, Virginia Commonwealth University) [23].
Elastic fiber concentration in respiratory airspaces
Left lungs were fixed at 20 cm H2O pressure for 16 hours at 4° in 4% paraformaldehyde and the volumes were determined by displacement [24]. The mean volumes of the left lungs did not vary significantly according to retinoid status and were 3.63 ± 0.18, 3.62 ± 0.12, and 3.83 ± 0.09 for VAS, VAD and VAD + 12d RA, respectively (n = 5 for each group). The fixed lungs were cut into sagittal slices of approximately 1.5 mm thickness. The slices were cut into strips of approximately 3 × 2 mm. The lungs were cut prior to dehydration, because it was difficult to uniformly dehydrate them. Therefore the displacement volumes were not measured after dehydration. The strips were then dehydrated in progressively increasing concentrations of ethanol (from 50 to 100%). The ethanol was replaced with 2 exchanges of LR-White resin, the strips were placed in gelatin capsules, and the LR-White was allowed to polymerize overnight at 60°. Sections were cut at a nominal thickness of 2 μm using a diamond-titanium knife and the actual thickness was determined using a stylus profilometer. Sections were hydrated, stained with orcein-hematoxylin, dehydrated and mounted in resinous medium. The intersections of alveolar septal elastic fibers with a test line were enumerated in 50 microscopic fields per section at 1000× magnification. The test line was a line spanning the width of a reticule placed in the ocular. The average number of intersections of a structure with a test line is one-half the ratio of the length to the volume [25]. Therefore the length of elastic fibers per unit volume (Lv) is equal to 2 times (average number of intersections / length of test line) times the thickness of the section. This value for elastic fiber length per unit volume is a measure of elastic fiber concentration and will be referred to as "concentration" [25]. The gas-exchange (included both alveoli and alveolar ducts) surface area was determined using previously described methods [1]. Randomly chosen paraffin blocks of the left lung were sectioned and stained with hematoxylin and eosin. One section per rat was randomly selected and 6 fields per section were photographed at 50 × at random avoiding blood vessels and airways. The photographs were uniformly enlarged, overlaid with transparent grids and analyzed using morphometric methods [26]. The volume densities of airspace and tissue were determined by point counting using a 10 by 10 grid with 100 evenly spaced points, ~42 μm apart, as described previously [27]. Mean cord lengths (Lm) were determined by counting intersections of airspace walls (including alveoli and alveolar ducts) with an array of 70 lines, each ~33 μm long [28]. The mean cord length is an estimate of the distance from one airspace wall to another airspace wall. The volume densities of the airspace and tissue, the mean cord length and the alveolar surface area were calculated as described previously [28]. Surface areas were expressed per cm3 of distal lung tissue.
Histological assessment of airway contraction
Approximately 20 minutes (the time required to measure the bulk and shear moduli) after administering the CCh (or in the absence of CCh-administration), the left lung was inflated to 16 cm H2O pressure with a stream of air, deflated to 5 cm, and then frozen in vapors of liquid nitrogen. The tissue was fixed by freeze substitution to maintain the architectural relationships that existed at the time of freezing. Carnoy's fixative (60% ethanol, 30% chloroform, and 10% acetic acid) was cooled with dry-ice and maintained overnight in a -20° freezer with excess dry-ice [14]. The following day, progressive concentrations of ethanol were substituted for the Carnoy's fixative while the lungs were maintained at -20° until 100% ethanol was reached [9]. The tissue was maintained in 100% ethanol overnight at -20° and then at 4° for 24 hours. The lungs were then embedded in paraffin, sectioned and stained with hematoxylin and eosin. Airways that contained a continuous circumference of smooth muscle and had been sectioned transversely were selected, photographed, and 35-mm slides were prepared. The 35 mm slides were digitized, the digitized images were analyzed using Image J (public domain software available at ), and the perimeter of the epithelial basement membrane, the lumen, and the inner and outer borders of the smooth muscle were traced. A stage micrometer was photographed at various magnifications and the micrometer-images were digitized using the same settings (scan resolution and enlargement) that were used for the airways. This allowed a conversion from pixels to microns. The actual area (A) of the airways that was luminal to the basement membrane was compared to the calculated area for the airway in the fully dilated (un-contracted) state (Ar). The details of the methods have been described and are predicated on the observation that the epithelial basement membrane circumference (perimeter) remains unchanged with constriction [29]. This allows one to relate all measurements to the ideally relaxed area that is contained within the circumference of the basement membrane, Ar = BM2/4π. The A/Ar is an index of the degree of airway narrowing and is influenced by both the fixation pressure and smooth muscle contraction [29,30]. Only airways with a ratio of the smallest to the largest diameter that was greater than 0.6 were used for the analysis of A/Ar. We stratified the A/Ar according to airway size because others have shown that airway diameter itself is a determinate of the contraction index [30].
Physiological assessment of lung parenchymal distortion
Immediately prior to euthanasia four VAD and four VAS rats were exposed to an aerosol of CCh for 60 seconds, whereas three VAS and three VAD rats were not exposed to CCh. The lungs were quickly removed and the left lung was inflated at 10 cm H2O pressure and fixed by freeze substitution, as described previously. Ten cm of pressure was used instead of 5 cm, because the lower inflation pressure was insufficient to provide uniform expansion, and an initial inspection of lungs fixed at 5 cm H2O suggested that the mean chord length could not be accurately determined. Paraffin embedded lungs were sectioned, 9 randomly selected fields from each lung, which contained alveoli and alveolar ducts were photographed at 25× magnification, and digitized images were prepared as described previously. The images were uniformly enlarged, overlaid with an array of lines, and the Lm was determined as previously described. To evaluate the variability of airspace size, the standard deviation of the Lm (SD Lm) was assessed for each lung. The means of the SD Lm determinations for four CCh-exposed and three unexposed lungs VAS and VAD lungs were calculated. To assess the proportion of alveolar and alveolar duct walls (as opposed to airspace) in the sections from lungs fixed at 10 cm H2O, the digitized images were subjected to uniform thresholding to separate air and tissue densities. The number of pixels that corresponded to tissue density (termed the atelectasis index or ATI) was determined for each microscopic field (the same images that were used to determine Lm) [9]. The proportion of pixels corresponding to tissue density was expressed relative to the total number of pixels in the microscopic field, which was the same for all of the images. To assess variability of the tissue density, the standard deviation of the ATI (SD ATI) was assessed for each lung. The means of the SD ATI determinations for four CCh-exposed and three unexposed lungs VAS and VAD lungs were calculated.
Statistics
The results were expressed as mean ± SEM and statistical comparisons were made using analysis of variance (ANOVA with a Student-Newman-Keuls post-hoc test). Differences were considered significant if p was less than 0.05. (n) is the number of animals in each treatment group, except for the morphometric studies in which (n) is the number of airways or lung parenchymal sections that were analyzed for each vitamin A-treatment group.
Results
VAD increases the elastance of excised lungs
The vitamin A deficient diet led to a decrease in the hepatic retinyl ester contents from 768 ± 248 nmol/g in VAS rats to 17.5 ± 5.2 nmol/g and 14.5 ± 1.9 nmol/g in VAD rats that remained unsupplemented or were supplemented with RA for 12 days, respectively, consistent with a vitamin A deficient state. The elastance of excised lungs that were ventilated at a tidal volume of 0.3 ml and 3 cm PEEP was significantly higher in VAD than in VAS rats in the absence of CCh (Figure 1). Following the administration of CCh, the elastance increased in all three categories of retinoid status. And the CCh-related increase in elastance was significantly higher for VAD and VAD rats that had received RA for 12 days than for VAS rats. These findings were consistent with our previous findings for the lungs in situ, using larger tidal volumes, except that the 12 days of RA-treatment did not lower the elastance of the excised lung to a level that was similar to that for VAS rats [1]. We next determined the effects of CCh on the bulk and shear modulus components of the elastic modulus.
Figure 1 Effects of vitamin A deficiency (VAD) on the elastance of excised lungs. After standardizing the volume history by inflating to 25 cm H2O, the excised lungs were ventilated at a tidal volume of 0.3 ml and 3 cm of PEEP. Elastance (mean ± SEM, n = 7 in each group) was calculated prior to (solid bars) and after (open bars) administration of aerosolized carbamylcholine (CCh). (#) p < 0.05, VAD compared to vitamin A sufficient (VAS), prior to CCh. (*) p < 0.05, VAD, VAD + 12 days (d) and VAD + 21 d of retinoic acid (RA) compared to VAS, after CCh. 2-way ANOVA, Student-Newman-Keuls post-hoc test.
VAD increases the elastic modulus after CCh-administration
The bulk modulus, measured at 3 cm PEEP, increased after the administration of CCh in both VAS and VAD rats, but the increase was approximately 2-fold greater in VAD rats (Figure 2A). Administration of RA for 12 or 21 days did not ameliorate the heightened CCh-mediated increase in bulk modulus, which remained significantly greater than VAS after both 12 and 21 days RA-treatment. There was a significant increase in the bulk modulus, in the absence of CCh, for VAD lungs that were treated with RA for 12 or 21 days, compared to VAS lungs. In VAD rats, the fold-increase in bulk modulus that was attributable to CCh was greater than the CCh-mediated increase that was observed in VAS rats (Figure 2B). However, VAD rats that received RA showed a smaller increase in bulk modulus after CCh compared to pre-CCh, and the fold-increases in these two groups were not significantly greater than for VAS rats. The static volumes of the lungs were not significantly altered by vitamin A-status and the increase in volume after CCh administration was only significant for VAD rats that received RA for 21 days (Figure 3). The lung volumes at 3 cm H2O did not vary among the various retinoid-treatment groups (Figure 3), so an increase in volume did not significantly contribute to the observed increase in bulk modulus in VAD rats. The volumes (including residual volume) of the lungs that had been inflated to 20 cm H2O also did not vary among retinoid treatment groups. They were 7.0 ± 0.9, 7.1 ± 0.4, and 7.2 ± 0.5 ml (mean ± SEM, n = 4) for VAS, VAD and VAD + 12 d RA, respectively
Figure 2 Bulk modulus is increased in vitamin A deficient rats (VAD). (A) Bulk modulus (mean ± SEM, n = 9 in each group) was increased (*, p < 0.05) by carbamylcholine (CCh) administration (open bars) in vitamin A sufficient (VAS) rats and VAD rats that had not received retinoic acid (RA) and VAD rats that had received RA for 12 or 21 days (d). The CCh-induced increase in bulk modulus was significantly (#, p < 0.05) higher in VAD rats that were untreated or treated for 12 or 21 d with RA, than in VAS rats. In the absence of CCh (solid bars), the bulk modulus was increased in VAD rats that had received RA for 12 or 21 d, compared to VAS rats (+, p < 0.05). (B) Comparing the ratio of bulk modulus after carbamylcholine (CCh) to before CCh, at 3 cm PEEP, showed that the CCh-induced increase in bulk modulus was significantly higher in vitamin A deficient (VAD) rats than in VAS rats (*), p < 0.05, n = 9 for each treatment group. 3-way ANOVA, Student-Newman-Keuls post-hoc test.
Figure 3 Volumes of excised lungs at 3 cm H2O did not vary with vitamin A status. In the absence of carbamycholine (CCh), residual volume (RV) was determined by volume displacement and the volume of air required to maintain 3 cm pressure was ascertained from the deflation pressure volume curve (open bars). A similar determination was made immediately after CCh-administration (hatched bars). Volumes (V, mean ± SEM, n = 8 for each vitamin A treatment group) shown are the sum of RV (volume at 0 cm) and the volume required to maintain 3 cm H2O pressure. (*) V after CCh greater than before CCh (p < 0.05, 2-way ANOVA, Student-Newman-Keuls post-hoc test).
As expected, the shear modulus increased after the administration of CCh for all categories of retinoid-status. Whereas VAD was associated with a larger increase in bulk modulus after CCh administration, the increase in shear modulus was smaller in VAD than in VAS rats (Figure 4). When measured after CCh administration, the shear modulus of the lungs of VAD rats that had received RA for 12 days was significantly smaller than that observed in VAS rats (Figure 3). In summary, these data indicate that VAD alters the mechanical properties of the lung parenchyma, and the alterations are most evident after CCh-administration. Repletion with RA for 12 or 21 days did not significantly restore the CCh-related changes in bulk modulus, although the bulk modulus in the absence of CCh was affected by RA-administration. After 21 days of RA-administration the shear modulus after CCh returned to a level that was similar to that of VAS rats.
Figure 4 Shear modulus is decreased in vitamin A deficiency (VAD). The shear modulus (mean ± SEM, n = 9 for each treatment group (the same as in Fig. 2) increased significantly after carbamylcholine (CCh), (*) p < 0.05 post-CCh (open bars) compared to pre-CCh (solid bars). (#) p < 0.05, post-CCh for VAD + 12d RA compared to post-CCh for VAS. (+) p < 0.05 VAD + 21 d RA compared to VAD + 12 d RA, post-CCh. 3-way ANOVA, Student-Newman-Keuls post-hoc test.
VAD reduces the concentration of elastic fibers and the quantity of lung elastin
The lungs of some rats from each retinoid-treatment group were fixed at 20 cm H2O inflation pressure and were dehydrated and embedded in LR-White resin, using the same methods for all of the lungs. The concentration of elastic fibers, which were detected by an orcein stain, was significantly lower in VAD than in VAS rats and administration of RA for 12 days restored the concentration of elastic fibers (Figure 5). The differences in elastic fiber concentration were not due to differences in the internal surface area. When the fiber concentration (mm fiber length /mm3 of lung) was divided by the internal surface area (mm2/mm3 of lung) of the respective lungs, the ratios of fiber length to surface area (mm/mm2) were 0.76 ± 0.06 (n = 11), 0.51 ± 0.03 (n = 9, p < 0.01 compared to VAS), and 0.92 ± 0.08 (n = 6, p < 0.01 compared to VAD) for VAS, VAD and VAD + 12d RA, respectively (1-way ANOVA). Elastin, which was resistant to hot alkali treatment, was also reduced in reduced in VAD rats, but unlike the density of elastic fibers that were visualized after orcein-staining, the elastin content was not restored by the administration of RA for 12 days (Figure 6).
Figure 5 Elastic fiber concentration (mm length / mm3 parenchyma) was decreased in VAD rats and was restored by retinoic acid (RA)-administration. The length (mean ± SEM) of elastic fibers per unit volume was decreased in lungs from VAD (n = 9 sections analyzed) rats compared to lungs from VAS rats (n = 11) that were fixed at the same pressure (*) p < 0.05, 1-way ANOVA, Student-Newman-Keuls post-hoc test. The fiber concentration in VAD rats that received RA for 12 days (VAD + 12d RA, n = 6) was significantly greater than for VAD (#), p < 0.05. 3 rats were used for each retinoid-treatment group.
Figure 6 Lung elastin contents were decreased in vitamin A deficiency (VAD). Lung elastin (mean ± SEM, n = 6 for each group), normalized to the dry-defatted lung weight was decreased in VAD compared to vitamin A sufficient (VAS) rats, which was not altered by retinoic acid (RA) treatment for 12 or 21 days. (*) p < 0.05, 1-way ANOVA. Student-Newman-Keuls post-hoc test.
Administration of RA to VAD rats increases tropoelastin mRNA
Because administering RA for 12 days increased the concentration of elastic fibers in VAD rats, we investigated the steady state-level of tropoelastin (TE) mRNA in lung and bronchial tissues that were isolated from VAS rats and VAD rats that were untreated or had received RA for 4 or 12 days. Northern analyses were preformed and the densities of the bands for tropoelastin were normalized to ribosomal phosphoprotein P-0 (RP-0), to account for inadvertent differences in the quantities of RNA that were loaded in various lanes. The results for lung and bronchial RNA shown in Figures 7A and 7B, respectively demonstrated that 12 days of RA-administration significantly increased TE mRNA in lung tissue, but not bronchial tissue. There was a trend towards an increase in TE mRNA in bronchial tissue after 4 days of RA administration (p = 0.1).
Figure 7 Tropoelastin (TE) mRNA was increased in the lung parenchyma after retinoic acid (RA) administration. Lung parenchymal (A) and bronchial (B) tissues were separated prior to RNA isolation. The filters from Northern analysis were re-probed for the constitutively expressed mRNA for ribosomal phosphoprotein P-0 (RP-0) to correct for differences in the amounts of RNA loaded. The density of TE mRNA was expressed relative to that for RP-0 for each lane, and normalized to the mean density for RNA from VAS rats within each Northern analysis. Data are mean ± SEM, n = 9 rats for each retinoid-treatment condition. Treatment with RA for 12 days (VAD + 12d RA) increased lung but not bronchial TE mRNA (*) p < 0.05, 2-way ANOVA. Treatment with RA for 4 days (VAD + 4d RA) did not significantly increase lung or bronchial TE mRNA.
VAD is associated with an increase in static lung elastance
VAD significantly increased the slope of the deflation pressure-volume curve and this was not restored by the administration of RA for 12 days (Figure 8). The slopes (ΔP/ΔV) were 1.136 ± 0014 (4), 1.297 ± 0.014 (4)*, and 1.241 ± 0.025 (4) for VAS, VAD and VAD + 12d RA, respectively. (*, VAS versus VAD) p < 0.05, 1-way ANOVA, Student-Newman-Keuls post hoc test. The effects of CCh-administration on the pressure-volume hysteresis for a representative VAS and VAD rat are shown in Figure 9A and 9B, respectively. In VAD rats methacholine-administration strikingly increased the pressure that was required to inflate the lungs compared to the effects of methacholine on VAS lungs. This rightward shift in the inflation portion of the pressure-volume curve contributed to a large CCh-mediated increase in the hysteresis of VAD (Figure 9B) compared to VAS lung (Figure 9A). This was a consistent finding in two other VAS and VAD rats, as indicated by the significant increase in the hysteresis ratio (mean ± SEM, n = 3), shown in Figure 9C.
Figure 8 Deflation pressure-volume analysis in the absence of carbamylcholine. Deflation pressure (P)-volume (V) curves are shown for four rats from each vitamin-A treatment group (mean ± SEM; VAS, vitamin A sufficient; VAD, vitamin A deficient; VAD + 12d RA, VAD treated for 12 days with RA).
Figure 9 Cholinergic administration produces a larger increase in hysteresis in VAD rats. The effects of cholinergic administration on pressure-volume hysteresis are shown for one representative VAS (A) and one representative VAD (B) rat. Solid line, prior to methacholine administration; broken line, immediately following methacholine administration. The mean ± SEM of the hysteresis ratio (C) was calculated for three VAS (open bars) and 3 VAD (checked bars) rats both before, and was significantly increased in VAD rats (p < 0.01, 1-way ANOVA, Student-Newman-Keuls post-hoc test) by cholinergic administration.
The airway contraction index was decreased in VAD rats
The airway contraction index is a morphometric assessment of reduction in airway caliber and compares the actual area internal to the epithelial basement membrane to the idealized maximal area if the bronchus was completely dilated. Therefore a smaller contraction index (A/Ar) correlates with a greater degree of luminal narrowing. Figure 10A shows the contraction index did not vary among the various retinoid treatment groups for bronchi that that had not been exposed to CCh. Figure 10B shows the contraction index for bronchi in lungs after exposure to CCh. Airways were stratified according to their diameter because the degree of contraction is dependent on the initial diameter, as well as the response to the cholinergic agent [30]. After CCh-administration, the index was significantly lower in VAD rats at both ranges of diameter than in VAS rats (Figure 10B). After 12 days of exposure to RA, the contraction index increased and was significantly higher than in untreated-VAD rats, for airways of diameter greater than 0.55 mm.
Figure 10 Airway contraction index is decreased in vitamin A deficient (VAD) rats. Airway contraction index (A/Ar, mean ± SEM, n = 9 rats for each retinoid treatment group). Solid bars are bronchi with diameters from 0.15 to 0.55 mm; open bars are bronchi with diameters greater than 0.55 mm. (A) shows that the contraction index prior to CCh-administration did not vary among the various retinoid treatment groups. After CCh-administration (B) the A/Ar was lower in VAD airways and was restored by retinoic acid (RA). (*) p < 0.05, VAD versus vitamin A sufficient (VAS). (#) p < 0.05, VAD + 12 days of RA versus VAD. Comparisons were between bronchi within the same range of diameters.
VAD accentuates the distortion of the gas-exchange region in VAD rats
Morphometric analysis of lungs from VAS and VAD rats, which had been inflated to 10 cm H2O pressure, without or immediately after exposure to CCh was performed to assess hyperinflation and atelectasis in the region of the alveoli and alveolar ducts. Representative photomicrographs of VAS and VAD lung are shown in Figure 11, which illustrates that VAD lungs (panels B and D) have more enlarged airspaces than VAS lungs (panels A and C) and that the enlargement is more pronounced after CCh administration. The results shown in Table 1 indicate that whereas the Lm was similar in CCh-unexposed VAS and VAD rats, CCh administration led to more pronounced airspace enlargement in VAD rats. This was evidenced as a larger Lm and SD Lm in VAD rats, indicating that the alveolar ducts and alveoli were more dilated with air and that the dilation was more heterogeneously distributed in the lungs of VAD rats. Whereas the percentage of alveolar and alveolar duct tissue (ATI), as opposed to air, was similar in VAS and VAD lungs after CCh-administration, there was more heterogeneity in the tissue density among different portions of the lungs of VAD rats (greater SD ATI).
Figure 11 Airway contraction and airspace distortion after carbamylcholine (CCh) administration. Lungs from vitamin A sufficient (VAS), A and C, or from vitamin A deficient (VAD), B and D, were fixed while inflated at 10 cm H2O pressure without, A and B, or with prior exposure to CCh, C and D. Alveoli and alveolar ducts were more dilated in VAD lungs particularly after CCh. Panel D also shows atelectatic areas which accompanied areas of hyperinflation in VAD lungs.
Table 1 Morphometric analysis of the gas-exchange region from carbamylcholine (CCh) exposed and unexposed rats
CCh (mg/ml) vitamin A status Lm mean SD Lm ATI mean SD ATI
0 VAS 40.15 ± 3.34 (n = 3) 4.89 20.49 ± 2.50 (n = 3) 2.78
0 VAD 40.74 ± 2.72 (n = 3) 3.23 16.69 ± 0.65 (n = 3) 1.63
12 VAS 36.10 ± 4.07 (n = 4) 5.88 20.41 ± 1.92 (n = 4) 2.81
12 VAD 44.41 ± 4.92* (n = 4) 9.84* 20.04 ± 1.12*† (n = 4) 4.15*†
Rats were exposed to carbamylcholine (CCh) or remained unexposed. Lm and ATI are expressed as mean ± SEM (n=number of rats used). Mean SD Lm was the mean of the standard deviations (SD) of the mean chord lengths (Lm) for the 9 sections, which were randomly photographed for the left lung of each rat. n = 4 lungs from CCh-exposed VAS and VAD rats. n = 3 lungs from unexposed VAS and VAD rats. The atelectasis index (ATI) was the percent of pixels that represent tissue (as opposed to air). The SD ATI was calculated from the SD of the ATI for the same 9 sections from each lung that were used to determine the Lm. (*) p < 0.05, VAS vs. VAD within CCh-treated group. (†) p < 0.05 VAD comparing 0 to 12 mg/ml CCh.
Discussion
Our previous studies of rat lungs in vivo have shown that the cholinergic-induced increase in total pulmonary elastance (which in this preparation is influenced by both the lung and the chest wall) is greater in VAD rats, and that RA-treatment restores the increase in elastance to a level, which is similar to that observed in VAS rats [1]. Elastance increases as tissue stiffness increases. In the lung, elastance is increased (a) when lung volumes approach total lung capacity, (b) by atelectasis, or (c) by an increase in rigid structural components (such as collagen) or (d) by an increase in hysteresis, which could result from alterations in alveolar surface tension or disruption of the elastic fibers [32,33]. In order to more specifically examine the contribution of the lung parenchyma to the exaggerated CCh-mediated increase in total pulmonary elastance that was observed VAD rats, we ventilated the lung ex vivo at a small tidal volume. This approach eliminated the contributions of the chest wall and of innervation and avoided the confounding effects of air-trapping that can be induced by large-volume oscillations. We found that the CCh-mediated increase in the elastic bulk modulus was exaggerated in VAD rats. This was manifest as an increase in the pressures required to expand the lung during inflation and a significant increase in hysteresis. In contrast, the CCh-mediated increase in shear modulus was diminished in VAD rats. Administration of RA for up to 21 days did not significantly reverse the effects of vitamin A deficiency on the bulk modulus, but there was a partial normalization of the shear modulus after 21 days of RA-treatment. The VAD-related alterations in the mechanical properties of the lung parenchyma were accompanied by a decrease in the concentration of parenchymal elastic fibers and in lung elastin. The administration of RA for 12 days increased TE mRNA but did not restore the 0.1 M NaOH-resistant lung elastin, although the concentration of parenchymal elastic fibers was increased. Therefore, decreases in lung parenchymal elastic fibers and total pulmonary elastin likely contribute to but do not completely account for to the exaggerated CCh-mediated increase in the elastance and bulk modulus in VAD rats.
Others have shown, using a qualitative pathologic grading system in rats, that VAD is associated with patchy atelectasis as well as emphysema [33]. Our previous morphometric study, using lungs that were inflated to 20 cm H2O confirmed the presence of emphysematous areas [1]. Terminal airway closure that occurs after the administration of aerosolized cholinergic agents results in a non-uniform distribution of atelectatic and hyperexpanded areas of parenchyma, which could exaggerate the pre-existing abnormalities that are associated with VAD [14]. The data in Table 1 are consistent with this statement, and show that both the SD Lm and SD ATI are increased in VAD relative to VAS lungs, following CCh administration. The exaggerated CCh-mediated increase that we observed in the bulk modulus reflects an increase in the elastance of the lung parenchyma of VAD rats. The deflation volume-pressure characteristics of the excised lung are also consistent with an increase in lung elastance in VAD. This differs from what one would expect in a uniformly emphysematous lung for which elastance would decrease. Furthermore, one might expect that the decrease in elastic fiber concentration (length per mm3 lung parenchyma) and lung elastin that we observed in VAD rats would be accompanied by a decrease in lung elastance. Therefore, another anatomical abnormality must contribute to the exaggerated increase in parenchymal lung elastance after CCh-administration. It is likely that this abnormality involves localized areas of atelectasis and hyperinflation, which are exaggerated by CCh-administration (see Figure 11 and Table 1). From Figure 9 it is clear that higher pressures are required to expand VAD lungs, compared to VAS lungs, after cholinergic administration. This is particularly obvious at low lung volumes that are similar to those which were used to ventilate the excised lungs during the measurement of the bulk and shear moduli. An increase in surface tension in atelectatic regions is probably the major contributor to this increase in elastance and therefore the bulk modulus. These increased inflationary pressures in cholinergic-exposed VAD lungs resulted in an increase in the hysteresis of VAD compared to cholinergic-exposed VAS lungs (Figure 9).
The VAD-induced suppression of the CCh-mediated increase in the shear modulus requires an alternate explanation (Figure 4). Although the shear modulus increased as expected after CCh-administration in VAD lungs, the increase was less than in VAS lungs. The shear modulus reflects the ability of the lung parenchyma to resist distortion. As the lung is progressively inflated, the "struts" which surround the airspaces become more distended and rigid [12]. This leads to a greater resistance to a distorting shear stress. CCh administration increases alveolar distortion resulting in hyperexpanded alveoli, which stiffens the lung and increases the shear modulus [13]. The cholinergic-induced hyperexpansion and stiffening of the struts appears to occur more uniformly (SD Lm is lower) in VAS lungs, which are not restricted by pre-existing distortion from atelectasis and airspace enlargement, than in VAD lungs. In the areas of VAD lung where alveolar hyperexpansion occurs, there is less elastic tissue to resist shape distortion, which would result in a lower shear modulus. The decrease in elastic tissue likely contributes to the smaller CCh-induced increase in the shear modulus of VAD. The blunted CCh-induced increase in shear modulus in VAD rats likely contributes to their airway hyperresponsiveness, because the shear modulus is thought to be the most important characteristic that mediates airway-parenchymal interdependent opposition to airway contraction [34].
We observed that VAD, which occurs after the major peak of pulmonary elastin synthesis has occurred, is accompanied by a decrease in lung parenchymal elastin. The loss of elastin in VAD was manifest as a decrease in the alveolar septal elastic fiber concentration (mm length per mm3 lung parenchyma) and in the quantity of elastin that was resistant to digestion in the presence of 0.1 M NaOH at 98° (Figures 5 and 6, respectively). We also made the novel observation that RA stimulates elastin synthesis and the deposition of elastic fibers, which are important determinants of the mechanical properties of the parenchyma. Northern analysis demonstrated that 12 days of RA-administration increased the steady-state level of tropoelastin mRNA in the lung parenchyma, which is consistent with the restoration of elastic fiber concentration after 12 days of RA-treatment (Figure 7). However, we did not observe an increase in alkali-resistant elastin after 12 days of RA-administration. This may result from one or more of several factors. First, orcein can stain "immature" elastic fibers that contain a larger proportion of microfibrils than thicker fully cross-linked "mature" elastic fibers [35]. Only the "mature" elastic fibers are resistant to the alkali treatment, which underestimates that quantity of newly formed, incompletely cross-linked elastin. Secondly, our morphometric analysis of elastic fiber concentration did not account for the thickness of the fibers, only the length per unit volume of lung. Therefore, thin newly formed elastic fibers would contain less elastin that could be detected by our biochemical analysis, but the fibers would be detected by our method for determining the concentration of elastic fibers, which is only dependent on the length of the fiber network and not on its thickness.
The airway contraction index after CCh-administration was lower in VAD rats relative to VAS controls. Administration of RA for 12 days was associated with a restoration of the contraction index (after CCh) of airways >0.55 mm diameter to a level that was similar to VAS rats. These data are consistent with our previous study, which demonstrated that 12 days of RA administration restored total (lung plus chest wall) pulmonary elastance and resistance [1]. The data for the contraction index should be considered in light of our observation that RA-administration does not reverse the exaggerated CCh-mediated alterations in bulk and shear moduli in VAD rats. This consideration suggests that the RA-mediated correction of the increase in total pulmonary elastance that we previously observed in VAD rats with intact chest walls was primarily due to factors within the airways or chest wall rather than the lung parenchyma [1]. If RA had corrected the lung parenchymal factors, then we would have expected to observe a correction in the VAD-related abnormalities in the parenchymal elastic modulus. These findings suggest that although VAD alters the elastic fiber system, alveolar architecture, and mechanical properties of the lung parenchyma; treatment with RA for 12 days corrects a VAD-related abnormality in the airway, rather than the parenchyma. When considered along with our prior observation that 12 days of RA treatment normalizes the expression of the muscarinic receptor-2, these findings suggest that the salutary effect of administering RA for 12 days is limited to the airways [1].
Conclusion
Multiple factors contribute to airway hyperreactivity in vitamin A deficiency including changes in the architecture of the alveoli and alveolar ducts. Aerosolization of cholinergic agents results in more distortion and heterogeneity of airspace size, and in particular atelectasis, that may contribute to the exaggerated CCh-mediated increase in bulk modulus in vitamin A deficient rats. There is also a decrease in elastin and the concentration of elastic fibers in vitamin A deficiency, which may reduce the ability of the parenchyma to resist deformation after CCh administration, resulting in a smaller CCh-mediated increase in the shear modulus than in vitamin A sufficient rats. Retinoic acid administration restores the parenchymal elastic fibers but does not restore the CCh-induced responses of the bulk and shear moduli to the pattern that was observed in VAS rats. Therefore, architectural changes that are not directly related to elastin also influence airway-parenchymal interdependence and enhance airway hyperreactivity.
Abbreviations
ATI, atelectasis index; CCh, carbamylcholine; E, elastance; HPLC, high performance liquid chromatography; k, bulk modulus; Lm, mean chord length; PBS, 15 ml of 137 mM NaCl, 8 mM Na2HPO4, 2.7 mM KCl, 1.5 mM KH2PO4, pH 7.4; PEEP, positive end-expiratory pressure; R, resistance; RA retinoic acid; TE, tropoelastin; VAD, vitamin A deficient or vitamin A deficiency; VAS, vitamin A sufficient; μ, shear modulus.
Competing interests
The three authors declare that neither has a completing interest that would influence the objectivity of these findings.
Grant Support
The authors thank the Veterans Affairs Research Service (Merit Review Grant) National Heart, Lung and Blood Institute (HL53430, HL62861) for supporting this research.
Authors' contributions
SEM planned the experiments, performed the physiological measurements, wrote the manuscript, and performed some of the morphometry. AJT performed the studies of elastic fiber concentration, lung elastin contents, airway contraction index and assisted with the preparation of the manuscript. AJH performed the Northern analyses and made substantive contributions to the writing of the manuscript.
Acknowledgements
The authors greatly appreciate the assistance of Dr. Chris Coretsopoulos, Director of the Microfabrication Laboratory, University of Iowa College of Engineering in using the prophilometer.
==== Refs
McGowan SE Holmes AJ Smith J Retinoic acid reverses the airway hyperresponsiveness but not the parenchymal defect that is associated with vitamin A deficiency Am J Physiol 2004 286 L437 L444
McGowan SE Smith J Holmes AJ Vitamin A deficiency promotes bronchial hyperreactivity in rats by altering muscarinic M(2) receptor function Am J Physiol 2002 282 L1031 L1039
Underwood BA Smitasiri S Micronutrient malnutrition: policies and programs for control and their implications Ann Rev Nut 1999 19 303 24 10.1146/annurev.nutr.19.1.303
Liu R Harvey CS McGowan SE Retinoic acid increases elastin in neonatal rat lung fibroblast cultures American Journal of Physiology (Lung Cellular and Molecular Physiology) 1993 265 L430 L437
McGowan SE Doro MM Jackson SK Endogenous retinoids increase perinatal elastin gene expression in rat lung fibroblasts and fetal explants Am J Physiol (Lung Cell Mol Physiol) 1997 273 L410 L416
Mariani TJ Sandefur S Pierce RA Elastin in Lung Development Experi Lung Res 1997 23 131 45
Godfrey M Barnes PJ, Drazen, Rennard, Thompson Extracellular Matrix Asthma and COPD Basic Mechanisms and Clinical Management 2002 Academic Press, Amsterdam 211 218
Shapiro SD Endicott SK Province MA Pierce JA Campbell EJ Marked longevity of human lung parenchymal elastic fibers deduced from prevalence of D-aspartate and nuclear weapons-related radiocarbon J Clin Invest 1991 87 1828 34 2022748
Eidelman DH Lei M Ghezzo RH Morphometry of methacholine-induced bronchoconstriction in the rat JAppl Physiol 1993 75 1702 10 8282623
Song S-H Three dimensional structures of pulmonary elastin; airway VS vascular elastin Yonsei Med 1994 35 475 83
Leick-Maldonado EA Lemos M Tiberio IF Differential distribution of elastic system fibers in control and bronchoconstricted intraparenchymatous airways in the guinea-pig lung J Submicrosc Cytol Patho 1997 29 427 34
Stamenovic D Micromechanical foundations of pulmonary elasticity Physiol Rev 1990 70 1117 34 2217556
Okazawa M D'Yachkova Y Pare PD Mechanical properties of lung parenchyma during bronchoconstriction J Appl Physiol 1999 86 496 502 9931182
Nagase T Moretto A Ludwig MS Airway and tissue behavior during induced constriction in rats: intravenous vs. aerosol administration J Appl Physiol 1994 76 830 838 8175596
Smith JE Colowick SP, Kaplan Methods in Enzymology 1990 Academic Press, San Diego 229 236 2087174
Bhat PV Labrecque J Boutin JM Lacroix A Yoshida A Cloning of a cDNA encoding rat aldehyde dehydrogenase with high activity for retinal oxidation Gene 1995 166 303 6 8543180 10.1016/0378-1119(96)81752-5
Nagase T Matsui H Sudo E Matsuse T Ludwig MS Fukuchi Y Effects of lung volume on airway resistance during induced constriction in papain-treated rabbits J Appl Physiol 1996 80 1872 79 8806889 10.1063/1.363000
Salerno FG Ludwig MS Elastic moduli of excised constricted rat lungs J Appl Physiol 1999 86 66 70 9887114
Stamenkovic I Yager D Elastic properties of air- and liquid-filled lung parenchyma J Appl Physiol 1988 65 2565 70 3215857
Elias H Hyde DM An elementary introduction to stereology (quantitative microscopy) Am J Anat 1980 159 411 46 10.1002/aja.1001590407
Soskel NT Wolt TB Sandberg LB Cunningham LW, Colowick SP, Kaplan NO Isolation and Characterization of Insoluble and Soluble Elastins Methods in Enzymology 1987 144 Orlando, Academic Press 196 214 3626871
Schroter RC Quantitative comparisons of mammalian lung pressure volume curves Respir Physio 1980 42 101 7 10.1016/0034-5687(80)90107-3
Fedullo AJ Jung-Legg Y Snider G Karlinsky JB Hysteresis ratio: A measure of the mechanical efficiency of fibrotic and emphysematous hamster lung tissue Am Rev Respi Dis 2004 122 47 52
Scherle W A simple method for volumetry of organs in quantitative stereology Mikroskopie 1970 26 57 60 5530651
Bruce MC Palowski R Tomashefki JF Changes in Lung Elastic Fiber Structure and Concentration Associated with Hyperoxic Exposure in the Developing Rat Lung Am Rev Respir Dis 1989 140 1067 74 2679260
Weibel ER Stereological Methods 1979 Academic, New York
McGowan SE Jackson SK Jenkins-Moore M Dai H-H Chambon P Snyder JM Mice bearing deletions of retinoic acid receptors demonstrate reduced lung elastin and alveolar numbers Am J Respir Cell Mol Bio 2000 23 162 67 10919981
Dunhill MS Quantitative methods in the study of pulmonary pathology Thorax 1962 17 320 333
James AL Hogg JC Dunn LA Pare PD The use of the internal perimeter to compare airway size and to calculate smooth muscle shortening Am Rev Respir Dis 1988 138 136 39 3202392
Lambert RK Pare PD Okazawa M Stiffness of peripheral airway folding membrane in rabbits J Appl Physiol 2001 90 2041 47 11356763
Macklem PT Eidelman D Reexamination of the Elastic Properties of Emphysematous Lungs Respiration 1990 57 187 92 2274717
Lai-Fook SJ Hyatt RE Effects of age on elastic moduli of human lungs J Appl Physiol 2000 89 163 68 10904048
Baybutt RC Hu L Molteni A Vitamin A deficiency injures lung and liver parenchyma and impairs function of rat type II pneumocytes J Nutr 2000 130 1159 65 10801913
Lai-Fook S Hyatt RE Rodarte J Effect of parenchymal shear modulus and lung volume on bronchial pressure-diameter behavior J Appl Physio 1978 44 859 68
Böck P Stockinger L Light and electron microscopic identification of elastic, elaunin and oxytalin fibers in human tracheal and bronchial mucosa Anat Embryo 1984 170 145 53 10.1007/BF00318999
|
16033655
|
PMC1199621
|
CC BY
|
2021-01-04 16:23:25
|
no
|
Respir Res. 2005 Jul 20; 6(1):77
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-77
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-811604279010.1186/1465-9921-6-81ResearchCD8+ T lymphocytes in lung tissue from patients with idiopathic pulmonary fibrosis Daniil Zoe [email protected] Panagiota [email protected] George [email protected] Maria [email protected] Androniki [email protected] Marilena [email protected] Joseph [email protected] Charis [email protected] Spyros A [email protected] Department of Critical Care and Pulmonary Services, National and Capodistrian University of Athens, "Evangelismos" Hospital, Athens, Greece2 Pathology Department, "Evangelismos" Hospital, Athens, Greece3 Hematology Department, "Evangelismos" Hospital, Athens, Greece4 Meakins-Cristie Laboratories, McGill University, Montreal, Quebec, Canada2005 24 7 2005 6 1 81 81 26 5 2005 24 7 2005 Copyright © 2005 Daniil et al; licensee BioMed Central Ltd.2005Daniil et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Several studies have implicated a role of inflammation in the pathogenesis of lung damage in idiopathic pulmonary fibrosis (IPF). Parenchymal lung damage leads to defects in mechanics and gas exchange and clinically manifests with exertional dyspnea. Investigations of inflammatory cells in IPF have shown that eosinophils, neutrophils and CD8+ TLs may be associated with worse prognosis. We wished to investigate by quantitative immunohistochemistry infiltrating macrophages, neutrophils and T lymphocytes (TLs) subpopulations (CD3+, CD4+ and CD8+) in lung tissue of patients with IPF and their correlation with lung function indices and grade of dyspnoea.
Methods
Surgical biopsies of 12 patients with IPF were immunohistochemically stained with mouse monoclonal antibodies (anti-CD68 for macrophages, anti-elastase for neutrophils, and anti-CD3, anti-CD4, anti-CD8 for CD3+TLs, CD4+TLs, and CD8+TLs respectively). The number of positively stained cells was determined by observer-interactive computerized image analysis (SAMBA microscopic image processor). Cell numbers were expressed in percentage of immunopositive nuclear surface in relation to the total nuclear surface of infiltrative cells within the tissue (labeling Index). Correlations were performed between cell numbers and physiological indices [FEV1, FVC, TLC, DLCO, PaO2, PaCO2 and P(A-a)O2)] as well as dyspnoea scores assessed by the Medical Research Council (MRC) scale.
Results
Elastase positive cells accounted for the 7.04% ± 1.1 of total cells, CD68+ cells for the 16.6% ± 2, CD3+ TLs for the 28.8% ± 7, CD4+ TLs for the 14.5 ± 4 and CD8+ TLs for the 13.8 ± 4. CD8+TLs correlated inversely with FVC % predicted (rs = -0.67, p = 0.01), TLC % predicted (rs = -0.68, p = 0.01), DLCO % predicted (rs = -0.61, p = 0.04), and PaO2 (rs = -0.60, p = 0.04). Positive correlations were found between CD8+TLs and P(A-a)O2 (rs = 0.65, p = 0.02) and CD8+TLs and MRC score (rs = 0.63, p = 0.02). Additionally, CD68+ cells presented negative correlations with both FVC % predicted (rs = -0.80, p = 0.002) and FEV1 % predicted (rs = -0.68, p = 0.01).
Conclusion
In UIP/IPF tissue infiltrating mononuclear cells and especially CD8+ TLs are associated with the grade of dyspnoea and functional parameters of disease severity implicating that they might play a role in its pathogenesis.
==== Body
Background
In usual interstitial pneumonia (UIP)/idiopathic pulmonary fibrosis (IPF) the role of inflammation in the pathogenesis of fibrosis is debatable [1-3]. Traditionally, UIP/IPF was regarded to develop in response to chronic inflammation of the lung parenchyma [4]. This view was advanced from previous studies implicating a role of the inflammatory cells including neutrophils, macrophages, eosinophils and T lymphocytes (TLs), based on the observation of their accumulation in sites of disease activity [5-8] or on their presence in high numbers in bronchoalveolar lavage [9-12].
Actually, the current pathogenetic theory that holds in UIP/IPF, implicates that fibrosis per se might progress despite a paucity of interstitial inflammation [13]. However, even in this case, recent data still indicate the contention that the type of the inflammatory response may modulate tissue injury, fibrosis or both [3,4]. Animal studies imply that TLs might play a role in the initiation and the evolution of pulmonary fibrosis. They also suggest that different TLs subpopulations, including both CD4+ and CD8+ subsets, might contribute through their ability to secrete fibrogenic cytokines [14,15]. Ultimately, in UIP/IPF, the inflammatory response is considered to resemble closely the type-2 T lymphocytic pattern [16-18] and drives the process in a profibrogenic direction.
The present study was designed to investigate by quantitative immunohistochemistry the inflammatory cell pattern in lung tissue of patients with UIP/IPF (macrophages, neutrophils, and CD3+, CD4+, CD8+ TLs) and to correlate their population numbers with the lung function indices and grade of dyspnoea.
Methods
1. Subjects
The study population consisted of 12 untreated patients with IPF and included 7 ex-smokers, and 5 never smokers (Table 1). They were recruited from the respiratory outpatient clinic of the "Evangelismos" General Hospital, Athens, Greece over a period of 3 years. The diagnosis of UIP/IPF was based on standard criteria [19], which included clinical findings (exertional dyspnoea, non-productive cough, fine bibasilar inspiratory crackles), pulmonary function tests (restrictive pattern and impaired gas exchange), and high resolution computerized tomography findings (bibasilar reticular abnormalities with minimal ground-glass opacities consistent with the diagnosis of IPF). The diagnosis of UIP/IPF was confirmed by video-assisted thoracoscopic lung biopsy in all patients. Pathology examination of these specimens clearly documented UIP according to Katzenstein's and American Thoracic Society – European Respiratory Society criteria (histologic variation with alternating zones of interstitial fibrosis, inflammation, honeycomb change, and normal lung [1,19]. A right thoracic approach was done and two or three samples were taken from the right lower or middle lobe in the region of the greater fissure. Biopsy of the lingular tip was avoided, as changes in this area may be particularly advanced and unrepresentative. All patients experienced a normal and uncomplicated postoperative course. Secondary causes of lung fibrosis were excluded: none of the patients included in this study had a history of environmental or occupational exposure, drug toxicity or connective tissue disease, as documented by patient's history and thorough clinical and immunological work out. The study was approved by the institutional ethics committee and informed consent was obtained from each patient.
Table 1 Demographic, clinical and lung function data of all patients at presentation
Variables Values
Age (yr) 60 ± 2
Sex (M/F) 5/7
MRC dyspnoea score 1.8 ± 0.3
FEV1 (% pr) 85 ± 4
FVC (% pr) 78 ± 4
FEV1/FVC (ratio × 100) 85 ± 4
TLC (% pr) 64 ± 3
RV (% pr) 54 ± 4
DLCO (% pr) 50 ± 5
PaO2 (mmHg) 75 ± 2
P(A-a)O2 30 ± 3
PaCO2 (mmHg) 36 ± 3
Data are presented as means ± SEM; M = Male; F = Female; pr: predicted
2. Pulmonary function tests
The pulmonary function tests included FEV1, FVC, FEV1/FVC ratio × 100, total lung capacity (TLC), residual volume (RV) and carbon monoxide transfer factor (DLCO). TLC and RV were measured by the helium dilution method with a Master Screen apparatus (Erich Jaeger GmbH, Wuerzburg, Germany), and DLCO by the single breathholding helium dilution method [20,21]. Lung function measurements (Table 1) were expressed as percentages of predicted values [20,21]. In all patients, the arterial PaO2 and PaCO2 were also measured at rest, and P(A-a)O2 calculated.
3. Dyspnea
Dyspnea was assessed with the modified (6-point) MRC dyspnoea self-administered questionnaire [22] that consists of six questions about perceived breathlessness: category 0, no dyspnoea; category 1, slight degree of dyspnoea (troubled by shortness of breath when hurrying on the level or walking up a slight hill); category 2, moderate degree of dyspnoea (walks slower than people of the same age on the level because of breathlessness); category 3, moderately severe degree of dyspnoea (has to stop because of breathlessness when walking at own pace on the level); category 4, severe degree of dyspnoea (stops for breath after walking about 100 yards or after a few minutes on the level); category 5, very severe degree of dyspnoea (too breathless to leave the house or breathless when dressing or undressing).
4. Histology
Open lung biopsies from the 12 patients were used. They were taken for diagnostic and staging purposes and were analyzed according to the Katzenstein's criteria [1] by two pathologists. Specimens were fixed in 4% formalin and after dehydration embedded in paraffin. Tissue sections were orientated and serial sections of 4 μm thickness were cut and immunohistochemistry was performed according to the Streptavidin-Biotin method.
5. Immunohistochemistry
To evaluate macrophages, neutrophils and the lymphocyte-subpopulations in lung tissues, 4 μm paraffin sections were immunohistochemically stained with the following mouse monoclonal antibodies (all from DAKO, Glostrup, Denmark): macrophages-histiocytes with anti CD68, (clone: M814, dilution 1:4000), neutrophils with anti-elastase (clone:M752 dilution 1:4000), T-cells with anti-CD3 (dilution 1:200), anti-CD4 (dilution 1:100), and anti-CD8 (dilution 1:4), according to the labeled Streptavidin-Biotin Complex method. The sections were deparaffinized and rehydrated with Tris-Buffered Saline (TBS: 0.005 M Tris, 0.15 M NaCl), pH = 7.6 for 10 minutes. Endogenous peroxidase was blocked with 3% hydrogen peroxide for 5 minutes. Then they were washed in TBS and incubated with primary antibodies at the appropriate dilutions for one hour. Biotinylated antimouse IgG was used as a secondary antibody (DAKO), followed by peroxidase-conjugated streptavidin (DAKO). The peroxidase reaction was developed using 3,3'-diaminbenzidine tetrachloride (0.25 mg dissolved in 1 ml of 0.02% hydrogen peroxide) for 3 min.
6. Lung Parenchyma Computer Image Analysis
The number of positively stained cells was determined by observer-interactive computerized image analysis (SAMBA microscopic image processor), whose hardware and software have been described by Brugal and associates [23]. This system is fitted with a standard Zeiss axioplan microscope, a color video camera (Sony Corporation Tokyo, Japan), an image analysis processor (matrox) and an IBM compatible Pentium 2, 166 MHZ computer. Estimation of the standard error of the mean within 95% confidence limits required a maximum of at least randomly selected 15 High Power fields (X400-Zeiss microscope) (Analysis per area of approximately 110000 μm2). The immunostaining was analyzed as dark brown color with counterstained cells as false blue. Formal scoring (labeling Index) for each antibody was then performed in one section for each paraffin block. Interobserver variability was very low (<0.03%). The results were expressed in percentage of immunopositive nuclear surface in relation to the total nuclear surface of infiltrative cells within the tissue (labeling Index), as previously described [24,25]. Blood vessels, connective tissue and cartilage structures were excluded.
7. Statistical Analysis
Data were expressed as means and standard error (SEM). Correlation coefficients were calculated using Spearman's rank method. A p-value of less than 0.05 was considered statistically significant. Analysis was performed using the SAS System software.
Results
Demographic characteristics, MRC dyspnoea score and lung function data of all patients are listed in Table 1. All patients claimed some degree of dyspnoea (MRC score > 0) and most patients had a restrictive lung function pattern characterized by a decrease in TLC (mean value was 64% of predicted) and an increased in FEV1/FVC ratio ×100 (mean value was 85% of predicted). The DLCO was decreased in all patients (mean value was 50% of predicted).
Among the inflammatory cells studied T lymphocytes (CD3+) appeared as the most numerous cells infiltrating the lung parenchyma. They were found in aggregates, within lymphoid follicles or diffusely within the fibrotic lung parenchyma and mainly in the areas with moderate or severe thickening of the alveolar wall. They were also observed within the wall of the alveoli. The CD4+ subpopulation was found in aggregates inside or around lymphoid follicles. A little portion of them infiltrated the lung parenchyma diffusely, especially the alveolar wall. On the other hand, the CD8+ cells infiltrated the parenchyma mainly diffusely (Fig 1); they were also found within the alveolar wall, around the fibrotic foci and in the areas with alveolar thickening. Less commonly they were distributed within aggregates or lymphoid follicles. Macrophages (CD68+ cells) were preferentially located in the lamina propria of the airways compared with the surface epithelium and the submucosa. They were also distributed in large aggregates in the dilated alveolar spaces. Neutrophils (Elastase+) were mainly observed in the surface epithelium and within the alveolar wall.
Figure 1 The CD8+TLs infiltrate diffusely the lung parenchyma (scale bar = 25 μm).
Elastase positive cells accounted for the 7.04% ± 1.1 of total cells, CD68+ cells for the 16.6% ± 2, CD3+ TLs for the 28.8% ± 7, CD4+ TLs for the 14.5 ± 4 and CD8+ TLs for the 13.8 ± 4. Among the infiltrating inflammatory cells, CD8+TLs were inversely correlated with FVC % predicted (rs = -0.67, p = 0.01) (Fig 2), TLC % predicted (rs = -0.68, p = 0.01) (Fig 3), DLCO % predicted (rs = -0.61, p = 0.04), PaO2 (rs = -0.60, p = 0.04). Positive and statistically significant correlations were found between CD8+TLs and P(A-a)O2 (rs = 0.65, p = 0.02) and CD8+TLs and MRC score (rs = 0.63, p = 0.02) (Fig 4). Additionally, the CD68+ cells presented significant negative correlations with the FVC % predicted (rs = -0.80, p = 0.002) and the FEV1 % predicted (rs = -0.68, p = 0.01). Elastase positive cells, CD3+ TLs, and the CD4+ TLs did not correlate with any of the functional indices as well as the MRC score.
Figure 2 Relationship between CD8+TLs (% of total cells) and FVC (% predicted), (rs = -0.67, p = 0.01).
Figure 3 Relationship between CD8+TLs (% of total cells) and TLC (% predicted), (rs = -0.68, p = 0.01).
Figure 4 Relationship between CD8+TLs (% of total cells) and MRC dyspnea score, (rs = 0.63, p = 0.02).
Discussion
Several studies have implicated a role of inflammation in the pathogenesis of lung damage in IPF. Parenchymal lung damage leads to defects in mechanics and gas exchange and consequently to exertional dyspnoea, the most prominent and disabling symptom in these patients. This study shows that the type of the inflammatory infiltrate in lung tissue of patients with UIP/IPF was predominantly mononuclear, and that among the different inflammatory cells, CD8+ TLs correlated significantly with both functional [FVC, TLC, DLCO, PaO2, P(A-a)O2] and clinical indices (the MRC chronic dyspnoea score) of the disease severity and extent studied and macrophages (CD68+ cells) with some of the functional indices studied (FVC and FEV1).
Previous studies have shown that a patchy chronic interstitial inflammatory process coexists with an abnormal extracellular matrix deposition, foci of fibroblasts, and alveolar collapse in lung tissue of subjects with IPF [1,2,8,26]. This inflammatory component in tissue specimens appeared to consist primarily of mononuclear cells (macrophages, lymphocytes and plasma cells) while the presence of the other inflammatory cells such as neutrophils and eosinophils appeared less numerous [5,8,26,27]. Previous studies of immunohistochemical analysis including the TLs subpopulations of lung tissue in IPF [5,8,28,29] and other interstitial pneumonias [30] have shown that the inflammatory process is mainly mononuclear, and that both CD4+ and CD8+ TLs were well represented and diffusely distributed in the interstitium, with an additional component of the CD4+ TLs observed inside lymphoid follicles. Our findings do not contrast these observations.
Few studies have attempted to evaluate the structure-function relationship in IPF, and their findings are not always in agreement [31-35]. The selection of patients and the different methodologies might, at least in part, explain discrepancies. Our findings come into agreement with the studies by Fulmer and coworkers [31] and Chinet and coworkers [35] who found significant correlations between the degree of inflammation and lung volumes and some index of gas exchange. Gaensler and Carrington [34] also reported correlations between physiological indices and an estimate of functional impairment from histology (designated by them as pathological severity) in a large but mixed population (502 patients) with "interstitial lung disorders" including 64 patients with UIP. However, comparisons with previous studies are not always possible for the following reasons. First, we used a different methodology than the above-cited works. This is the first study attempted to correlate cell counts including TLs subpopulations in tissue biopsies with clinical and physiological parameters. This should be in relation to the fact that reliable cell counts were not feasible with the past technologies. Second, older studies might have included a mixed population of patients with UIP/IPF, patients with non-specific interstitial pneumonia and patients with pulmonary fibrosis associated with collagen vascular disorders since there were not defined strict criteria for these entities, by that time [36]. Finally, the patients' selection and the effect of previous treatment might have influenced the results.
Inflammatory cells including subpopulations of TLs have been also studied in IPF by bronchoalveolar lavage (BAL) [4] and many of the conclusions regarding the role of inflammation in interstitial lung disorders have been drawn from these studies. The possible role of lymphocytes in the pathogenesis of IPF traditionally received little investigation since increase in their number is an uncommon finding in BAL samples. Hence, early BAL studies have driven attention into neutrophils as well as macrophages. However, evidence from animal models appears to suggest that lymphocytes do play a role in fibrosis [4]. Furthermore, studies on BAL lymphocytes have shown that CD8+ TLs are prominent in BAL in IPF [36] and may also be associated with a worse prognosis [37]. T lymphocytes and their phenotypic and functional characteristics have been more extensively studied in scleroderma fibrosis [38-43]. IPF and scleroderma fibrosis are two fibroses with different prognoses [41]. This might be related to the fact that most patients with scleroderma develop a less aggressive form of fibrosing interstitial pneumonia, the non-specific interstitial pneumonia (NSIP) [42]. Indeed, studies that compared the prognosis of patients with "idiopathic" NSIP to that of patients with usual interstitial pneumonia type/IPF have clearly shown that the former present a far better prognosis than the latter [43]. Recent studies with BAL in scleroderma patients have shown that a subset of them, who present more than 15% lymphocytes in BAL [38], or have activated, long-lived CD8+ T cells [40], or produce type 2 cytokines (IL-4 and IL-5) by the CD8+ TLs [39], present a more aggressive form of interstitial pneumonia. Hence, TLs and in particular the CD8+ subset may be associated with progressive fibrosis in scleroderma resembling more patients with IPF.
Notwithstanding correlations do not imply direct cause-effect relationships, we think that the significant negative correlations observed in this group of patients with IPF between CD8+TLs and functional indices and the positive correlation observed between the same cells and clinical indices estimating the disease severity as well as the correlation between CD68+ cells with FVC and FEV1 might suggest a potential pathophysiologic relevance for mononuclear cells and especially CD8+TLs in the pathogenesis of pulmonary fibrosis. However, the mechanisms related to these correlations and the relationship of inflammatory and immune parameters to structural changes in the lung parenchyma still remain unknown and further studies are needed for their clarification.
The increase in CD8+ TLs observed in lung surgical biopsies in patients with IPF appears intriguing. Classically, the major role of CD8+ TLs in the inflammatory response has been considered the rapid resolution of viral infections [44]. It has also become evident that CD8+ TLs may contribute to lung injury [45,46]. Viruses have been implicated in the pathogenesis of IPF, and a higher incidence of viral infections (Ebstein Barr Virus, influenza, cytomegalovirus, and possibly Hepatitis C virus) has been reported in these patients [19]. Recently, it has been hypothesized that in patients with IPF an excessive recruitment of CD8+ TLs may occur in response to recurrent or persistent viral infections, and this excessive response may play a role for the development of lung damage [47]. The above hypothesis has received some experimental confirmation by the studies of Enelow and coworkers [47] and Small and coworkers [48] who have shown that antigen-specific CD8+ T cell recognition of an alveolar epithelial "autoantigen" is sufficient to trigger an inflammatory cascade that results in the histological and physiological manifestations of interstitial pneumonia. CD8+ TLs can differentiate into cells that make IFN-γ but no IL-4 (Tc1 cells) promoting attenuation of fibrosis and cells that make IL-4 but not IFN-γ (Tc2 cells) leading to exuberant fibrosis [49]. Though further studies are necessary to address the specific role of the Tc2 cells in pulmonary fibrosis, some data such as the upregulation of genes encoding immunoglobulins and extracellular matrix proteins in IPF lung tissue [50] appear to suggest that the predominance of type-2 immune response in IPF [51] is what drives the process in profibrogenic direction.
Conclusion
We found that the type of the inflammatory cell infiltrate in surgical biopsies of patients with UIP/IPF was predominantly mononuclear. Among the different inflammatory cells revealed by immunohistochemistry, the CD8+TLs correlated significantly with both functional [FVC, TLC, DLCO, PaO2, P(A-a)O2] and clinical indices (the MRC chronic dyspnoea score) of disease severity and extent studied and the CD68+ cells with FVC and FEV1. These data might suggest a potential role for mononuclear cells and especially CD8+TLs in the pathogenesis of pulmonary fibrosis. However, because of the relatively small size of the population studied, further studies are needed to support our findings.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
ZD participated in the design of the study and collection of the clinical data, performed the statistical analysis and drafted the manuscript. PK carried out the histology and the immunohistochemical analysis and revised the article. GK participated in the collection of the data and helped to draft the manuscript. MD helped in biopsy evaluation, diagnosis and analysis. AK participated in tissue collection and data analysis. MK participated in data analysis. JM-E helped in the interpretatiuon of the data and revised the article. CR participated in the interpretation of the data and revised the article. SP conceived of the study, participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript.
Acknowledgements
Supported by the "Thorax" Foundation, Athens, Greece.
==== Refs
Katzenstein A Zisman AD Litzky AL Nguyen TB Kotloff MR Usual interstitial pneumonia. Histologic study of biopsy and explant specimens Am J Surg Pathol 2002 26 1567 1577 12459623 10.1097/00000478-200212000-00004
Flaherty KR Travis WD Colby TV Toews BG Kazerooni AE Gross HB Jain A Strawderman LR IIIFlint A Lynch PJ IIIMartinez JF Histopathologic variability in usual and nonspecific interstitial pneumonias Am J Respir Crit Care Med 2001 164 1722 1727 11719316
Gross TJ Hunninghake GW Idiopathic pulmonary fibrosis N Engl J Med 2001 345 517 525 11519507 10.1056/NEJMra003200
Riches WHD Worthen GS Augustin A Lapadat R Chan DE Schwartz JM, King ET Jr Inflammation in the pathogenesis of interstitial lung disease Interstitial Lung Disease 2003 Hamilton London BC Decker 187 220
Kradin RL Divertie MB Colvin RB Ramirez J Ryu J Carpenter HA Bhan AK Usual interstitial pneumonitis is a T-cell alveolitis Clin Immunol Immunopathol 1986 40 224 235 3521973 10.1016/0090-1229(86)90025-5
Crystal RG Fulmer JD Roberts WC Moss ML Line BR Reynolds HY Idiopathic pulmonary fibrosis. Clinical, histologic, radiographic, physiologic, scintigraphic, cytologic and biochemical aspects Ann Intern Med 1976 85 769 788 793466
Hunninghake GW Kawanami O Ferrans VJ Young RC Roberts WC Crystal RG Characterization of the inflammatory and immune effector cells in the lung parenchyma of patients interstitial lung disease Am Rev Respir Dis 1981 123 407 412 7224352
Campell DA Poulter LW Janossy G du Bois RM Immunohistological analysis of lung tissue from patients with cryptogenic fibrosing alveolitis suggesting local expression of immune hypersensitivity Thorax 1985 40 405 411 3875162
Haslam PL Turton CW Lukoszek A Salsbury AJ Dewar A Collins JV Turner-Warwick M Bronchoalveolar lavage fluid cell counts in cryptogenic fibrosing alveolitis and their relation to therapy Thorax 1980 35 328 339 7434282
Reynolds HY Fulmer JD Kazmierowski JA Roberst WC Frank MM Crystal RG Analysis of cellular and protein content of broncho-alveolar lavage fluid from patients with idiopathic pulmonary fibrosis and chronic hypersensitivity pneumonitis J Clin Invest 1977 59 165 175 830661
Peterson MW Monick M Hunninghake GW Prognostic role of eosinophils in pulmonary fibrosis Chest 1987 92 51 56 3595249
Lynch JP 3rdStandiford TJ Rolfe MW Kunkel SL Strieter RM Neutrophilic alveolitis in idiopathic pulmonary fibrosis. The role of interleukin-8 Am Rev Respir Dis 1992 145 1433 1439 1596015
Selman M King TE JrPardo A Idiopathic pulmonary fibrosis: prevailing and evolving hypotheses about its pathogenesis and implications therapy Ann Intern Med 2001 134 136 151 11177318
Piguet PF Collart MA Grau GE Kapanci Y Vassali P Tumor necrosis factor/cachectin plays a key role in bleomycin induced pneumopathy and fibrosis J Exp Med 1989 170 655 663 2475571 10.1084/jem.170.3.655
Hu H Stein-Streilein J Hapten-immune pulmonary interstitial fibrosis (HIPIF) in mices requires both CD4+ and CD8+ T-lymphocytes J Leukocyte Biol 1993 54 414 422 7901313
Furuie H Yamasaki H Suga M Ando M Altered accessory cell function of alveolar macrophages: a possible mechanism for induction of Th 2 secretory profile in idiopathic pulmonary fibrosis Eur Respir J 1997 10 787 794 9150314
Hancock A Armstrong L Gama R Millar A Production of interleukin 13 by alveolar macrophages from normal and fibrotic lung Am J Respir Cell Mol Biol 1998 18 60 65 9448046
Romagnani S Th1/Th2 cells Inflamm Bowel Dis 1999 5 285 294 10579123
American Thoracic Society Idiopathic pulmonary fibrosis: diagnosis and treatment. International consensus statement Am J Respir Crit Care Med 2000 161 646 664 10673212
Quanjer PhH Tammeling GJ Cotes JE Pedersen OF Peslin R Yernault J-C Lung volumes and forced ventilatory flows. Report working party, Standardization of lung function tests, European Community for steel and coal. Official Statement of the European respiratory Society Eur Respir J 1993 Suppl 16 5 40 8381090
Cotes JE Chinn DJ Quanjer PhH Roca J Yernault J-C Standardization of the measurement of transfer factor (Diffusing Capacity). Report working party, Standardization of lung function tests, European Community for steel and coal. Official Statement of the European respiratory Society Eur Respir J 1993 Suppl 16 41 52
Eltayara L Becklake MR Volta CA Milic-Emili J Relationship between chronic dyspnea and expiratory flow limitation in patients with chronic obstructive pulmonary disease Am J Respir Crit Care Med 1996 154 1726 1734 8970362
Brugal T Arnorssen R Bengtsson A Wilander E A double scanning micromorphometer for image analysis: hardware, software and biomedical applications J Histochem Cytochem 1979 27 144 152 374569
Charpin C Martin P Jacquemin J Jacquemier J Lavaut MN Pourreau-Schneider N Toga M Estrogen receptor immunocytochemical assay (ER-ICA): Computerized image analysis system, immunoelectron microscopy and comparisons with estradiol binding assays in 115 breast assays in 115 breast carcinomas Cancer Res 1986 Suppl 8 4271 4277
Costes V Marty-Ane C Picot MC Serre I Pujol J-L Mary H Baldet P Typical and Atypical Bronchopulmonary Carcinoid Tumors: A Clinicopathologic and Ki-67-Labeling Study Hum Pathol 1995 26 740 745 7628845 10.1016/0046-8177(95)90221-X
White SR Lazar HM Thannichal JV Pathogenetic mechanisms in usual interstitial pneumonia/idiopathic pulmonary fibrosis J Pathol 2003 201 343 354 14595745 10.1002/path.1446
Haslam LP Turton WGC Heard B Lukoszek A Collins VJ Slasbury A Turner-Warwick Bronchoalveolar lavage in pulmonary fibrosis: comparison of cells obtained with lung biopsy and clinical features Thorax 1980 35 9 18 7361291
Haslam LP Evaluation of alveolitis by studies of lung biopsies Lung 1990 Suppl 168 984 992 2117219
Paradis LI Dauber HJ Rabin SB Lymphocyte phenotypes in bronchoalveolar lavage and lung biopsies in sarcoidosis and idiopathic pulmonary fibrosis Am Rev Respir Dis 1986 133 855 860 3486613
Yamadori I Fujiata J Kajitani H Tokuda M Yang Y Ohtsuki Y Yoshinouchi T Kamei T Ishida T Lymphocytic subsets in lung tisuue of non-specific interstitial pneumonia and pulmonary fibrosis associated with collagen vascular disorders: correlation with CD4/CD8 ratio in bronchoalveolar lavage Lung 2000 178 361 370 11361059 10.1007/s004080000037
Fulmer JD Roberts WC Von Gal R Crystal RG Morphologic-physiologic correlates of the severity of fibrosis and degree of cellular infiltration in idiopathic pulmonary fibrosis J Clin Invest 1979 63 665 676 438328
Green GM Graham GB Hanson JS Gump DW Phillips CA Brody AR Sylvester DW Landis JN Davis GS Craighead JE Correlate studies of interstitial pulmonary disease Chest 1976 suppl 2 263 1248299
Gaensler EA Carrington CB Coutu RE FitzGerald MX Radiologic-physiologic-pathologic correlations in interstitial pneumonias Prog Respir Res 1975 8 223 241
Gaensler EA Carrington CB Open lung biopsy for chronic diffuse infiltrative lung disease: clinical, roentgenographic and physiologic correlations in 502 patients Ann Thorac Surg 1980 30 411 426 7436611
Chinet T Jaubert F Dusser D Danel C Chretien J Huchon JG Effects of inflammation and fibrosis on pulmonary function in diffuse lung fibrosis Thorax 1990 45 675 768 2218974
Groen H Hamstra M Aalbert R Van Der Mark WT Koëter HG Postma SD Clinical evaluation of lymphocyte sub-populations and oxygen radical production in sarcoidosis and idiopathic pulmonary fibrosis Respir Med 1994 88 55 64 8029515 10.1016/0954-6111(94)90175-9
Fireman E Vardinon N Burke M Soizer S Levin S Endler A Stav D Topilsky M Mann A Schwarz S Greif J Predictive value of response to treatment of T-lymhocyte subpopulations in idiopathic pulmonary fibrosis Eur Respir J 1998 11 706 711 9596125
Behr J Vogelmeier C Beinert T Meurer M Krombach F Konig G Fruhmann G Bronchoalveolar lavage for evaluation and management of scleroderma disease of the lung Am J Respir Crit Care Med 1996 154 400 6 8756813
Atamas SP Yurovsky VV Wise R Wigley FM Goter Robinson CJ Henry P Alms WJ White B Production of type 2 cytokines by CD8+ lung cells is associated with greater decline in pulmonary function in patients with systemic sclerosis Arthritis Rheum 1999 42 1168 78 10366109 10.1002/1529-0131(199906)42:6<1168::AID-ANR13>3.0.CO;2-L
Luzina IG Atamas SP Wise R Wigley FM Choi J Xiao HQ White B Occurrence of an activated, profibrotic pattern of gene expression in lung CD8+ T cells from scleroderma patients Arthritis Rheum 2003 48 2262 74 12905481 10.1002/art.11080
Papiris SA Vlachoyiannopoulos PG Maniati MA Karakostas KX Constantopoulos SH Moutsopoulos HH Idiopathic pulmonary fibrosis and pulmonary fibrosis in diffuse systemic sclerosis: two fibroses with different prognoses Respiration 1997 64 81 5 9044481
Flaherty KR Colby TV Travis WD Toews GB Mumford J Murray S Thannickal VJ Kazerooni EA Gross BH Lynch JP 3rdMartinez FJ Fibroblastic foci in usual interstitial pneumonia: idiopathic versus collagen vascular disease Am J Respir Crit Care Med 2003 167 1410 5 12615630 10.1164/rccm.200204-373OC
Daniil ZD Gilchrist FC Nicholson AG Hansell DM Harris J Colby TV du Bois RM A histologic pattern of nonspecific interstitial pneumonia is associated with a better prognosis than usual interstitial pneumonia in patients with cryptogenic fibrosing alveolitis Am J Respir Crit Care Med 1999 160 899 905 10471616
Ramsay A Ruby J Ramshaw I A case for cytokine as effector molecules in the resolution of virus infection Immunol Today 1993 14 155 157 8499074 10.1016/0167-5699(93)90277-R
Alwan WH Kozlowska WJ Opensaw PJ Distinct types of lung disease caused by functional subsets of antiviral T cells J Exp Med 1994 179 81 89 8270885 10.1084/jem.179.1.81
Guidotti LG Chisari FV To kill or to cure: options in host defense against viral infection Curr Opin Immunol 1996 8 478 483 8794011 10.1016/S0952-7915(96)80034-3
Enelow RI Mohammed AZ Stoler MH Ning Liu A Young JS Lou Y-H Structural and functional consequences of alveolar cell recognition by CD8+ T Lymphocytes in experimental lung disease J Clin Invest 1998 102 1653 1661 9802879
Small BA Dressel SA Lawrence CW Drake DR IIIStoler MH Enelow RI Braciale TJ CD8+ T Cell-mediated Injury In Vivo Progresses in the Absence of Effector T Cells J Exp Med 2001 194 1835 1846 11748284 10.1084/jem.194.12.1835
Croft M Carter L Swain SL Dutton RWL Generation of polarized antigen-specific CD8 effector populations: reciprocal action of interleukin (IL)-4 and IL-12 in promoting type 2 versus type 1 cytokine profiles J Exp Med 1994 180 1715 1728 7525836 10.1084/jem.180.5.1715
Zuo F Kaminski N Eugui E Allard J Yakhini Z Ben-Dor A Lollini L Morris D Kim Y DeLustro B Sheppard D Pardo A Selman M Heller RA Gene expression analysis reveals matrilysin as a key regulator of pulmonary fibrosis in mice and humans Proc Natl Acad Sci USA 2002 99 6292 6297 11983918 10.1073/pnas.092134099
Strieter RM Keane MP Innate immunity dictates cytokine polarization relevant to the development of pulmonary fibrosis J Clin Invest 2004 114 165 168 15254582 10.1172/JCI200422398
|
16042790
|
PMC1199622
|
CC BY
|
2021-01-04 16:23:25
|
no
|
Respir Res. 2005 Jul 24; 6(1):81
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-81
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-821604279710.1186/1465-9921-6-82ResearchThe effect of prior statin use on 30-day mortality for patients hospitalized with community-acquired pneumonia Mortensen Eric M [email protected] Marcos I [email protected] Antonio [email protected] Jacqueline [email protected] VERDICT Research Center, Audie L Murphy VA Hospital, San Antonio, Texas, USA2 Division of General Medicine, The University of Texas Health Science Center at San Antonio, USA3 Division of Pulmonary and Critical Care Medicine, The University of Texas Health Science Center at San Antonio, USA2005 25 7 2005 6 1 82 82 26 4 2005 25 7 2005 Copyright © 2005 Mortensen et al; licensee BioMed Central Ltd.2005Mortensen et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Recent studies suggest that HMG-CoA reductase inhibitors ("statins") may have beneficial effects for patients at risk for some types of infections. We examined the effect of prior outpatient use of statins on mortality for patients hospitalized with community-acquired pneumonia.
Methods
A retrospective cohort study conducted at two tertiary teaching hospitals. Eligible subjects were admitted with a diagnosis of, had a chest x-ray consistent with, and had a discharge ICD-9 diagnosis of pneumonia. Subjects were excluded if they were "comfort measures only" or transferred from another acute care hospital. Subjects were considered to be on a medication if they were taking it at the time of presentation.
Results
Data was abstracted on 787 subjects at the two hospitals. Mortality was 9.2% at 30-days and 13.6% at 90-days. At presentation 52% of subjects were low risk, 34% were moderate risk, and 14% were high risk based on the pneumonia severity index. In the multivariable regression analysis, after adjusting for potential confounders including a propensity score, the use of statins at presentation (odds ratio 0.36, 95% confidence interval 0.14–0.92) was associated with decreased 30-day mortality.
Discussion
Prior outpatient statin use was associated with decreased mortality in patients hospitalized with community-acquired pneumonia despite their use being associated with comorbid illnesses likely to contribute to increased mortality. Confirmatory studies are needed, as well as research to determine the mechanism(s) of this protective effect.
==== Body
Background
Community-acquired pneumonia is the seventh leading cause of death and the leading cause of infectious death in the United States [1]. Although mortality due to community-acquired pneumonia decreased significantly with the introduction of antibiotics in the 1950s, since that time mortality has been stable or increasing [2]. Despite this, only a few new classes of antibiotics have been added to the armamentarium for treating community-acquired pneumonia in the last 20 years and no new classes of medications beyond antibiotics have been added since the 1950s.
Recent studies have demonstrated that inhibitors of HMG-CoA reductase ("statins") have significant immunomodulatory effects and reduce systemic cytokine levels [3-8]. These cytokines play an important role in host defense mechanisms for patients with community-acquired pneumonia but under certain conditions may lead to septic shock or acute respiratory distress syndrome (ARDS) [9-11]. Recent studies have demonstrated that in patients hospitalized with bacteremia or diabetic lower extremity infections those patients who were taking statins had a significantly decreased odds of death after adjusting for other potential confounders [12,13].
The study aim was to assess the effects of prior outpatient statin use on 30-day mortality for patients hospitalized with community-acquired pneumonia after adjusting for other potential confounders including a propensity score based upon the use/non-use of statins at presentation.
Methods
This a retrospective cohort study of patients hospitalized with community-acquired pneumonia at 2 academic tertiary care hospitals in San Antonio, Texas. Both hospitals are teaching affiliates of the University of Texas Health Science Center at San Antonio. The Institutional Review Board of the University Health Science Center at San Antonio approved the research protocol with exempt status.
Study Sites/Inclusion and Exclusion Criteria
We identified all patients admitted to the study hospitals between January 1, 1999 and December 1, 2002 with a primary discharge diagnosis of pneumonia (ICD-9 codes 480.0–483.99 or 485–487.0) or secondary discharge diagnosis of pneumonia with a primary diagnosis of respiratory failure (518.81) or sepsis (038.xx). Subjects were included if they were 1) greater than 18 years of age, 2) had an admission diagnosis of community-acquired pneumonia, and 3) had a radiographically confirmed infiltrate or other finding consistent with community-acquired pneumonia on chest x-ray or CT obtained within 24 hours of admission.
Exclusion criteria included 1) having been discharged from an acute care facility within 14 days of admission, 2) transfer after being admitted to another acute care hospital, and 3) being comfort measures only on this admission. If a subject was admitted more than once during the study period, only the first hospitalization was abstracted.
Data Abstraction
Chart review data included: demographics, comorbid conditions, physical examination findings, laboratory data, and chest radiograph reports. In addition, data on important processes of care measures for patients hospitalized with community-acquired pneumonia were also abstracted: first dose of antibiotics within 4 hours and 8 hours of admission, collection of blood cultures prior to antibiotic administration, and obtaining blood cultures and oxygen saturation measurement within 24 hours of presentation [14]. Antimicrobial therapy was considered guideline-concordant if it agreed with either the 2000 Infectious Diseases Society of America or 2001 American Thoracic Society guidelines [15,16]. Information on all outpatient medications that were either 1) reported as currently being taken by the patient at presentation, or 2) listed in the electronic medical record, were recorded. Patients were defined as taking a statin if they had a statin listed on the electronic medical record (as an outpatient medication) or history and physical under outpatient medications.
Mortality was assessed using information from the Texas Department of Health and Department of Veteran Affairs clinical database. Mortality status was assessed through December 2002.
Risk Adjustment
The pneumonia severity index score was used to assess severity of illness at presentation [17]. The pneumonia severity index is a validated prediction rule for 30-day mortality in patients with community-acquired pneumonia. This rule is based on three demographic characteristics, five comorbid illnesses, five physical examination findings, and seven laboratory and radiographic findings from the time of presentation. Patients are classified into five risk classes with 30-day mortality ranging from 0.1% for class I to 27% for class V for patients enrolled in the PORT cohort study [17].
Outcome
We used 30-day mortality as the outcome for this study. Previous research has demonstrated that 30-day mortality is primarily due to the community-acquired pneumonia rather than other co-existing co-morbid conditions [18,19]. Therefore by using 30-day mortality as our outcome we are able to minimize the effect of statin use on other co-morbid conditions.
Sample Size
Sample size calculations were based on an assumption of a 30% overall utilization of statins and a 40% difference in use between those who died and survived. We calculated that 800 subjects were needed to have an 80% probability to detect a significant mortality difference at 30-days (with an α of 0.05 and β of 0.20).
Statistical Analyses
Univariate statistics were used to test the association of sociodemographic and clinical characteristics with all-cause 30-day mortality. Categorical variables were analyzed using the Chi-square test and continuous variables were analyzed using Student's t-test.
A propensity score technique was used to balance covariates associated with statin use between groups [20]. The use of the propensity score technique provides a way, in non-randomized studies, to control for pretreatment differences by defining sets of comparable patients. The propensity score was derived from a logistic regression model. A dichotomous indicator variable indexing whether a patient was on a statin was our response variable. The covariates used in the propensity score model were the pneumonia severity index score (which includes comorbid conditions such as congestive heart failure, liver disease, and history of stroke), history of alcoholism, history of diabetes mellitus, coronary artery disease, and current tobacco use. Variables were entered, and maintained, in the model if they had a p-value <0.20 in the univariate analysis (with statin use as the dependent variable) and had a p-value <0.20 in the final model.
We used a Cox proportional hazard model to estimate, and graph, the baseline survivor functions after adjusting for the propensity score and processes of care including use of guideline-concordant antibiotics, initial dose of antibiotics within 4 hours, obtaining blood cultures prior to antibiotics and within 24 hours, and assessing oxygenation at presentation.
A multivariable logistic regression model was derived with 30-day mortality as the dependent variable, and the propensity score, use of statin at presentation, and process of care measures (initial antibiotics within 4 hours and obtaining blood cultures prior to initial dose of antibiotics, and whether antimicrobial therapy was guideline concordant) as potential confounding variables. Interactions were assessed using cross-product terms between the medications and all of the other variables retained in the models. No significant interactions terms were noted, so they were excluded from the final models. All analyses were performed using STATA version 8 (Stata Corporation, College Station, Texas).
Results
Data was abstracted on 787 patients at the two hospitals. The mean age was 60 years with a standard deviation of 16 years. The population was 79% male, 84% were admitted through the emergency department, and 20% were admitted to the intensive care unit (ICU) within the first 24 hours after admission. Mortality was 9.2% at 30-days and 13.6% at 90-days. By pneumonia severity index, 52% were low risk (pneumonia severity index classes I-III), 34% were moderate risk (pneumonia severity index class IV), and 14% were high risk (pneumonia severity index class V). Regarding community-acquired pneumonia-related processes of care, 28% received the initial dose of antibiotics within 4 hours of presentation and an additional 22% received the initial antibiotic dose within 8 hours, 76% of patients had blood cultures obtained within 24 hours and prior to antibiotics, and oxygenation was assessed at presentation in 91%.
Table 1 shows the demographic factors, clinical characteristics, and processes of care data for this population by 30-day mortality. In the univariate analysis numerous individual components of the PSI were significantly associated with 30-day mortality including age, nursing home residency, history of congestive heart failure, history of malignancy, altered mental status, systolic blood pressure < 90 mmHg, tachycardia> 125 beats per minute, arterial acidosis, elevated blood urea nitrogen 30 mg/dl, serum sodium < 130 meq/l, and pleural effusion on chest radiograph. The only processes of care that were statistically significant were the assessment of oxygenation within 24 hours or presentation and use of guideline-concordant antibiotics. Statin use had only a borderline significance (p = 0.07) in the univariate analysis.
Table 1 Subject Demographic and Clinical Characteristics by 30-Day Mortality*
Variable 30-Day Mortality
Alive (n= 715) Dead (n= 72) p-valzue
Age, years +/- standard deviation 60.2+/-16.4 62.9 +/-16.4 0.09
Men 561 (79) 60 (83) 0.3
Nursing home resident 41 (6) 13 (18) <0.001
Admitted through emergency department 598 (84) 58 (81) 0.5
Admitted to intensive care within 24 hours 118 (17) 36 (50) <0.001
Preexisting Comorbid Conditions
Congestive heart failure 105 (15) 18 (25) 0.02
Chronic pulmonary disease 195 (27) 23 (31) 0.4
History of stroke 93 (13) 12 (17) 0.4
Chronic liver disease 83 (12) 11 (15) 0.4
History of malignancy 58 (8) 20 (28) <0.001
Renal insufficiency 74 (10) 13 (18) 0.05
History, Physical, Laboratory, and Radiographic Data
Altered mental status 68(10) 17(24) <0.001
Respiratory rate > 30 per minute 71 (10) 11 (15) 0.2
Systolic blood pressure < 90 mmHg 16 (2) 5 (7) 0.02
Heart rate > 125 per minute 86 (12) 19 (26) 0.001
Temperature < 95° or > 104°F 19 (3) 2 (3) 0.9
Arterial pH < 7.35 37 (5) 12 (17) <0.001
Arterial oxygenation saturation < 90% 149 (21) 27 (38) 0.001
Hematocrit < 30% 64 (9) 8 (11) 0.5
Serum blood urea nitrogen > 30 mg/dL 135 (19) 33 (46) <0.001
Serum glucose > 250 mg/dL 71 (10) 5 (7) 0.4
Serum sodium < 130 meq/L 98 (14) 18 (25) 0.01
Pleural effusion on chest radiograph 160 (11) 29 (35) 0.001
Pneumonia Severity Index
Class I-III 393 (54) 16 (22)
Class IV 240 (34) 26 (36)
Class V 82 (12) 30 (42) <0.001
Processes of Care
Initial antibiotics within 4 hours 201 (28) 22 (31) 0.7
Initial antibiotics within 8 hours 358 (50) 36 (50) 1.0
Blood cultures prior to antibiotics 540 (76) 55 (76) 0.9
Oxygenation assessed ≤ 24 hours 538 (75) 65 (90) 0.004
Guideline-concordant antibiotics used 574 (80) 51 (71) 0.05
Outpatient Medications
Statin 105 (15) 5 (7) 0.07
* Data are presented as number (%) or mean +/-standard deviation
Of the 787 subjects, 110 subjects (14%) were on statins at presentation. Table 2 demonstrates the association between clinical and demographic variables and the use/non-use of statins. Components of the PSI that were significantly associated with statin use include increased age, history of congestive heart failure, history of stroke, systolic blood pressure < 90 mmHG, and an elevated serum glucose. History of diabetes mellitus was also associated with statin use. Conditions inversely associated with statin use include: nursing home residence, history of alcoholism, chronic liver disease, current tobacco use, and pleural effusion on chest x-ray. Figure 1 displays the % survival by statin-use versus non-use over 30-days after adjusting for the propensity score and processes of care, showing that statin use is associated with higher survival at 30-days (p = 0.001).
Table 2 Use versus non-use of statin by demographic and clinical characteristics*
Variable Statin
Not on statin (n = 677) On statin (n = 110) p-value
Age, years +/- standard deviation 59.4+/-16.8 66.3+/-12.3 <0.001
Men 529(78) 92(84) 0.2
Nursing home resident 51(8) 3(3) 0.06
Admitted through emergency department 570(84) 86(78) 0.1
Admitted to intensive care within 24 hours 135(20) 19(17) 0.5
Preexisting Comorbid Conditions
Diabetes Mellitus 168(25) 62(56) <0.001
Alcoholism 79(12) 5(5) 0.03
Current tobacco use 217(32) 18(16) 0.001
Congestive heart failure 99(15) 24(22) 0.05
History of stroke 78(12) 27(25) <0.001
Chronic liver disease 91(13) 3(3) 0.001
History of malignancy 68(10) 10(9) 0.8
Renal insufficiency 74(11) 13(12) 0.8
History, Physical, Laboratory, and Radiographic Data
Altered mental status 76(11) 9(8) 0.3
Respiratory rate > 30 per minute 69(10) 13(11) 0.6
Systolic blood pressure < 90 mmHg 21(3) 0(0) 0.06
Heart rate > 125 per minute 96(14) 9(8) 0.09
Temperature < 95° or > 104°F 17(3) 4(4) 0.5
Arterial pH < 7.35 39(6) 10(9) 0.2
Arterial oxygenation saturation < 90% 151(22) 25(22) 0.9
Hematocrit < 30% 65(10) 7(6) 0.3
Serum blood urea nitrogen > 30 mg/dL 146(22) 22(20) 0.7
Serum glucose > 250 mg/dL 58(9) 18(16) 0.01
Serum sodium < 130 meq/L 105(15) 11(10) 0.13
Pleural effusion on chest radiograph 172(25) 17(15) 0.02
Pneumonia Severity Index
Class I-III 357(53) 52(47)
Class IV 222(33) 44(40)
Class V 98(14) 14(13) 0.3
* Data are presented as number (%) or mean +/-standard deviation
Figure 1 Proportion of surviving patients hospitalized with community-acquired pneumonia by use of statin versus non-use after adjusting for the propensity score and other potential confounders (p = 0.001).
In the multivariable regression analysis, after adjusting for the propensity score and processes of care, the use of statins at presentation (odds ratio 0.36, 95% confidence interval 0.14–0.92) was significantly associated with decreased 30-day mortality.
Discussion
We found that prior outpatient use of statins was associated with decreased 30-day mortality for subjects hospitalized with community-acquired pneumonia. Our findings provide further support to previous work that demonstrate that statin use is associated with decreased mortality for patients with acute bacterial illnesses [12,13]. Further studies are needed to examine the impact of statins, both pre-hospitalization and acute, on patients hospitalized with community-acquired pneumonia and other bacterial illnesses.
Our study, with a methodologically stronger cohort design, supports the findings of the recent case-control study which demonstrated that patients hospitalized with bacteremia who were on statins at admission had a significant reduction in in-hospital mortality (28% versus 6%, p<0.002) [12]. In the multivariate analysis, after adjustment for confounding factors (including comorbid conditions, age, concurrent medications, site of infection, vital signs, and laboratory data) not being on a statin (odds ratio 7.6, 95% confidence interval 1.01–57.5) was significantly associated with mortality. This prior research, combined with our results, supports the need for further research to examine the impact of statins in the treatment of infectious diseases.
Although our study was retrospective and subject to the recognized limitations of such studies, we carefully assembled our cohort from complete patient discharge data to avoid ascertainment bias. Additionally, during chart abstraction we encountered a very small amount (<5%) of missing data. Our sample was predominantly men due to the inclusion of a VA hospital and it is possible, but unlikely, that women may have differential responsiveness to statins as compared to men. Also we are unable to assess factors such as duration of statin use, inpatient continuation of the statin, or the dose effect due to the design of this study. In addition we are unable to control for quality of health care that patients had prior to hospitalization. Further research is needed to examine these factors. Finally, as in any non-experimental study, we are unable to state conclusively that the prior outpatient use of statin is the cause of decreased mortality in this cohort. However, since patients on statins have numerous medical conditions that are significantly associated with increased short-term morality we feel that we have good evidence that these medications may have beneficial effects for patients hospitalized with community-acquired pneumonia.
Conclusion
Our study finds that prior outpatient use of statins reduces mortality for patients hospitalized with community-acquired pneumonia. Our results add an additional potential benefit of statin use to the already compelling data for their use in patients with coronary artery disease, hypercholesterolemia, diabetes, and peripheral vascular disease. Additionally, patients with diabetes and vascular disease are at higher risk for either contracting pneumonia or dying from pneumonia when they do contract it. Further studies are needed to confirm the magnitude of the impact of statins, either pre-hospitalization or acute, on patients hospitalized with community-acquired pneumonia and to elucidate the mechanism by which they may work.
Competing interests
None of the authors, except for Dr. Anzueto, have any conflicts of interests to disclose regarding this paper. Dr. Anzueto is currently a consultant with Pfizer, Ortho-McNeil, and Bayer Pharma.
Authors' contributions
EMM originated and coordinated the study, obtained funding, contributed to the analysis of the data, and preparation of the paper.
MIR contributed to the design of the study, contributed to the analysis of the data, and preparation of the paper.
AA contributed to the design of the study and preparation of the paper.
JP contributed to the design of the study, contributed to the analysis of the data, and preparation of the paper.
Acknowledgements
Dr. Mortensen was supported by Howard Hughes Medical Institute faculty-start up grant 00378-001 and a Department of Veteran Affairs Veterans Integrated Service Network 17 new faculty grant. Dr. Pugh was supported by Department of Veteran Affairs grant HFP98-002. This material is the result of work supported with resources and the use of facilities at the South Texas Veterans Health Care System. The funding agencies had no role in conducting the study, or role in the preparation, review, or approval of the manuscript.
The views expressed in this article are those of the authors and do not necessarily represent the views of the Department of Veterans Affairs.
==== Refs
Hoyert DL Arias E Smith BL Deaths: Final Data for 1999 Natl Vital Statistics Report 2001 49 8 1 113
Gilbert K Fine MJ Assessing prognosis and predicting patient outcomes in community-acquired pneumonia Seminars in Respiratory Infections 1994 9 3 140 152 7831536
Jialal I Stein D Balis D Grundy SM Adams-Huet B Devaraj S Effect of hydroxymethyl glutaryl coenzyme a reductase inhibitor therapy on high sensitive C-reactive protein levels Circulation 2001 103 15 1933 1935 11306519
Musial J Undas A Gajewski P Jankowski M Sydor W Szczeklik A Anti-inflammatory effects of simvastatin in subjects with hypercholesterolemia International Journal of Cardiology 2001 77 2-3 247 253 11182189 10.1016/S0167-5273(00)00439-3
de Bont N Netea MG Rovers C Smilde T Demacker PN van der Meer JW Stalenhoef AF LPS-induced cytokine production and expression of LPS-receptors by peripheral blood mononuclear cells of patients with familial hypercholesterolemia and the effect of HMG-CoA reductase inhibitors Atherosclerosis 1998 139 1 147 152 9699902 10.1016/S0021-9150(98)00074-4
Rosenson RS Tangney CC Casey LC Inhibition of proinflammatory cytokine production by pravastatin Lancet 1999 353 9157 983 984 10459915 10.1016/S0140-6736(98)05917-0
Ridker PM Rifai N Pfeffer MA Sacks FM Moye LA Goldman S Flaker GC Braunwald E Inflammation, pravastatin, and the risk of coronary events after myocardial infarction in patients with average cholesterol levels. Cholesterol and Recurrent Events (CARE) Investigators Circulation 1998 98 9 839 844 9738637
Strandberg TE Vanhanen H Tikkanen MJ Effect of statins on C-reactive protein in patients with coronary artery disease Lancet 1999 353 9147 118 119 10023901 10.1016/S0140-6736(05)76154-7
Moussa K Michie HJ Cree IA McCafferty AC Winter JH Dhillon DP Stephens S Brown RA Phagocyte function and cytokine production in community acquired pneumonia. Thorax 1994 49 2 107 111 8128397
Puren AJ Feldman C Savage N Becker PJ Smith C Patterns of cytokine expression in community-acquired pneumonia Chest 1995 107 5 1342 1349 7750329
Bauer TT Monton C Torres A Cabello H Fillela X Maldonado A Nicolas JM Zavala E Comparison of systemic cytokine levels in patients with acute respiratory distress syndrome, severe pneumonia, and controls Thorax 2000 55 1 46 52 10607801 10.1136/thorax.55.1.46
Liappis AP Kan VL Rochester CG Simon GL The effect of statins on mortality in patients with bacteremia Clinical Infectious Diseases 2001 33 8 1352 1357 11565076 10.1086/323334
Seraphin LM Liappis AP Kan VL Simon GL Increased incidence of lower extremity infections among diabetic patients receiving statins.: September and Decmeber 2001.
2001 76
Meehan TP Fine MJ Krumholz HM Scinto JD Galusha DH Mockalis JT Weber GF Petrillo MK Houck PM Fine JM Quality of care, process, and outcomes in elderly patients with pneumonia JAMA 1997 278 23 2080 2084 9403422 10.1001/jama.278.23.2080
Niederman MS Mandell LA Anzueto A Bass JB Broughton WA Campbell GD Dean N File T Fine MJ Gross PA Martinez F Marrie TJ Plouffe JF Ramirez J Sarosi GA Torres A Wilson R Yu VL Guidelines for the management of adults with community-acquired pneumonia. Diagnosis, assessment of severity, antimicrobial therapy, and prevention Am J Respir Crit Care Med 2001 163 7 1730 1754 11401897
Bartlett JG Dowell SF Mandell LA File Jr TM Musher DM Fine MJ Practice guidelines for the management of community-acquired pneumonia in adults. Infectious Diseases Society of America Clin Infect Dis 2000 31 2 347 382 10987697 10.1086/313954
Fine MJ Auble TE Yealy DM Hanusa BH Weissfeld LA Singer DE Coley CM Marrie TJ Kapoor WN A prediction rule to identify low-risk patients with community-acquired pneumonia N Engl J Med 1997 336 4 243 250 8995086 10.1056/NEJM199701233360402
Mortensen EM Kapoor WN Chang CC Fine MJ Assessment of mortality after long-term follow-up of patients with community-acquired pneumonia Clin Infect Dis 2003 37 12 1617 1624 14689342 10.1086/379712
Mortensen EM Coley CM Singer DE Marrie TJ Obrosky DS Kapoor WN Fine MJ Causes of death for patients with community-acquired pneumonia: results from the Pneumonia Patient Outcomes Research Team cohort study Arch Intern Med 2002 162 9 1059 1064 11996618 10.1001/archinte.162.9.1059
Stone RA Obrosky DS Singer DE Kapoor WN Fine MJ Propensity score adjustment for pretreatment differences between hospitalized and ambulatory patients with community-acquired pneumonia. Pneumonia Patient Outcomes Research Team (PORT) Investigators Med Care 1995 33 4 Suppl AS56 66 7723462
|
16042797
|
PMC1199623
|
CC BY
|
2021-01-04 16:23:25
|
no
|
Respir Res. 2005 Jul 25; 6(1):82
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-82
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-871605353210.1186/1465-9921-6-87ResearchAlveolar macrophage-epithelial cell interaction following exposure to atmospheric particles induces the release of mediators involved in monocyte mobilization and recruitment Ishii Hiroshi [email protected] Shizu [email protected] James C [email protected] Takeshi [email protected] Yukinobu [email protected] Noriho [email protected] Hiroshi [email protected] Renaud [email protected] Eeden Stephan F [email protected] James Hogg iCAPTURE Centre for Cardiovascular and Pulmonary Research, St. Paul's Hospital, University of British Columbia, 1081 Burrard Street, Vancouver, BC, V6Z 1Y6, Canada2 Second Department of Internal medicine, Nagasaki University School of Medicine, Nagasaki, Japan3 Environmental Health Directorate, Health Canada, Ottawa, Ontario, Canada2005 1 8 2005 6 1 87 87 4 1 2005 1 8 2005 Copyright © 2005 Ishii et al; licensee BioMed Central Ltd.2005Ishii et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Studies from our laboratory have shown that human alveolar macrophages (AM) and bronchial epithelial cells (HBEC) exposed to ambient particles (PM10) in vitro increase their production of inflammatory mediators and that supernatants from PM10-exposed cells shorten the transit time of monocytes through the bone marrow and promote their release into the circulation.
Methods
The present study concerns co-culture of AM and HBEC exposed to PM10 (EHC-93) and the production of mediators involved in monocyte kinetics measured at both the mRNA and protein levels. The experiments were also designed to determine the role of the adhesive interaction between these cells via the intercellular adhesion molecule (ICAM)-1 in the production of these mediators.
Results
AM/HBEC co-cultures exposed to 100 μg/ml of PM10 for 2 or 24 h increased their levels of granulocyte-macrophage colony-stimulating factor (GM-CSF), M-CSF, macrophage inflammatory protein (MIP)-1β, monocyte chemotactic protein (MCP)-1, interleukin (IL)-6 and ICAM-1 mRNA, compared to exposed AM or HBEC mono-cultures, or control non-exposed co-cultures. The levels of GM-CSF, M-CSF, MIP-1β and IL-6 increased in co-cultured supernatants collected after 24 h exposure compared to control cells (p < 0.05). There was synergy between AM and HBEC in the production of GM-CSF, MIP-1β and IL-6. But neither pretreatment of HBEC with blocking antibodies against ICAM-1 nor cross-linking of ICAM-1 on HBEC blocked the PM10-induced increase in co-culture mRNA expression.
Conclusion
We conclude that an ICAM-1 independent interaction between AM and HBEC, lung cells that process inhaled particles, increases the production and release of mediators that enhance bone marrow turnover of monocytes and their recruitment into tissues. We speculate that this interaction amplifies PM10-induced lung inflammation and contributes to both the pulmonary and systemic morbidity associated with exposure to air pollution.
==== Body
Background
Exposure to ambient particulate matter with a diameter of less than 10 μm (PM10) is strongly associated with increased morbidity and mortality, particularly in subjects with pre-existing pulmonary and cardiovascular diseases [1,2]. This increase in mortality induced by PM10 exposure was present even when adjusted for the other major risk factors such as cigarette smoking [1]. A recent report [3] has shown that environmentally relevant concentrations of PM2.5 induced airway inflammation even in healthy subjects with a selective influx of monocytes.
Although the biological mechanisms are still unclear, PM10 are known to stimulate the production of reactive oxygen species and inflammatory mediators by alveolar macrophages (AM) [4-7] and epithelial [7-10] and other lung cells [11]. When AM and airway epithelial cells are directly exposed to inhaled atmospheric particles these small particles are phagocytized by both cells [10,12]. Both cell types can synthesize a variety of pro-inflammatory cytokines that induce airway inflammation and contribute to the airway lesions in asthma and chronic obstructive pulmonary diseases [9]. In vitro, AM and lung epithelial cells interact in response to PM10 and this interaction has been implicated in amplifying their mediator production [7,13]. Studies from our laboratory have shown that the PM10(EHC-93)-induced interaction of human AM and bronchial epithelial cells (HBEC) enhances the synthesis and release of a variety of pro-inflammatory cytokines and that supernatants from these co-cultures instilled into rabbit lungs induces a systemic inflammatory response [13].
We recently showed that deposition of PM10 (EHC-93 and inert carbon particles) in the lung shortened the transit time of monocytes through the bone marrow and enhanced their release into the circulation [14,15]. Furthermore, we also showed that monocytes are the predominant inflammatory cells that accumulate in the alveoli following repeated PM10 exposure [16]. The present study was designed to determine whether, and if so, which interactions between AM and HBEC (AM/HBEC co-cultures) amplify the response to PM10 exposure, especially the synthesis of inflammatory mediators that enhance bone marrow turnover of monocytes and their recruitment into the lung. We used primary cultures of HBEC and human AM freshly isolated from lobectomy or pneumonectomy specimens and measured the expression of inflammatory mediators relevant to monocyte kinetics. We further evaluated the potential role of the intercellular adhesion molecule (ICAM)-1 in the production of mediators by AM/HBEC co-cultures exposed to PM10.
Methods
Urban air particles (PM10)
PM10 particles were collected in an urban environment (EHC-93) and obtained from the Environmental Health Directorate, Health Canada, Ottawa, Ontario. A detailed analysis of the EHC-93 has been presented elsewhere [17]. Particles were suspended at a concentration of 1 mg/ml in hydrocortisone-free supplemented bronchial epithelial cell growth medium (BEGM; Clonetics, San Diego, CA) and sonicated 3 times for 1 min each at maximal power on a Vibra Cell VC-50 sonicator (Sonics and Materials Inc., Danbury, CT) prior to adding to the cells. The endotoxin content of the PM10 suspension of 100 μg/ml was 6.4 ± 1.8 EU/ml or less than 3.0 ng/ml [10,13]. This dose of LPS has been shown not to activate either AM or lung epithelial cells to produce cytokines [10].
Isolation of HBEC and human AM
Bronchial tissue and broncho-alveolar lavage (BAL) fluid was obtained from a total of ten patients who underwent lobectomy or pneumonectomy for small peripheral nodules at St. Paul's Hospital, Vancouver. Informed consent was obtained from all subjects and these studies were approved by the Human Ethics Committee of the University of British Columbia. All subjects were current smokers and were asked to abstain from smoking for 6 weeks prior to the operation. Their mean age was 67.2 yr (range 56–74 yr) (6 women and 4 men). Primary HBEC were isolated from bronchial tissues according to a previously described procedure [10]. In brief, pieces of excised human bronchial tissue approximately 1 cm long were incubated at 4°C for 24 h with 0.1% protease (Type14; Sigma) solution prepared in BEGM containing Fungizone (1 μg/ml; GIBCO BRL, Gaithersburg, MD). The epithelial cells were harvested, washed with BEGM with added antibiotics(100 U/ml of penicillin and 100 μ/gml of streptomycin; Sigma)and Fungizone, and cultured in a 25-cm2 cell culture flask until 80 to 90% confluent. Then the cells were trypsinized and placed in 100-mm cell culture dishes and cultured in BEGM. Light microscopy showed that 95% of the isolated cells had features of bronchial epithelial cells, that is they formed a monolayer of ciliated cells. Also, by trypan blue exclusion, >95% of these cells were viable. Human AM were harvested from BAL fluid obtained from lung segments or lobes that were free of the tumor using a method previously described in detail [7,13]. The BAL fluid cells were >90% viable (trypan blue exclusion method) and consisted of 90–95% AM (as assessed by Wrights-Giemsa stain) and less than 2% neutrophils. AM mono-cultures and AM/HBEC co-cultures were suspended in BEGM. BEGM used throughout this study was without hydrocortisone.
Exposure of cells to PM10
Primary HBEC from the third or fourth passage of cells from each patient were cultured to 90–100 % confluence in 100-mm cell culture dishes (approximately 2.5–3.0 × 106 cells/dish) then exposed for 2 and 24 h to fresh stock suspensions of 100 μg/ml PM10 (EHC-93) prepared in BEGM.
AM (1.0 × 107) from each patient were placed in 100-mm cell culture dishes and allowed to adhere to the plastic dish for 30 min in humidified incubator (5% CO2 at 37°C). The non-adherent cells less than 1.0 × 106) were then removed by rinsing twice with BEGM and adherent AM (>98% AM) were incubated in 10 ml of BEGM with or without 100 μg/ml of PM10 for 2 and 24 h.
In co-culture experiments, freshly prepared AM (5.0 × 106) were directly placed on the confluent HBEC monolayers which were grown in 100-mm cell culture dishes. The AM were allowed to adhere to HBEC and the non-adherent cells were removed by washing twice with BEGM. The AM/HBEC co-cultured cells were incubated in 10 ml of BEGM with or without 100 μg/ml of PM10 for 2 and 24 h. Cell viability was determined following the 24 h PM10 exposure in all experiments using the trypan blue exclusion method.
RNase protection assay (RPA)
After 2 or 24 h treatment, total RNA was isolated from the cells using a single-step phenol/chloroform extraction procedure (Trizol, Life Technologies, Inc., Grand Island, NY). The levels of inflammatory mediator mRNA were determined using the RiboQuant™ multi-probe system (PharMingen, San Diego, CA) following the instructions of the supplier. Two customized template sets were used that included mRNAs of the following inflammatory mediators: human regulated on activation, normal T-cells expressed and secreted (RANTES), macrophage inflammatory protein (MIP)-1β, granulocyte-macrophage colony-stimulating factor (GM-CSF), M-CSF, monocyte chemotactic protein (MCP)-1, interleukin (IL)-6 and leukemia inhibitory factor (LIF). Human ICAM-1 mRNA was determined using a separate template set. Internal controls included mRNAs of the ribosomal protein L32 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH). In brief, 10 μg of total cellular RNA was hybridized overnight to the [α-32P] UTP-labeled riboprobes which had been synthesized from the supplied template sets. Single-stranded RNA and free probe remaining after hybridization were digested by a mixture of RNase A and T1. The protected RNA was then phenolized, precipitated, and analyzed on a 5% denaturing polyacrylamide gel. Following electrophoresis, the gel was dried under vacuum and subjected to autoradiography. The quantity of protected labeled RNA was determined using densitometry and the NIH image 1.63 software (National Institutes of Health, Bethesda, MD). Results were normalized to the expression of the internal control, GAPDH. For the densitometric analysis each RPA was repeated four to six times.
ELISA measurements
Cell culture supernatants were collected 24 h after addition of 100 μg/ml of PM10 suspension, centrifuged, filtered through a syringe filter with pore size of 0.22 μm (Corning, Cambridge, MA) to eliminate as much as possible any remaining particles and stored at -80°C until use. MIP-1β, GM-CSF, M-CSF, MCP-1 and IL-6 levels were measured by the Cytokine Core Laboratory (Baltimore, MD) using an ELISA based on a biotin-strepavidin-peroxidase detection system as previously described [10]. All measurements were done in triplicate and values corrected for the number of AM used in each experiment are reported as the means of five experiments.
Immunocytochemistry
To demonstrate cell surface ICAM-1 (CD54) expression on HBEC and CD11b on AM, cells were placed or grown on coverslips in 6-well plates and incubated for 2 or 24 h with 100 μg/ml of PM10. Cells were fixed with 1% paraformaldehyde for 10 min and immunocytochemistry was performed by the alkaline phosphatase anti-alkaline phosphatase method using mouse anti-human CD54 monoclonal antibody (Immunotech, Marseille, France) and mouse anti-human CD11b monoclonal antibody (DAKO, Copenhagen, Denmark) to identify cell surface expression of ICAM-1 and CD11b.
Cell adhesion blockers and ICAM-1 cross-linking
In experiments testing whether anti-CD54 and anti-CD11b block mediator production by co-cultured AM/HBEC, HBEC and AM were preincubated for 1 h before PM10 exposure with control IgG F(ab')2 fragments (2 μg/ml; Jackson ImmunoResearch Laboratories, PA), mouse anti-human monoclonal CD54 F(ab')2 fragments, and/or monoclonal CD11b F(ab')2 fragments (1 μg/ml, respectively; Caltag Laboratories, CA). Cells were then co-cultured and exposed to PM10 for 24 h in the presence of the blocking antibodies before analysis by RPA. To determine whether ligand binding to CD54 on HBEC in of itself contributes to the enhanced mediator response of these cells to PM10 stimulation cross-linking antibodies to CD54 were used to simulate this possibility. We used previously reported methods of cross-linking CD54 which resulted in intracellular signaling [18,19]. After 2 h of exposure to PM10, HBEC were incubated for 1 h with 1 μg/ml mouse anti-human CD54 or 1 μg/ml control mouse non-specific IgG (DAKO). Cells were washed and then incubated for 4 h with 10 μg/ml rabbit anti-mouse IgG (DAKO) to cross-link the bound anti-CD54 and mRNA mediator expression was assessed as above.
Statistical Analysis
Data are expressed as mean values ± SE. The minimum number of replicates for all measurement was at least three. For RPA and ELISA, differences between matched pairs (control versus PM10 treated) were compared by Wilcoxon signed ranked test. To compare mediator production by co-cultures to that by AM plus HBEC mono-cultures, we used the Mann-Whitney U test. Differences between multiple groups were compared by one-way analysis of variance (ANOVA). The post hoc test for multiple comparisons was the Dunnett's test. Significance was assumed at p < 0.05.
Results
AM/HBEC co-cultures and PM10
We previously showed that the majority of AM and HBEC were in contact with each other in our co-culture system [13]. Both cells internalized PM10 particles with many cells containing more than one particle. The 100 μg/ml concentration of PM10 used throughout this experiment was not toxic to either AM or HBEC and >90% of cells were viable after 24 h exposure as assessed by the trypan blue exclusion method.
Expression of mRNA induced by PM10
Representative autoradiographs of mRNA expression by AM or HBEC mono-cultures and AM/HBEC co-cultures after 2 and 24 h incubation in medium alone (control) or a 100 μg/ml of PM10 suspension (PM10) are shown in Figure 1. Because the RPA kit is not provided with an internal control to account for variation between autoradiographs of different pairs of control versus PM10 treated cells such as those shown in Figure 1 and due to that fact that cells from a single but different patient are represented in each different pair, as documented by the differences in intensity of the control L32 band(s) compared to that of the corresponding GAPDH bands as well as differences in the L32 banding pattern (Fig. 1), expected large variations in densitometric data between corresponding pairs of autoradiographs were found. Despite these variations the compiled densitometric analyses of these autoradiographs yielded statistically significant results. However, because of the unavoidable variations, the compiled densitometric results for a few mediators differed from that depicted in the representative autoradiographs. In Figure 1 mRNA expression of the inflammatory mediators of interest, RANTES, MIP-1β, GM-CSF, M-CSF, MCP-1, IL-6 and LIF, was not altered after 2 h of PM10 exposure of neither AM nor HBEC mono-cultures and this result was confirmed after densitometric analysis (n = 4, data not shown). Only the expression of ICAM-1 mRNA by HBEC at this time-point appears to be marginally increased (Fig. 1) but after densitometric analysis this change was not found to be significant (n = 4, data not shown). In contrast to the results from the mono-cultures, PM10 exposure for 2 h of co-cultured AM/HBEC increased MIP-1β, GM-CSF, M-CSF, IL-6, LIF and ICAM-1 mRNA expression (Fig. 1) and, of these, densitometric analysis of six such RPA experiments confirmed that increases in MIP-1β, GM-CSF, IL-6, and ICAM-1 were significant (Fig. 2A), as well as that of MCP-1 (Fig. 2A) which was not detected in the representative autoradiograph (Fig. 1).
Figure 1 RNase protection assay of mRNA expression by AM and HBEC. Representative autoradiographs of RNase protection assays (RPAs) showing mediator expression by AM/HBEC co-cultures, AM mono-cultures and HBEC mono-cultures after 2 and 24 h incubation in medium alone (control) or a 100 μg/ml of PM10 suspension (PM10). After 24 h exposure AM showed increased expression of LIF and ICAM-1 mRNA. Expression of GM-CSF, LIF and ICAM-1 mRNA by HBEC was increased by 24 h PM10 stimulation compared to their respective controls. MIP-1β, GM-CSF, M-CSF, MCP-1, IL-6, LIF and ICAM-1 mRNA expression by AM/HBEC co-cultures was increased 2 and/or 24 h after incubation with PM10 compared to control. L32 and GAPDH were used as controls for lane loading.
Figure 2 Densitometric analysis of bands on RPAs. (A): the density of the bands representing the mediator mRNAs in AM/HBEC co-cultures on autoradiographs such as that shown in Figure 1A was compared to that of the GAPDH mRNA band in the same lane and the resulting ratio (PM10; black bars) is shown as the percentage change from control values (white bars). The mean densitometric value confirmed that the mRNA levels of MIP-1β, GM-CSF, MCP-1, IL-6 and ICAM-1 at 2 h and those of MIP-1β, GM-CSF, M-CSF and ICAM-1 after 24 h exposure were significantly higher than control values. Values are means ± SE of six experiments representing the AM/HBEC co-culture group. (B): the mean densitometric value confirmed that the mRNA levels of GM-CSF, LIF and ICAM-1 at 24 h exposure were significantly higher than control values. Values are means ± SE of four experiments representing the HBEC mono-culture group. *p < 0.05 compared with control.
After 24 h exposure to PM10 the representative autoradiographs showed that increases in mRNA expression by AM mono-cultures were restricted to that of LIF and ICAM-1 (Fig. 1) but this was not confirmed after statistical analysis of the densitometric results (n = 4, data not shown). In contrast, the increases in GM-CSF, LIF and ICAM-1 by HBEC mono-cultures (Fig. 1) were found to be statistically significant (p < 0.05 and n = 4, respectively)(Fig. 2B). Co-cultures exposed to PM10 at this time-point showed strong increases in MIP-1β, GM-CSF, M-CSF, IL-6 and ICAM-1 mRNA and minor increases in those of MCP-1 and LIF (Fig. 1). Except for IL-6, the strong increases were confirmed by the densitometric analysis of six RPA experiments (Fig. 2A).
Mediator production induced by PM10
Figure 3 shows the GM-CSF, IL-6, MIP-1β, MCP-1 and M-CSF protein levels in supernatants of AM/HBEC co-cultures, AM mono-cultures and HBEC mono-cultures incubated for 24 h with medium alone (control) or with 100 μg/ml of PM10. GM-CSF, IL-6, MIP-1β and M-CSF production by AM/HBEC co-cultures stimulated by PM10 were significantly increased compared to control levels. GM-CSF and IL-6 production by AM mono-cultures stimulated with PM10 suspension increased significantly compared to controls. MIP-1β production by HBEC mono-culture stimulated by PM10 were significantly increased over control levels.
Figure 3 Mediator protein levels in supernatants of AM and HBEC. GM-CSF, IL-6, MIP-1β, MCP-1, and M-CSF protein levels in supernatants of AM mono-cultures, HBEC mono-cultures and AM/HBEC co-cultures incubated for 24 h with medium alone (control; white bars) or 100 μg/ml of PM10 (black bars). GM-CSF and IL-6 production by AM mono-cultures and AM/HBEC co-cultures stimulated by PM10 increased significantly compared to controls. Exposure to PM10 also increased MIP-1β production by HBEC mono-cultures and AM/HBEC co-cultures and M-CSF production by AM/HBEC co-cultures. The GM-CSF, IL-6 and MIP-1β produced by exposed AM/HBEC co-cultures significantly exceeded the sum of those produced by AM and HBEC mono-cultures exposed separately. Values are means ± SE of five experiments. * p < 0.05 compared with control. † p < 0.05 for exposed AM/HBEC co-cultures compared to the sum of the exposed HBEC and AM mono-cultures.
The GM-CSF, IL-6 and MIP-1β produced by AM/HBEC co-cultures in response to PM10 stimulation were more than the sum of the respective mediator produced by PM10 exposed AM and HBEC mono-cultures alone suggesting a synergistic effect in production of these cytokines (p < 0.05). This synergistic effect was not seen in the production of M-CSF. MCP-1 production was not significantly increased by PM10 in either mono-cultures or AM/HBEC co-cultures but its expression by AM tended to be decreased by co-culturing (Fig. 3).
Expression of ICAM-1 induced by PM10
Figure 4 shows immunocytochemically stained CD54 on HBEC (to identify ICAM-1) and CD11b on AM. In the absence of PM10, HBEC express low levels of ICAM-1 on their cell surface (Fig. 4A). After stimulation with 100 μg/ml of PM10 for 24 h many more cells stained positively for ICAM-1 and their intensity of staining was increased (Fig. 4B). Most AM expressed surface CD11b and this expression was unaffected by 2 and 24 h stimulation with PM10 (Fig. 4C, D and data not shown).
Figure 4 Surface expression of ICAM-1 on HBEC and CD11b on AM. Photomicrographs of primary cultured HBEC and human AM on coverslips. Immunocytochemistry was performed using mouse anti-human CD54 monoclonal antibody on HBEC and mouse anti-human CD11b monoclonal antibody on AM. In the absence of PM10 stimulation HBEC rarely expressed CD54 (A). After stimulation with 100 μg/ml of PM10 for 24 h the majority of cells stained positively (arrows, pink cells) for CD54 (B). Expression of surface CD11b on AM (C) was unaffected by 2 h stimulation with PM10 (D). The scale bars represent 20 μm.
ICAM-1 and PM10-induced mediator production by AM/HBEC co-cultures
To determine the role of β2-integrin/ICAM-1 interaction in mediator production by AM/HBEC co-cultures, AM and HBEC were incubated with inhibitors of these adhesion molecules before PM10 exposure. Representative autoradiographs of mRNA expression by such AM/HBEC co-cultures after 24 h incubation in a 100 μg/ml of PM10 suspension are shown in Figure 5A. They include pretreatment of neither cell before co-culture, of only AM with control IgG or anti-CD11b antibody, of only HBEC with control IgG or anti-CD54 antibody, and of both cell types with both antibodies. The increased mRNA expression in PM10-stimulated AM/HBEC co-cultures was not affected by any of the pretreatments with these antibodies. In addition, as shown in Figure 5B, CD54 cross-linking itself in the absence of AM did not induce mediator expression in HBEC exposed to PM10.
Figure 5 AM/HBEC co-culture responses after pretreatment with cell adhesion blockers. (A): autoradiographs from RNase protection assay of mediator mRNA expression by AM/HBEC co-cultures pretreated before 24 h incubation in a 100 μg/ml of PM10 suspension including no pretreatment (no treatment) before co-culture, AM pretreated with control IgG (AM-IgG), HBEC with control IgG (HBEC-IgG), AM with anti-CD11b antibody (AM-CD11b), HBEC with anti-CD54 antibody (HBEC-CD54) and both cell types with respective antibodies (both antibodies). The mRNA expression in PM10-exposed AM/HBEC co-cultures was not affected by any pretreatments with these antibodies. (B): in the absence of AM, pretreatment of HBEC to cross-link CD54 with antibody followed by 2 h exposure to PM10 (PM10-CD54) did not alter mediator expression compared with HBEC pretreated with control IgG (PM10-IgG) and non-pretreated HBEC (control).
Discussion
AM and lung epithelial cells play a key role in processing inhaled particulate matter. In the present study we confirmed that exposing co-cultures of human AM and HBEC atmospheric particles to for 2 hr increased mRNA expression of GM-CSF, MCP-1 and IL-6 [13]. The current addition of mRNAs, that of M-CSF and MIP-1β, to this list of these mediators involved in the marrow production, mobilization and recruitment of monocytes that are increased in response to PM10 exposure reinforces the hypothesis that exposure of the lung to environmental pollutants can stimulate a systemic inflammatory response [4]. Besides these bone marrow oriented mediators, mRNA expression of ICAM-1, an adhesion molecule potentially involved in an interaction between AM and HBEC to amplify marrow-related mediator expression, was increased. Another important hitherto unreported finding, that of sustained increased expression of many of these mediator mRNAs, MIP-1β, GM-CSF, M-CSF, and ICAM-1, over 24 h of exposure, supports the robust increase in the expression of the corresponding mediator proteins that we observed. These included MIP-1β, M-CSF, ICAM-1 as well as the previously reported GM-CSF and IL-6 [13]. Furthermore, the synergistic increases in GM-CSF, IL-6 and MIP-1β secretion by the co-cultures compared to the sum of the mono-cultures in response to PM10 exposure indicate an interaction between these cells with ICAM-1 possibly contributing to this interaction.
IL-6, the hematopoietic growth factors GM-CSF and M-CSF, and the C-C chemokine MIP-1 are important mediators in the production and mobilization of monocytes from the bone marrow [20-22]. IL-6 is considered an important multifunctional cytokine involved in the regulation of a variety of cellular responses, including being a permissive factor for monocytic colony formation by human hematopoietic progenitor cells in combination with GM-CSF [23]. Monocytes recruited into the lung play a critical important role in clearing foreign material such as particles from the lung which underscores the importance of mediators such as GM-CSF as both a pro-inflammatory but also an anti-inflammatory mediator. This anti-inflammatory role is supported by studies that showed that GM-CSF has a protective role against pulmonary fibrosis [24] or hyperoxic lung injury [25] in animal models. Both IL-6 and GM-CSF stimulate the marrow to produce and release monocytes while the acute response cytokines, IL-1 and TNF-α, secreted in response to PM10 stimulation by AM [7,13] induce the production of monocytic chemoattractants such as MCP-1 [20,21,26-29]. MIP-1β is a chemotactic factor for human monocytes similar to MIP-1α [22]. Because PM10 did not induce MIP-1β production in human AM [4] or its mRNA in HBEC in the current study, increased MIP-1β expression in the co-cultures most likely relies on an interaction between these two cells. The significance of such an interaction is reinforced by our finding that the production of this chemokine in response to PM10, along with that of GM-CSF and IL-6, is synergistically increased, as noted above, when AM and HBEC are co-cultured. Such a synergistic increase in mediator production could augment the release of both monocytes and polymorphonuclear leukocytes from the bone marrow observed after stimulation by mediators produced by AM incubated alone with EHC-93 ex vivo [6] and thus contribute to a similar response to in vivo exposure to the ambient particles [6,15].
MCP-1 was the other C-C chemokine that we studied. Along with additional support from results from our laboratory [15], Rosseau and colleagues [30] have shown that the induction of MCP-1 in AM is a major contributor to the recruitment of peripheral blood monocytes into the alveolar compartment. In the present study we showed that production of MCP-1 by AM was just marginally increased by PM10 exposure (p = 0.07). Interestingly, the production and release of MCP-1 by AM/HBEC co-cultures tended (not significant) to be lower than by AM alone (Fig. 3). In AM/HBEC co-cultures, expression of MCP-1 mRNA was significantly increased by PM10 after 2 h but not 24 h exposure suggesting suppression of MCP-1 expression following prolonged exposure of lung cells to particles. This suggests a translational or post-translational control of MCP-1 production and could be an important immunomodulatory pathway by which the local inflammatory reaction in the lung is controlled after PM10 exposure. Together, our findings suggest that both colony stimulating factors and chemokines are released from lung cells following the inhalation of atmospheric particles and that these mediators are critically important in the production and the release of monocytes from the marrow as well as their recruitment into the lung.
The close proximity of AM and epithelial cells in the lung suggests that interaction between these cells is critically important in generating inflammatory mediators in response to noxious stimuli. Previous studies from our laboratory [13] support this concept showing that AM and epithelial cells in co-culture interact to amplify their pro-inflammatory mediator mRNA generation in response to PM10 exposure compared to exposure of mono-cultures of these cells. That soluble factors contribute to this interaction was shown when conditioned media from PM10-stimulated AM induced increases in mRNA expression of many of these mediators in HBEC [13]. On the other hand, that cellular contact between different lung cells (e.g., epithelial, endothelial cells, and fibroblast) is necessary for cell activation and cytokine production [11,31] has also been demonstrated. Along these lines, we recently showed increased expression of ICAM-1 mRNA after incubation of HBEC with conditioned media from PM10-stimulated AM and this response was blocked by neutralizing antibodies to TNF-α and IL-1β [7]. While TNF-α and IL-1β appear to be major players in the interaction between AM and HBEC in response to PM10, our results suggest that ICAM-1 may play an important role in facilitating the AM-HBEC interaction via these soluble factors.
We postulated that adhesive interactions between CD11/CD18 on AM with ICAM-1 on HBEC contribute to the amplified production of cytokines from AM/HBEC co-cultures observed in the current study. Previous studies have demonstrated that cross-linking CD11b/CD18 on the surface of phagocytes using a combination of either its ligand ICAM-1 or anti-ICAM-1 antibodies primes phagocytes for increased respiratory burst and release of reactive oxygen intermediates [32-34]. In the present study we showed increased epithelial cell surface expression of ICAM-1 induced by PM10 exposure, while CD11b was constitutively expressed on the surface of AM. Cross-linking ICAM-1 on HBEC did not change their PM10-induced mRNA expression. Furthermore, blocking CD11b/CD18 and one of its ligands, ICAM-1 (CD54), did not block or decrease the PM10-induced mRNA expression in AM/HBEC co-cultures. These results are consistent with those of Tao and co-workers [35] who demonstrated that TNF-α and MIP-2 responses to urban air particles in rat AM and RLE (rat alveolar type II epithelial cell line) co-cultures were not blocked with anti-CD18 (β2-integrins)/CD54, arginine-glycine-aspartate peptide (against β1/β3-integrins) and heparin (non-specific anti-inflammatory agent). Paine and colleagues [36] demonstrated that blocking ICAM-1 (anti-CD54 F(ab')2 fragments) decreased rat AM phagocytosis of beads and their planar chemotaxis over the surface of rat alveolar type I epithelial cells. This suggests that ICAM-1 is important for the efficient phagocytosis of particles by AM and promotes mobility of AM on airway epithelial cell surface in the alveolus. Together these studies showed that the β2-integrin/ICAM-1 interaction between AM and lung epithelial cells are important in the chemotaxis of AM in the lung and their phagocytosis of inhaled particles but that this adhesive interaction may not contribute to the mediator production and release by AM and lung epithelial cells. These findings do not exclude the possibility that other adhesive interactions or simultaneous adhesive interactions of more that one adhesion molecule are involved in the particle-induced AM-bronchial epithelial cell mediator response.
Conclusion
Exposure of AM/HBEC co-cultures to ambient particles increased the expression and release of a variety of inflammatory mediators including GM-CSF, M-CSF, IL-6 and MIP-1β that enhance bone marrow production of monocytes and their recruitment into the lung. In addition this type of exposure resulted in synergistic production of GM-CSF and IL-6 in the co cultured cells. The adhesive interaction between ICAM-1 on epithelial cells with the β2-integrin CD11b on AM did not contribute to this synergistic mediator production. We speculate that the interaction between AM and lung epithelial cells amplifies PM10-induced lung inflammation and contributes to the pulmonary morbidity associated with exposure to particulate matter air pollution. This enhanced lung inflammation may also contribute to the systemic inflammatory response as well as the cardiovascular morbidity and mortality induced by air pollution [1,2,37].
Authors' contributions
HI carried out all through the experiments and drafted the manuscript. TF, YG, NS and HM participated in the design of the study. SH and JCH participated in its design and helped to draft the manuscript. SFVE conceived of the study, participated in its design and coordination and helped to draft the manuscript. RV provided EHC-93. All authors read and approved the final manuscript.
Acknowledgements
The authors thank Dr. W. Mark Elliott for technical support and Health Canada for providing the EHC-93. The work was supported by grants from the National Institutes of Health (HL407201), BC Lung Association and the Wolfe & Gita Churg Foundation. SF van Eeden is the recipient of a Career Investigators Award from the American Lung Association and the William Thurlbeck Distinguished Researcher Award.
==== Refs
Dockery DW Pope CA Xu X Spengler JD Ware JH Fay ME Ferris BG Speizer FE An association between air pollution and mortality in six U.S. cities N Engl J Med 1993 329 1 753 759 8350884 10.1056/NEJM199309093291102
Pope CA 3rd Epidemiology of fine particulate air pollution and human health: biologic mechanisms and who's at risk? Environ Health Perspect 2000 108 713 723 10931790
Schaumann F Borm PJ Herbrich A Knoch J Pitz M Schins RP Luettig B Hohlfeld JM Heinrich J Krug N Metal-rich Ambient Particles (Particulate Matter2.5) Cause Airway Inflammation in Healthy Subjects Am J Respir Crit Care Med 2004 170 898 903 15229099 10.1164/rccm.200403-423OC
van Eeden SF Tan WC Suwa T Mukae H Terashima T Fujii T Qui D Vincent R Hogg JC Cytokines involved in the systemic inflammatory response induced by exposure to particulate matter air pollutants (PM(10)) Am J Respir Crit Care Med 2001 164 826 830 11549540
Monn C Becker S Cytotoxicity and induction of proinflammatory cytokines from human monocytes exposed to fine (PM2.5) and coarse particles (PM10-2.5) in outdoor and indoor air Toxicol Appl Pharmacol 1999 155 245 252 10079210 10.1006/taap.1998.8591
Mukae H Hogg JC English D Vincent R van Eeden SF Phagocytosis of particulate air pollutants by human alveolar macrophages stimulates the bone marrow Am J Physiol Lung Cell Mol Physiol 2000 279 L924 931 11053029
Ishii H Fujii T Hogg JC Hayashi S Mukae H Vincent R van Eeden SF Contribution of IL-1 beta and TNF-alpha to the initiation of the peripheral lung response to atmospheric particulates (PM10) Am J Physiol Lung Cell Mol Physiol 2004 287 L176 183 15003925 10.1152/ajplung.00290.2003
Carter JD Ghio AJ Samet JM Devlin RB Cytokine production by human airway epithelial cells after exposure to an air pollution particle is metal-dependent Toxicol Appl Pharmacol 1997 146 180 188 9344885 10.1006/taap.1997.8254
Mills PR Davies RJ Devalia JL Airway epithelial cells, cytokines, and pollutants Am J Respir Crit Care Med 1999 160 S38 43 10556168
Fujii T Hayashi S Hogg JC Vincent R Van Eeden SF Particulate matter induces cytokine expression in human bronchial epithelial cells Am J Respir Cell Mol Biol 2001 25 265 271 11588002
Lang DS Schocker H Hockertz S Effects of crocidolite asbestos on human bronchoepithelial-dependent fibroblast stimulation in coculture: the role of IL-6 and GM-CSF Toxicology 2001 159 81 98 11250057 10.1016/S0300-483X(00)00415-7
Mukae H Vincent R Quinlan K English D Hards J Hogg JC van Eeden SF The effect of repeated exposure to particulate air pollution (PM10) on the bone marrow Am J Respir Crit Care Med 2001 163 201 209 11208647
Fujii T Hayashi S Hogg JC Mukae H Suwa T Goto Y Vincent R van Eeden SF Interaction of alveolar macrophages and airway epithelial cells following exposure to particulate matter produces mediators that stimulate the bone marrow Am J Respir Cell Mol Biol 2002 27 34 41 12091243
Goto Y Hogg JC Shih CH Ishii H Vincent R van Eeden SF Exposure to ambient particles accelerates monocyte release from bone marrow in atherosclerotic rabbits Am J Physiol Lung Cell Mol Physiol 2004 287 L79 85 15003928 10.1152/ajplung.00425.2003
Goto Y Ishii H Hogg JC Shih CH Yatera K Vincent R van Eeden SF Particulate Matter Air Pollution Stimulates Monocyte Release from the Bone Marrow Am J Respir Crit Care Med 2004 170 891 897 15256391 10.1164/rccm.200402-235OC
Goto Y Hogg JC Suwa T Quinlan KB van Eeden SF A novel method to quantify the turnover and release of monocytes from the bone marrow using the thymidine analog 5'-bromo-2'-deoxyuridine Am J Physiol Cell Physiol 2003 285 C253 259 12672652
Vincent R Bjarnason SG Adamson IY Hedgecock C Kumarathasan P Guenette J Potvin M Goegan P Bouthillier L Acute pulmonary toxicity of urban particulate matter and ozone Am J Pathol 1997 151 1563 1570 9403707
Koyama Y Tanaka Y Saito K Abe M Nakatsuka K Morimoto I Auron PE Eto S Cross-linking of intercellular adhesion molecule 1 (CD54) induces AP-1 activation and IL-1beta transcription J Immunol 1996 157 5097 5103 8943419
Wang Q Doerschuk CM Neutrophil-induced changes in the biomechanical properties of endothelial cells: roles of ICAM-1 and reactive oxygen species J Immunol 2000 164 6487 6494 10843706
Oppenheim J Human chemokines: an update Annu Rev Immunol 1998 15 675 705
Shanley TP Warner RL Ward PA The role of cytokines and adhesion molecules in the development of inflammatory injury Mol Med Today 1995 1 40 45 9415137 10.1016/1357-4310(95)80019-0
Wang JM Sherry B Fivash MJ Kelvin DJ Oppenheim JJ Human recombinant macrophage inflammatory protein-1 alpha and -beta and monocyte chemotactic and activating factor utilize common and unique receptors on human monocytes J Immunol 1993 150 3022 3029 8454871
Jansen JH Kluin-Nelemans JC Van Damme J Wientjens GJ Willemze R Fibbe WE Interleukin 6 is a permissive factor for monocytic colony formation by human hematopoietic progenitor cells J Exp Med 1992 175 1151 1154 1552286 10.1084/jem.175.4.1151
Moore BB Coffey MJ Christensen P Sitterding S Ngan R Wilke CA McDonald R Phare SM Peters-Golden M Paine R 3rdToews GB GM-CSF regulates bleomycin-induced pulmonary fibrosis via a prostaglandin-dependent mechanism J Immunol 2000 165 4032 4039 11034414
Paine R 3rdWilcoxen SE Morris SB Sartori C Baleeiro CE Matthay MA Christensen PJ Transgenic overexpression of granulocyte macrophage-colony stimulating factor in the lung prevents hyperoxic lung injury Am J Pathol 2003 163 2397 2406 14633611
Le J Vilcek J Tumor necrosis factor and interleukin 1: cytokines with multiple overlapping biological activities Lab Invest 1987 56 234 248 3029503
Dinarello CA Interleukin-1 and its biologically related cytokines Adv Immunol 1989 44 153 205 2466396
Keatings VM Collins PD Scott DM Barnes PJ Differences in interleukin-8 and tumor necrosis factor-alpha in induced sputum from patients with chronic obstructive pulmonary disease or asthma Am J Respir Crit Care Med 1996 153 530 534 8564092
Jiang Y Beller DI Frendl G Graves DT Monocyte chemoattractant protein-1 regulates adhesion molecule expression and cytokine production in human monocytes J Immunol 1992 148 2423 2428 1348518
Rosseau S Hammerl P Maus U Walmrath HD Schutte H Grimminger F Seeger W Lohmeyer J Phenotypic characterization of alveolar monocyte recruitment in acute respiratory distress syndrome Am J Physiol Lung Cell Mol Physiol 2000 279 L25 35 10893199
Lee YC Rannels DE Alveolar macrophages modulate the epithelial cell response to coal dust in vitro Am J Physiol 1996 270 L123 132 8772535
Liles WC Ledbetter JA Waltersdorph AW Klebanoff SJ Cross-linking of CD18 primes human neutrophils for activation of the respiratory burst in response to specific stimuli: implications for adhesion-dependent physiological responses in neutrophils J Leukoc Biol 1995 58 690 697 7499967
Rothlein R Kishimoto TK Mainolfi E Cross-linking of ICAM-1 induces co-signaling of an oxidative burst from mononuclear leukocytes J Immunol 1994 152 2488 2495 7510744
Shappell SB Toman C Anderson DC Taylor AA Entman ML Smith CW Mac-1 (CD11b/CD18) mediates adherence-dependent hydrogen peroxide production by human and canine neutrophils J Immunol 1990 144 2702 2711 2181020
Tao F Kobzik L Lung macrophage-epithelial cell interactions amplify particle-mediated cytokine release Am J Respir Cell Mol Biol 2002 26 499 505 11919087
Paine R 3rdMorris SB Jin H Baleeiro CE Wilcoxen SE ICAM-1 facilitates alveolar macrophage phagocytic activity through effects on migration over the AEC surface Am J Physiol Lung Cell Mol Physiol 2002 283 L180 187 12060575
Suwa T Hogg JC Quinlan KB Ohgami A Vincent R van Eeden SF Particulate air pollution induces progression of atherosclerosis J Am Coll Cardiol 2002 39 935 942 11897432 10.1016/S0735-1097(02)01715-1
|
16053532
|
PMC1199624
|
CC BY
|
2021-01-04 16:23:27
|
no
|
Respir Res. 2005 Aug 1; 6(1):87
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-87
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-911609113610.1186/1465-9921-6-91ResearchPPARα downregulates airway inflammation induced by lipopolysaccharide in the mouse Delayre-Orthez Carine [email protected] Julien [email protected] Isabelle [email protected] Vincent [email protected] Johan [email protected] Nelly [email protected] Françoise [email protected] EA 3771, Inflammation et environnement dans l'asthme, Faculté de Pharmacie, Université Louis Pasteur-Strasbourg I, Illkirch, France2 INSERM U620, Faculté des Sciences Pharmaceutiques, Université de Rennes 1, Rennes, France3 Institut de Génétique et de Biologie Moléculaire et Cellulaire, CNRS/Inserm/ULP, Illkirch, France2005 9 8 2005 6 1 91 91 26 1 2005 9 8 2005 Copyright © 2005 Delayre-Orthez et al; licensee BioMed Central Ltd.2005Delayre-Orthez et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Inflammation is a hallmark of acute lung injury and chronic airway diseases. In chronic airway diseases, it is associated with profound tissue remodeling. Peroxisome proliferator-activated receptor-α (PPARα) is a ligand-activated transcription factor, that belongs to the nuclear receptor family. Agonists for PPARα have been recently shown to reduce lipopolysaccharide (LPS)- and cytokine-induced secretion of matrix metalloproteinase-9 (MMP-9) in human monocytes and rat mesangial cells, suggesting that PPARα may play a beneficial role in inflammation and tissue remodeling.
Methods
We have investigated the role of PPARα in a mouse model of LPS-induced airway inflammation characterized by neutrophil and macrophage infiltration, by production of the chemoattractants, tumor necrosis factor-α (TNF-α), keratinocyte derived-chemokine (KC), macrophage inflammatory protein-2 (MIP-2) and monocyte chemoattractant protein-1 (MCP-1), and by increased MMP-2 and MMP-9 activity in bronchoalveolar lavage fluid (BALF). The role of PPARα in this model was studied using both PPARα-deficient mice and mice treated with the PPARα activator, fenofibrate.
Results
Upon intranasal exposure to LPS, PPARα-/- mice exhibited greater neutrophil and macrophage number in BALF, as well as increased levels of TNF-α, KC, MIP-2 and MCP-1, when compared to PPARα+/+ mice. PPARα-/- mice also displayed enhanced MMP-9 activity. Conversely, fenofibrate (0.15 to 15 mg/day) dose-dependently reduced the increase in neutrophil and macrophage number induced by LPS in wild-type mice. In animals treated with 15 mg/day fenofibrate, this effect was associated with a reduction in TNF-α, KC, MIP-2 and MCP-1 levels, as well as in MMP-2 and MMP-9 activity. PPARα-/- mice treated with 15 mg/day fenofibrate failed to exhibit decreased airway inflammatory cell infiltrate, demonstrating that PPARα mediates the anti-inflammatory effect of fenofibrate.
Conclusion
Using both genetic and pharmacological approaches, our data clearly show that PPARα downregulates cell infiltration, chemoattractant production and enhanced MMP activity triggered by LPS in mouse lung. This suggests that PPARα activation may have a beneficial effect in acute or chronic inflammatory airway disorders involving neutrophils and macrophages.
PPARαlipopolysaccharideinflammationneutrophilmacrophagematrix metalloproteinasemouse
==== Body
Background
Inflammation is a feature of both acute lung injury and chronic airway diseases. In chronic airway diseases such as chronic obstructive pulmonary disease (COPD), it is associated with profound tissue remodeling that contributes to impaired lung function [1]. Lipopolysaccharides (LPS), which are biological active components of the outer membrane of gram-negative bacteria, are important inducers of lung inflammation. Inflammatory response triggered by LPS is characterized by neutrophil and macrophage recruitment and by the release of chemoattractants including tumor necrosis factor-α (TNF-α), and the CXC and CC chemokines, interleukin-8 (IL-8) and monocyte chemoattractant protein-1 (MCP-1), respectively [2-5]. These inflammatory events reproduce some of the features of the inflammatory response observed during acute lung injury or COPD [1,6].
In mice, airway inflammation induced by LPS is associated with an increase of the matrix metalloproteinases (MMP), MMP-2 and MMP-9 [7,8]. MMP are a family of zinc- and calcium-dependent endopeptidases that play a major role in tissue remodeling [9,10]. Indeed, MMP degrade the majority of the extracellular matrix (ECM) proteins, including collagens, gelatins and proteoglycans, an activity which may contribute to lung injury by promoting infiltration accross basement membrane and activation of inflammatory cells [9,11]. Among MMP, MMP-2 (gelatinase A) preferentially produced by fibroblasts and other connective tissue cells, and MMP-9 (gelatinase B) mainly found in inflammatory cells, such as neutrophils and macrophages are of particular interest, since they cleave the major constituent of basement membrane, type IV collagen [9,10].
With the exception of neutrophils, normal tissues do not store MMP and constitutive expression is minimal. However, during inflammation and tissue remodeling, MMP expression is upregulated [9]. Levels or activity of several MMP have been found to be raised in animal models of acute lung injury (for review: [12]). Upregulation of MMP was also observed in chronic airway diseases associated with tissue remodeling, such as asthma and COPD (for review: [1,13]). Indeed, increased levels of MMP-9 have been reported in bronchoalveolar lavage fluid (BALF), blood or sputum from patients with asthma or COPD [14-17].
Peroxisome proliferator-activated receptor-α (PPARα) is a ligand-activated transcription factor, that belongs to the nuclear receptor family. PPARα regulates gene expression by binding as a heterodimeric complex with the retinoid X receptor to specific DNA sequences known as peroxisome proliferator response elements. PPARα was first identified for its role in the regulation of lipid and carbohydrate metabolism (for reviews: [18,19]). However, subsequent data have demonstrated that it exhibits also a potent anti-inflammatory activity. Indeed, mice deficient in PPARα (PPARα-/-) were reported to display an exacerbated reaction to various inflammatory stimuli, including LPS in the skin and the vessel [20-22]. Conversely, animals treated with PPARα activators such as fibrates exhibited a decreased response. Anti-inflammatory activity of fibrates appeared as unrelated to their lipid-lowering activity, since treatment with fenofibrate was shown to reduce inflammatory response associated with cerebral injury in absence of any improvement in plasma lipid levels in the mouse [23]. More recently, PPARα agonists were shown to reduce LPS- and cytokine-induced MMP-9 secretion in human monocytes and rat mesangial cells, suggesting that PPARα may also play a beneficial role in tissue remodeling [24,25].
We have here investigated the role of PPARα in a mouse model of LPS-induced airway inflammation characterized by cell infiltration, production of chemoattractants and increased MMP activity. This study was undertaken using both PPARα-deficient mice and mice treated with the PPARα activator, fenofibrate.
Materials and methods
Animals
Male wild-type (PPARα+/+) and homozygous knockout (PPARα-/-) mice (SV/129/C57BL/6) were expanded from breeding pairs [26] and used at the age of 9 weeks. Nine-week-old male C57BL/6 mice were purchased from Charles River Laboratories (Saint-Germain-sur-l'Arbresle, France). Animals were maintained under controlled environmental conditions with a 12 h/12 h light/dark cycle according to the EU guide for use of laboratory animals. Food (UAR-Alimentation, Villemoisson, France) and tap water were available ad libitum. Animal experimentation was conducted with the approval of the government body that regulates animal research in France.
LPS administration
LPS (Escherichia coli, serotype 055:B5, Sigma Chemical, Saint Quentin Fallavier, France) prepared in saline was administered by i.n. instillation for 4 consecutive days at the dose of 40 μg/kg. Control animals received saline instead of LPS. Instillations (12.5 μl per nostril) were carried out under anaesthesia (50 mg/kg ketamine and 3.33 mg/kg xylazine given i.p.).
Treatment with fenofibrate
Fenofibrate (Sigma Chemical) suspended in 1% carboxymethylcellulose (low viscosity, Sigma) in water was administered per os once daily for 10 days at increasing doses (0.15 to 15 mg/day), as previously described [27]. Duration of treatment was selected from a previous study showing protection against myocardial injury in mice [28]. Control animals received equivalent volumes (100 μl) of 1% carboxymethylcellulose (CMC) in similar conditions.
Collection of bronchoalveolar lavage fluids
Eighteen to twenty-four hours after the last LPS administration, mice were anaesthetized by i.p. injection of ketamine (150 mg/kg) and xylazine (10 mg/kg). A plastic cannula was inserted into the trachea and airways were lavaged by 10 instillations of 0.5 ml ice-cold saline supplemented with 2.6 mM EDTA (saline-EDTA). BALF recovered from the two first instillations were centrifuged (4100 rpm for 5 min at 4°C), and the resulting supernatant was stored at -20°C until MMP and cytokine measurements.
Determination of total and differential cell counts
BALF were centrifuged (1200 rpm for 5 min at 4°C) to pellet cells and erythrocytes were lysed by hypotonic shock. Cells were then resuspended in 500 μl ice-cold saline-EDTA and total cell counts were determined using a hemocytometer (Neubauer's chamber). Differential cell counts were assessed on cytologic preparations obtained by cytocentrifugation (Cytospin 3, Shandon Ltd, Runcorn, Chershire, UK) of 200 μl of diluted BALF (250 000 cells/ml in ice-cold saline-EDTA). Slides were stained with Hemacolor (Merck, Dormstadt, Germany) and determinations were performed by counting at least 400 cells for each preparation. Cells were identified as macrophages and neutrophils, and expressed as absolute numbers from total cell counts.
Determination of cytokine and chemokine levels
Tumor necrosis factor-α (TNF-α), keratinocyte derived-chemokine (KC), macrophage inflammatory protein-2 (MIP-2) and monocyte chemoattractant protein-1 (MCP-1) were quantified in BALF using capture ELISA kits according to instructions provided by the manufacturers (PharMingen for TNF-α and R&D Systems Europe (Lille, France) for KC, MIP-2 and MCP-1).
Gelatin zymography for determination of gelatinase activity
BALF samples were separated under non-reducing conditions by electrophoresis on a 7% acrylamide-separating gel containing 1 mg/ml gelatin and sodium dodecyl sulfate, as previously described [7]. After electrophoresis, gels were washed twice with 2.5% Triton X-100, rinsed with water and incubated overnight at 37°C in 50 mM Tris pH 8.0 containing 5 mM CaCl2 and 1 nM ZnCl2. Gels were stained with Coomassie Brilliant blue and destained in a 25% ethanol and 10% acetic acid solution. Gelatinase (MMP-2 and MMP-9) activities that appeared as clear bands against a blue background were quantified by measuring intensity of the bands by densitometry using the Densylab software (Bioprobe Systems, Les Ulis, France). Results were expressed as percentages of the intensity of a given sample loaded as internal standard onto each gel.
Histology
Lungs were perfused in situ, collected and immersed in 4% paraformaldehyde for 24 h at 4°C. Fixed lungs were rinsed in phosphate-buffered saline, dehydrated and embedded in paraffin using standard procedures. Five-micrometer tissue sections were stained with hematoxylin-eosin and observed under light microscopy.
Statistical analysis
Data are presented as means ± SEM. Statistical differences were analyzed from raw data by analysis of variance followed by unpaired two-tailed Student's t-test with a Bonferroni correction.
Results
Increased cell infiltration, chemoattractant production and MMP activity in PPARα-/- mice upon exposure to LPS
Saline-exposed PPARα-/- mice exhibited no differences in total cell and macrophage count in BALF when compared to saline-exposed PPARα+/+ animals (Figure 1). Upon exposure to LPS, both PPARα+/+ and PPARα-/- mice displayed a significant increase in total cell, neutrophil and macrophage number, when compared to animals exposed to saline (Figure 1). However, these increases were 2.9- (p < 0.0001), 5.0- (p < 0.0001) and 1.9-fold (p < 0.0001) greater, respectively in PPARα-/- mice than in PPARα+/+ mice (Figure 1).
Figure 1 Number of total cells, neutrophils and macrophages in BALF from PPARα+/+(+/+) and PPARα-/- (-/-) mice exposed to LPS or saline. Data are mean ± SEM of n = 10–13 animals. Statistically significant differences at α = 0.05: (*) when compared to PPARα+/+ mice treated with saline; (#) when compared to PPARα-/- mice treated with saline; and ($) when compared to PPARα+/+ mice treated with LPS.
Cell infiltration induced by LPS was associated with a significant increase in BALF levels of the chemoattractants, TNF-α, KC and MCP-1 in both PPARα+/+ and PPARα-/- mice (Figure 2). These levels were however 1.5- (p = 0.0003), 2.3- (p = 0.0008) and 3.5-fold (p = 0.0012) greater, respectively in PPARα-/- animals when compared to PPARα+/+ mice (Figure 2). PPARα-/- mice exposed to LPS also displayed a significant rise in MIP-2 in BALF (2.0-fold, p = 0.0065), whereas LPS-treated PPARα+/+ animals exhibited no changes in this chemokine.
Figure 2 Chemoattractant levels in BALF from PPARα+/+ (+/+) and PPARα-/- (-/-) mice exposed to LPS or saline. Data are mean ± SEM of n = 9–12 animals. Statistically significant differences at α = 0.05: (*) when compared to PPARα+/+ mice treated with saline; (#) when compared to PPARα-/- mice treated with saline; and ($) when compared to PPARα+/+ mice treated with LPS.
Saline-exposed PPARα-/- mice exhibited similar low MMP-2 (76 kDa) and MMP-9 (105 kDa) activity in BALF when compared to saline-exposed PPARα+/+ animals (Figure 3). Upon exposure to LPS, PPARα+/+ and PPARα-/- mice displayed a significant increase in both MMP-2 and MMP-9 activity, when compared to animals exposed to saline (Figure 3). MMP-2 levels were similar in LPS-treated PPARα-/- and PPARα+/+ mice (61 ± 8 vs 58 ± 4). In contrast, MMP-9 levels were 1.8-fold (p < 0.0001) greater in PPARα-/- animals than in PPARα+/+ mice.
Figure 3 MMP-2 (76 kDa) and MMP-9 (105 kDa) activity in BALF from PPARα+/+ (+/+) and PPARα-/- (-/-) mice exposed to LPS or saline. Upper panel shows gelatin zymogram from two representative animals in each group. Lower panel shows data of all animals in each group (n = 10–13) expressed as mean ± SEM. Statistically significant differences at α = 0.05: (*) when compared to PPARα+/+ mice treated with saline; (#) when compared to PPARα-/- mice treated with saline; and ($) when compared to PPARα+/+ mice treated with LPS.
Reduced cell infiltration, chemoattractant production and MMP activity in wild-type mice upon PPARα activation by fenofibrate
Exposure to LPS resulted in marked increases in total cell, neutrophil and macrophage number in BALF from C57BL/6 mice (Figure 4). These increases were dose-dependently reduced by fenofibrate (0.15 to 15 mg/day). Reduction in total cell, neutrophil and macrophage number reached 80% (p < 0.0001), 91% (p < 0.0001) and 64% (p < 0.0001), respectively in BALF from mice treated with 15 mg/kg of the PPARα activator when compared to mice treated with the vehicle, CMC (Figure 4). Fenofibrate (15 mg/day) inhibited also total cell (p = 0.0055), neutrophil (p < 0.0001) and macrophage (p = 0.0064) infiltrate induced by LPS in PPARα+/+ mice (Table 1). In contrast, LPS-exposed PPARα-/- mice treated with 15 mg/day fenofibrate failed to exhibit changes in inflammatory cell infiltrate, demonstrating that PPARα mediates the anti-inflammatory activity of fenofibrate (Table 1).
Table 1 Cell infiltration in LPS-exposed PPARα+/+ and PPARα-/- mice treated with fenofibrate.
Group Number of cells (×106)
Total Neutrophils Macrophages
(+/+)-LPS-CMC 1.93 ± 0.11 0.95 ± 0.11 0.98 ± 0.13
(+/+)-LPS-FF 0.73 ± 0.08 (*) 0.23 ± 0.07 (*) 0.50 ± 0.05 (*)
(-/-)-LPS-CMC 3.46 ± 0.38 (*) 1.85 ± 0.28 (*) 1.60 ± 0.25 (*)
(-/-)-LPS-FF 3.13 ± 0.54 (n.s.) 1.73 ± 0.30 (n.s.) 1.39 ± 0.28 (n.s.)
Data are mean ± SEM of n = 6–8 animals. (*): statistically significant differences at α = 0.05 when compared to PPARα+/+ mice treated with CMC. (n.s.): non statistically different when compared to PPARα-/- mice treated with CMC.
Figure 4 Dose-dependent reduction of cell infiltration in wild-type mice exposed to LPS upon PPARα activation by fenofibrate. Number of total cells, neutrophils and macrophages in BALF from C57BL/6 mice exposed to LPS and treated with increasing doses of fenofibrate (0.15 to 15 mg/day) or its vehicle (1% CMC), when compared to mice exposed to saline and treated with CMC. Data are mean ± SEM of n = 6 animals. Statistically significant differences at α = 0.05: (*) when compared to mice exposed to saline and treated with CMC; (#) when compared to mice exposed to LPS and treated with CMC.
Histological examination of lung tissue confirmed the anti-inflammatory effect of fenofibrate. Indeed, whereas a massive inflammatory cell infiltration was observed in perivascular and alveolar tissue of C57BL/6 mice exposed to LPS and treated with CMC when compared to mice exposed to saline (Figure 5A et 5B), a marked reduction in cell infiltration was observed on lung sections from mice exposed to LPS and treated with fenofibrate (Figure 5C).
Figure 5 Histological analysis of lung tissue from wild-type mice. Lung sections showing a massive inflammatory cell infiltrate in perivascular and alveolar tissue of C57BL/6 mice exposed to LPS and treated with CMC (B), when compared to mice exposed to saline (A). Reduced cell infiltrate in lung tissue from mice exposed to LPS and treated with fenofibrate (C).
C57BL/6 mice exposed to LPS and treated with CMC displayed also increases in TNF-α, KC, MIP-2 and MCP-1 in BALF when compared to saline-exposed mice (Figure 6A). Treatment with fenofibrate (15 mg/day) inhibited these increases by 59% (p < 0.0001), 50% (p = 0.0015), 30% (p = 0.0058) and 69% (p < 0.0001), respectively (Figure 6A).
Figure 6 Reduced chemoattractant production and MMP activity in wild-type mice upon PPARα activation by fenofibrate. Chemoattractant levels (A) and MMP-2 and MMP-9 activity (B) in BALF from C57BL/6 mice exposed to LPS and treated with fenofibrate (15 mg/day, black bars) or its vehicle (1% CMC, grey bars), when compared to mice exposed to saline and treated with CMC (open bars). Data are mean ± SEM of n = 7–8 animals. Statistically significant differences at α = 0.05: (*) when compared to mice exposed to saline and treated with CMC; (#) when compared to mice exposed to LPS and treated with CMC.
Treatment with fenofibrate (15 mg/day) also dramatically reduced LPS-induced increase in MMP-2 and MMP-9 activity (Figure 6B). Indeed, whereas MMP-2 and MMP-9 activity was increased by 1.8- (p < 0.0001) and 3.6-fold (p < 0.0001), respectively in BALF from LPS-exposed mice treated with CMC when compared to saline-exposed mice, animals exposed to LPS and treated with fenofibrate displayed MMP levels similar to those measured in saline-exposed animals.
Discussion
In this study, we have addressed the role of PPARα in a mouse model of LPS-induced airway inflammation. Using both genetic and pharmacological approaches, our data clearly showed that PPARα downregulates cell infiltration, chemoattractant production and enhanced MMP activity triggered by LPS in mouse lung.
As expected, wild-type mice exposed to LPS exhibited a massive recruitment of inflammatory cells in the airways, composed of neutrophils and macrophages. This cell infiltration was associated with an increase in BALF levels of the pro-inflammatory and chemoattractant cytokine, TNF-α and by a rise in the levels of the CXC chemokines, MIP-2 and KC and of the CC chemokine, MCP-1. Exposure to LPS also induced a marked increase in MMP-2 and MMP-9 activity in BALF, when compared to saline exposure. This model reproduced several features of the inflammatory response observed during acute lung injury or COPD [1,6,13]. Using this model, we found that PPARα-/- mice exposed to LPS displayed enhanced neutrophil and macrophage number in BALF when compared to PPARα+/+ animals, whereas wild-type mice treated with the PPARα activator, fenofibrate exhibited reduced cell infiltrate. Furthermore, we demonstrated fenofibrate selectivity by showing absence of effect of fenofibrate in PPARα-/- animals. Taken together, these results suggest that PPARα activation may have a beneficial effect in airway inflammatory diseases involving neutrophil and monocyte recruitment. In agreement with our results, Birrell et al. recently proposed that agonists of another PPAR receptor, PPARγ may have a therapeutic potential in respiratory diseases involving neutrophilia [29]. Our study adds to these previous findings by showing that PPARα agonists may also be effective in blocking recruitment of monocytes, which play a pivotal role in the pathophysiology of COPD, as well as of pulmonary fibrosis. By contrast, Trifilieff et al. found that PPARα ligands failed to inhibit neutrophil recruitment induced by LPS in BALF from mice [30]. Differences in the mode of exposure to LPS could explain this discrepancy. Indeed, whereas these authors exposed female mice intranasally to a single high dose of LPS (0.3 mg/kg) for a short period of time (3 h), the present study was carried out in male animals using four repeated instillations of a 7.5-fold lower dose of LPS (40 μg/kg). Indeed, these modes of exposure may trigger different inflammatory responses. Likewise, nature (GW 9578 vs fenofibrate) and route of delivery (local vs oral) of PPARα agonists may be another source of discrepancy. Therefore, by both genetic and pharmacological approaches, our data clearly demonstrate that PPARα downregulates neutrophil and monocyte infiltration in mouse lung.
We also found that PPARα-/- mice exposed to LPS displayed increased levels of TNF-α in BALF when compared to PPARα+/+ animals, whereas wild-type mice treated with fenofibrate exhibited reduced TNF-α levels. As a pro-inflammatory cytokine, TNF-α that is released by macrophages or airway epithelial cells upon activation plays an important role in neutrophilic inflammation induced by LPS [4]. Indeed, TNF-α triggers the release of CXC chemokines, such as MIP-2 and KC that are involved in LPS-induced intrapulmonary recruitment of neutrophils [2,3]. As well, MCP-1, which plays a central role in monocyte recruitment to inflamed tissues, is produced by pulmonary macrophages and airway epithelial cells in response to TNF-α or LPS [31,32]. In the present study, release of MIP-2, KC and MCP-1 triggered by LPS instillation was greater in BALF from PPARα-/- mice when compared to PPARα+/+ animals. Conversely, wild-type mice treated with fenofibrate displayed decreased levels of these chemokines when compared to vehicle-treated animals. Taken together, our results suggest that downregulation of TNF-α and of the CXC and C-C chemokines, MIP-2, KC and MCP-1 contributes to PPARα-induced inhibition of neutrophil and macrophage airway recruitment in our model.
PPARα agonists were recently reported to reduce LPS- and IL-1β-induced secretion of MMP-9 in human monocytes and rat mesangial cells, respectively [24,25]. However, the effect of PPARα on MMP production in vivo is so far unknown. In the present study, we demonstrate that PPARα downregulates increase in MMP-2 and MMP-9 activity triggered by LPS in mouse lung. Indeed, whereas PPARα-/- mice displayed a greater increase in MMP activity in BALF upon exposure to LPS when compared to PPARα+/+ animals, wild-type mice exposed to LPS exhibited decreased levels of MMP when treated by fenofibrate. Sources of MMP in the lung are numerous, particularly under inflammatory conditions. Among them, neutrophils and macrophages are considered as the major sources of MMP-9 [11]. Therefore, downregulation of MMP-9 production by PPARα may result from decreased cell infiltration. In neutrophils, MMP-9 is stored in specific granules from which it is readily released, in particular upon stimulation by LPS or chemoattractant factors, like IL-8 [33]. Downregulation of MMP-9 production by PPARα could alternatively result from decreased neutrophil activation. MMP-9 is believed to play a major role in lung remodeling. Indeed, in addition to digestion of extracellular matrix proteins, MMP-9 increases the activity of other proteases, as well as of chemoattractants and growth factors (for review: [34]). By providing evidence that PPARα downregulates MMP activity in vivo, our study reinforces the idea that the nuclear receptor PPARα may play a beneficial role in tissue remodeling.
Several studies have reported that PPARα inhibits the NF-κB pathway, which plays a critical role in LPS signaling as well as in the expression of the chemokines, MIP-2, KC and MCP-1 and of MMP-9 [35]. This property could account for the beneficial effect of PPARα observed in the present study. However, several other mechanisms could be involved. This includes production of anti-inflammatory mediators, such as IL-10. Indeed, fenofibrate was reported to suppress autoimmune myocarditis in rats by stimulating expression of this cytokine [36]. As well, inhibition of cell recruitment could be implicated. Thus, activation of PPARα was reported to inhibit chemotaxis of inflammatory cells, including macrophages [37,38]. Finally, resolution of inflammation through stimulation of inflammatory cell apoptosis may also be involved, since activation of PPARα was shown to induce apoptosis of macrophages [39].
Conclusion
In conclusion, using both genetic and pharmacological approaches, our study provides evidence that PPARα downregulates neutrophil and monocyte infiltration induced by LPS in mouse lung. Our data also demonstrated that this beneficial effect of PPARα involves downregulation of the production of neutrophil and monocyte chemoattractants, including the CXC and C-C chemokines, MIP-2, KC and MCP-1, and of MMP that play a major role in tissue remodeling. We postulate that PPARα agonists, and in particular fenofibrate may have a therapeutic potential in airway inflammatory disorders involving neutrophil and monocyte, such as acute lung injury and COPD.
List of abbreviations
BALF: bronchoalveolar lavage fluid
CMC: carboxylmethylcellulose
COPD: chronic obstructive pulmonary disease
EDTA: ethylenediaminetetraacetic acid
IL: interleukin
KC: keratinocyte derived-chemokine
LPS: lipopolysaccharide
MIP-2: macrophage inflammatory protein-2
MMP: matrix metalloproteinase
PPAR: peroxisome proliferator-actived receptor
MCP-1: monocyte chemoattractant protein-1
TNF-α: tumor necrosis factor-α
Authors' contributions
CDO, JB and IG have made substantial contributions to acquisition and analysis of data.
CDO, VL and FP have made substantial contributions to conception and design of the study.
CDO and FP have been involved in drafting the article.
JA, NF and VL have been involved in revising the article critically for important intellectual content.
Acknowledgements
This work was supported by the Institut National de la Santé et de la Recherche Médicale, Université Louis Pasteur and Fonds de Recherche GIP Aventis. Carine Delayre-Orthez was supported by a joint PhD grant from ADEME and Région Alsace, and by the Société de Pneumologie de Langue Française. The PPARα-/- mice used in this study were a kind gift of Dr F.Gonzalez at the NHCI in Bethesda.
==== Refs
Barnes PJ Shapiro SD Pauwels RA Chronic obstructive pulmonary disease: molecular and cellular mechanisms Eur Respir J 2003 22 672 688 14582923
Frevert CW Huang S Danaee H Paulauskis JD Kobzik L Functional characterization of the rat chemokine KC and its importance in neutrophil recruitment in a rat model of pulmonary inflammation J Immunol 1995 154 335 344 7995953
Schmal H Shanley TP Jones ML Friedl HP Ward PA Role for macrophage inflammatory protein-2 in lipopolysaccharide-induced lung injury in rats J Immunol 1996 156 1963 1972 8596051
Skerrett SJ Martin TR Chi EY Peschon JJ Mohler KM Wilson CB Role of the type 1 TNF receptor in lung inflammation after inhalation of endotoxin or Pseudomonas aeruginosa Am J Physiol 1999 276 L715 27 10330027
Kopydlowski KM Salkowski CA Cody MJ van Rooijen N Major J Hamilton TA Vogel SN Regulation of macrophage chemokine expression by lipopolysaccharide in vitro and in vivo J Immunol 1999 163 1537 1544 10415057
Goodman RB Pugin J Lee JS Matthay MA Cytokine-mediated inflammation in acute lung injury Cytokine Growth Factor Rev 2003 14 523 535 14563354 10.1016/S1359-6101(03)00059-5
Corbel M Theret N Caulet-Maugendre S Germain N Lagente V Clement B Boichot E Repeated endotoxin exposure induces interstitial fibrosis associated with enhanced gelatinase (MMP-2 and MMP-9) activity Inflamm Res 2001 50 129 135 11339500 10.1007/s000110050736
Corbel M Germain N Lanchou J Molet S PM RS Martins MA Boichot E Lagente V The selective phosphodiesterase 4 inhibitor RP 73-401 reduced matrix metalloproteinase 9 activity and transforming growth factor-beta release during acute lung injury in mice: the role of the balance between Tumor necrosis factor-alpha and interleukin-10 J Pharmacol Exp Ther 2002 301 258 265 11907182 10.1124/jpet.301.1.258
Nagase H Woessner JFJ Matrix metalloproteinases J Biol Chem 1999 274 21491 21494 10419448 10.1074/jbc.274.31.21491
Murphy G Docherty AJ The matrix metalloproteinases and their inhibitors Am J Respir Cell Mol Biol 1992 7 120 125 1497900
Shapiro SD Senior RM Matrix metalloproteinases. Matrix degradation and more Am J Respir Cell Mol Biol 1999 20 1100 1102 10340927
Corbel M Lagente V Boichot E Pulmonary inflammation and tissue remodelling : role of metalloproteinases Eur Respir Rev 2000 10 260 263
Belvisi MG Bottomley KM The role of matrix metalloproteinases (MMPs) in the pathophysiology of chronic obstructive pulmonary disease (COPD): a therapeutic role for inhibitors of MMPs? Inflamm Res 2003 52 95 100 12755372 10.1007/s000110300020
Belleguic C Corbel M Germain N Lena H Boichot E Delaval PH Lagente V Increased release of matrix metalloproteinase-9 in the plasma of acute severe asthmatic patients Clin Exp Allergy 2002 32 217 223 11929485 10.1046/j.1365-2222.2002.01219.x
Mautino G Oliver N Chanez P Bousquet J Capony F Increased release of matrix metalloproteinase-9 in bronchoalveolar lavage fluid and by alveolar macrophages of asthmatics Am J Respir Cell Mol Biol 1997 17 583 591 9374109
Segura-Valdez L Pardo A Gaxiola M Uhal BD Becerril C Selman M Upregulation of gelatinases A and B, collagenases 1 and 2, and increased parenchymal cell death in COPD Chest 2000 117 684 694 10712992 10.1378/chest.117.3.684
Vignola AM Riccobono L Mirabella A Profita M Chanez P Bellia V Mautino G D'Accardi P Bousquet J Bonsignore G Sputum metalloproteinase-9/tissue inhibitor of metalloproteinase-1 ratio correlates with airflow obstruction in asthma and chronic bronchitis Am J Respir Crit Care Med 1998 158 1945 1950 9847290
Desvergne B Wahli W Peroxisome proliferator-activated receptors: nuclear control of metabolism Endocr Rev 1999 20 649 688 10529898 10.1210/er.20.5.649
Francis GA Fayard E Picard F Auwerx J Nuclear receptors and the control of metabolism Annu Rev Physiol 2003 65 261 311. Epub 2002 May 1. 12518001 10.1146/annurev.physiol.65.092101.142528
Devchand PR Keller H Peters JM Vazquez M Gonzalez FJ Wahli W The PPARalpha-leukotriene B4 pathway to inflammation control Nature 1996 384 39 43 8900274 10.1038/384039a0
Delerive P De Bosscher K Besnard S Vanden Berghe W Peters JM Gonzalez FJ Fruchart JC Tedgui A Haegeman G Staels B Peroxisome proliferator-activated receptor alpha negatively regulates the vascular inflammatory gene response by negative cross-talk with transcription factors NF-kappaB and AP-1 J Biol Chem 1999 274 32048 32054 10542237 10.1074/jbc.274.45.32048
Sheu MY Fowler AJ Kao J Schmuth M Schoonjans K Auwerx J Fluhr JW Man MQ Elias PM Feingold KR Topical peroxisome proliferator activated receptor-alpha activators reduce inflammation in irritant and allergic contact dermatitis models J Invest Dermatol 2002 118 94 101 11851881 10.1046/j.0022-202x.2001.01626.x
Deplanque D Gele P Petrault O Six I Furman C Bouly M Nion S Dupuis B Leys D Fruchart JC Cecchelli R Staels B Duriez P Bordet R Peroxisome proliferator-activated receptor-alpha activation as a mechanism of preventive neuroprotection induced by chronic fenofibrate treatment J Neurosci 2003 23 6264 6271 12867511
Shu H Wong B Zhou G Li Y Berger J Woods JW Wright SD Cai TQ Activation of PPARalpha or gamma reduces secretion of matrix metalloproteinase 9 but not interleukin 8 from human monocytic THP-1 cells Biochem Biophys Res Commun 2000 267 345 349 10623622 10.1006/bbrc.1999.1968
Eberhardt W Akool el S Rebhan J Frank S Beck KF Franzen R Hamada FM Pfeilschifter J Inhibition of cytokine-induced matrix metalloproteinase 9 expression by peroxisome proliferator-activated receptor alpha agonists is indirect and due to a NO-mediated reduction of mRNA stability J Biol Chem 2002 277 33518 33528 12093797 10.1074/jbc.M202008200
Lee SS Pineau T Drago J Lee EJ Owens JW Kroetz DL Fernandez-Salguero PM Westphal H Gonzalez FJ Targeted disruption of the alpha isoform of the peroxisome proliferator- activated receptor gene in mice results in abolishment of the pleiotropic effects of peroxisome proliferators Mol Cell Biol 1995 15 3012 3022 7539101
Schoonjans K Peinado-Onsurbe J Lefebvre AM Heyman RA Briggs M Deeb S Staels B Auwerx J PPARalpha and PPARgamma activators direct a distinct tissue-specific transcriptional response via a PPRE in the lipoprotein lipase gene Embo J 1996 15 5336 5348 8895578
Tabernero A Schoonjans K Jesel L Carpusca I Auwerx J Andriantsitohaina R Activation of the peroxisome proliferator-activated receptor alpha protects against myocardial ischaemic injury and improves endothelial vasodilatation BMC Pharmacol 2002 2 10 11940253 10.1186/1471-2210-2-10
Birrell MA Patel HJ McCluskie K Wong S Leonard T Yacoub MH Belvisi MG PPAR-gamma agonists as therapy for diseases involving airway neutrophilia Eur Respir J 2004 24 18 23 15293600 10.1183/09031936.04.00098303
Trifilieff A Bench A Hanley M Bayley D Campbell E Whittaker P PPAR-alpha and -gamma but not -delta agonists inhibit airway inflammation in a murine model of asthma: in vitro evidence for an NF- kappaB-independent effect Br J Pharmacol 2003 139 163 171 12746235 10.1038/sj.bjp.0705232
Brieland JK Flory CM Jones ML Miller GR Remick DG Warren JS Fantone JC Regulation of monocyte chemoattractant protein-1 gene expression and secretion in rat pulmonary alveolar macrophages by lipopolysaccharide, tumor necrosis factor-alpha, and interleukin-1 beta Am J Respir Cell Mol Biol 1995 12 104 109 7811465
Standiford TJ Kunkel SL Phan SH Rollins BJ Strieter RM Alveolar macrophage-derived cytokines induce monocyte chemoattractant protein-1 expression from human pulmonary type II-like epithelial cells J Biol Chem 1991 266 9912 9918 2033076
Masure S Proost P Van Damme J Opdenakker G Purification and identification of 91-kDa neutrophil gelatinase. Release by the activating peptide interleukin-8 Eur J Biochem 1991 198 391 398 1645657 10.1111/j.1432-1033.1991.tb16027.x
Atkinson JJ Senior RM Matrix metalloproteinase-9 in lung remodeling Am J Respir Cell Mol Biol 2003 28 12 24 12495928 10.1165/rcmb.2002-0166TR
Delerive P Fruchart JC Staels B Peroxisome proliferator-activated receptors in inflammation control J Endocrinol 2001 169 453 459 11375115 10.1677/joe.0.1690453
Maruyama S Kato K Kodama M Hirono S Fuse K Nakagawa O Nakazawa M Miida T Yamamoto T Watanabe K Aizawa Y Fenofibrate, a peroxisome proliferator-activated receptor alpha activator, suppresses experimental autoimmune myocarditis by stimulating the interleukin-10 pathway in rats J Atheroscler Thromb 2002 9 87 92 12236317
Kintscher U Goetze S Wakino S Kim S Nagpal S Chandraratna RA Graf K Fleck E Hsueh WA Law RE Peroxisome proliferator-activated receptor and retinoid X receptor ligands inhibit monocyte chemotactic protein-1-directed migration of monocytes Eur J Pharmacol 2000 401 259 270 10936484 10.1016/S0014-2999(00)00461-1
Woerly G Honda K Loyens M Papin JP Auwerx J Staels B Capron M Dombrowicz D Peroxisome proliferator-activated receptors alpha and gamma down-regulate allergic inflammation and eosinophil activation J Exp Med 2003 198 411 421 12900517 10.1084/jem.20021384
Chinetti G Griglio S Antonucci M Torra IP Delerive P Majd Z Fruchart JC Chapman J Najib J Staels B Activation of proliferator-activated receptors alpha and gamma induces apoptosis of human monocyte-derived macrophages J Biol Chem 1998 273 25573 25580 9748221 10.1074/jbc.273.40.25573
|
16091136
|
PMC1199625
|
CC BY
|
2021-01-04 16:23:24
|
no
|
Respir Res. 2005 Aug 9; 6(1):91
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-91
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-931609552910.1186/1465-9921-6-93ResearchA role for MCP-1/CCR2 in interstitial lung disease in children Hartl Dominik [email protected] Matthias [email protected] Thomas [email protected] Gernot [email protected] Christine [email protected] Dietrich [email protected] Dolores J [email protected] Susanne [email protected] Childrens' Hospital of the Ludwig-Maximilians-University, Munich, Germany2 Department of Pneumology, Medical Center, Albert-Ludwigs-University, Freiburg, Germany3 Institute of Molecular Immunology and Immune Monitoring Platform, GSF National Research Center for Environment and Health, Munich, Germany2005 11 8 2005 6 1 93 93 19 4 2005 11 8 2005 Copyright © 2005 Hartl et al; licensee BioMed Central Ltd.2005Hartl et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Interstitial lung diseases (ILD) are chronic inflammatory disorders leading to pulmonary fibrosis. Monocyte chemotactic protein 1 (MCP-1) promotes collagen synthesis and deletion of the MCP-1 receptor CCR2 protects from pulmonary fibrosis in ILD mouse models. We hypothesized that pulmonary MCP-1 and CCR2+ T cells accumulate in pediatric ILD and are related to disease severity.
Methods
Bronchoalveolar lavage fluid was obtained from 25 children with ILD and 10 healthy children. Levels of pulmonary MCP-1 and Th1/Th2-associated cytokines were quantified at the protein and the mRNA levels. Pulmonary CCR2+, CCR4+, CCR3+, CCR5+ and CXCR3+ T cells were quantified by flow-cytometry.
Results
CCR2+ T cells and MCP-1 levels were significantly elevated in children with ILD and correlated with forced vital capacity, total lung capacity and ILD disease severity scores. Children with lung fibrosis had significantly higher MCP-1 levels and CCR2+ T cells in bronchoalveolar lavage fluid compared to non-fibrotic children.
Conclusion
The results indicate that pulmonary CCR2+ T cells and MCP-1 contribute to the pathogenesis of pediatric ILD and might provide a novel target for therapeutic strategies.
ChemokinesMCP-1CCR2Bronchoalveolar LavageChildrenInterstitial Lung Diseases
==== Body
Background
Interstitial lung diseases (ILD) are chronic inflammatory disorders characterized by restrictive lung disease and diffuse pulmonary infiltrates. Although the precise incidence is not known, ILD are less frequent in children than adults [1-3]. Lungs of ILD patients show inflammation with alveolar wall thickening by leukocytes and pulmonary fibrosis. Despite immunosuppressive treatment and supportive measures, the progressive course leading to irreversible lung fibrosis sometimes can not be prevented. Therefore, the development of additional therapeutic strategies is of high importance.
Monocyte chemotactic protein 1 (MCP-1, CCL2) is produced in response to inflammatory stimuli by a variety of cells, including monocytes/macrophages, lymphocytes and airway epithelial cells [4-6]. MCP-1 stimulates collagen synthesis and production of the pro-fibrotic factor transforming growth factor β (TGF-β) in fibroblasts, while MCP-1 antisense oligonucleotides reduce TGF-β production[7,8]. Application of MCP-1 into murine lungs induces an inflammatory cytokine response and pulmonary leukocyte accumulation. In adult patients with ILD, increased levels of MCP-1 were observed in serum[9,10] and bronchoalveolar lavage fluid (BALF) [11-14]. Although MCP-1 was originally described for its chemotactic activity on monocytes, in vitro studies revealed an even higher activity on T cells[15]. This occurs through MCP-1 binding to its sole receptor CCR2[16]. Deletion of the CCR2-gene or receptor blockade with anti-CCR2 antibodies leads to a dramatic inhibition of leukocyte accumulation in murine lungs[17]. Furthermore, CCR2-/- mice are protected from fluorescein (FITC) or bleomycin induced lung fibrosis[18]. Thus far, CCR2+ T cells in BALF of patients with fibrotic lung diseases have not been determined.
In addition to the MCP-1/CCR2 axis, Th2 cytokines seem to mediate pulmonary fibrosis [19-22]. IL-4 stimulates fibroblast proliferation and collagen synthesis[23,24], while IFN-γ inhibits this process [25-28]. In a Th2 mouse model fibroblasts expressed more CCR2 protein and higher levels of MCP-1 and TGF-β as compared to fibroblasts from a Th1-mouse model[8]. Furthermore, increased levels of IL-4 were observed in animal models of pulmonary fibrosis[29] and lungs of patients with idiopathic pulmonary fibrosis (IPF)[30] or cryptogenic fibrosing alveolitis[31].
The contribution of MCP-1 to ILD has been investigated exclusively in adults. However, the spectrum of ILD differs considerably between adults and children and some forms are unique to children while others, such as idiopathic pulmonary fibrosis (IPF), are extremely rare in childhood[32].
Therefore, we asked whether levels of MCP-1 and frequencies of CCR2+ T cells are increased in BALF of children with ILD and, if so, how levels of MCP-1 and CCR2+ T cells relate to disease severity in pediatric ILD.
To address these questions levels of MCP-1 and frequencies of CCR2+ T cells in BALF were compared between children with ILD and children without lung disease.
To evaluate the contribution of the pulmonary Th1/Th2 micromilieu to the pathogenesis of pediatric ILD, CCR4+ and CCR3+ (Th2) and CCR5+ and CXCR3+ (Th1) cells were determined in BALF together with an array of pulmonary Th1- and Th2-associated cytokines.
Our results indicate that pulmonary CCR2+ T cells and levels of MCP-1 are characteristic components in BALF of children with ILD. A pathophysiological role in pediatric ILD seems likely as their levels relate to restrictive lung function and ILD disease severity.
Methods
Characterization of the patients
Children attending the Department of Pulmonology and Allergology of the University Children's Hospital of Munich during 1999–2004 were considered for inclusion in this study. Children suspective of ILD underwent a comprehensive clinical evaluation, including patient history, physical examination, routine laboratory tests, lung function testing, chest radiography, high resolution computed tomography (HRCT) and bronchoalveolar lavage (BAL). Children were assigned to the ILD group according to the criteria of Fan[33]: (i) ≥3 months of respiratory symptoms characteristic for ILD, i.e. non-productive cough, dyspnoea, tachypnea, crackles and/or rales, exercise intolerance and/or hypoxemia, (ii) diffuse infiltrates on chest radiographs and HRCT and (iii) restrictive lung function (decreased forced vital capacity (FVC) and total lung capacity (TLC)) according to the ATS criteria[34].
The diagnosis of the specific form of ILD was established by patient history, physical examination, HRCT, BAL and/or lung biopsy according to consensus criteria[33,35]. Two thoracic radiologists independently evaluated all lobes on HRCT for ground glass opacity and pulmonary fibrosis as described previously[36,37]. A pathologist specialized on pediatric ILD[38] evaluated the lung sections systematically[39,40]. Furthermore, the disease severity of each ILD patient was characterized using the clinical ILD score of Fan[41]: 1 = asymptomatic, no desaturation; 2 = symptomatic but normoxic (>90%) under all conditions; 3 = symptomatic with desaturation during sleep or with exercise; 4 = symptomatic with desaturation at rest. None of the included children had familial idiopathic pulmonary fibrosis. Patients with congenital heart disease or suspected or proven bacterial pulmonary infection were excluded from the study.
Twenty-five children with ILD (median age: 7 ± 3.6 years; male/female = 16/9) were included (Table 1).
Table 1 Patients' characteristics
No Sex Age [years] Interstitial lung disease Diagnosis finding Radiographic findings Fibrotic changes (CT) ILD Score* Dyspnoe Cough Cyanosis Exercise Intolerance Failure to thrive Medication FVC [% of pred.] TLC % [of pred.]
1 F 7 LIP CT, LB • diffuse interstitial involvement + 4 ++ + + + + CS, AZT 34 56
• reticular-nodular pattern
• follicular bronchiolitis
2 M 14 U-ILD, IPH CT, BAL patchy interstitial involvement - 2 + - - - - CS 77 89
3 M 8 U-ILD CT, LB • ground-glass opacity + 3 ++ - - + - 46 74
4 M 4 IPH CT, BAL, LB interstitial involvement - 3 + - - - - 77 168
5 M 16 U-ILD CT, BAL interstitial involvement + 2 + - - + + 76 95
6 F 7 U-ILD CT, BAL interstitial involvement - 2 + + - - + AZT 50 68
7 M 4 CPI CT, LB • diffuse infiltrates + 3 + - - + + AZT 58 64
• ground-glass opacity
8 F 3 NSIP CT, LB • interstitial involvement + 3 ++ - + + + CS n.d. n.d.
• alveolar infiltrates
9 M 8 Sarcoidosis CT, BAL, LB • interstitial involvement + 2 ++ + - + + CS 56 63
• perivascular nodules
10 F 8 U-ILD CT, BAL, LB ground-glass opacity - 1 - + - + - 76 87
11 F 8 CPI CT, LB • interstitial involvement + 2 + + - + + CS 37 74
• ground-glass opacity
12 M 9 U-ILD CT interstitial involvement - 2 + - - - - 70 98
13 M 5 NSIP CT, LB • interstitial involvement + 3 ++ - - + - CS 61 76
• ground-glass opacity
14 F 6 U-ILD CT reticular-nodular pattern + 3 ++ + - - - AZT 60 68
15 F 4 U-ILD CT interstitial involvement + 2 + + - - + n.d. n.d.
16 M 12 U-ILD CT interstitial involvement - 2 + - - - - 68 75
17 M 3 PAP† CT, BAL, LB • ground glass opacity - 4 +++ + + + + CS n.d. n.d.
18 M 6 NSIP CT, BAL, LB • alveolar infiltrates + 4 +++ + + + + CS, AZT 63 72
PAP • ground glass opacification
19 F 3 PAP† CT, BAL, LB • ground glass opacity + 4 ++ - + + + CS n.d. n.d.
• alveolar infiltrates
20 F 9 NSIP CT, LB • interstitial involvement + 3 ++ + + + + CS, AZT 55 74
• honeycombing
21 M 7 U-ILD CT reticular-interstitial pattern + 3 + + - + + AZT, MT 38 59
22 M 7 Cholesterol CT, BAL, LB • interstitial involvement + 4 +++ + + + + CS 16 24
pneumonitis† • reticular-interstitial pattern
23 M 4 U-ILD CT, LB • interstitial involvement - 2 + - - + + CS 102 99
• honeycombing
24 M 8 U-ILD CT interstitial involvement - 2 + + - + - CS 63 78
25 M 7 NSIP CT, LB • interstitial involvement + 3 + + - + - CS 60 76
ILD-NC: children with interstitial lung disease without systemic corticosteroid treatment; ILD-C: children with interstitial lung disease with systemic corticosteroid treatment;
U-ILD: undefined/idiopathic interstitial lung disease: no specific diagnosis could be made; PAP: pulmonary alveolar proteinosis; CGD: chronic granulomatous disease; IPH: idiopathic pulmonary hemosiderosis; LIP: lymphocytic interstitial pneumonia; CPI: Chronic pneumonitis of infancy
CS: corticosteroids, AZT: azathioprine, MT: methotrexat
n.d.: lung function testing not done (children < 5 years); † symbolizes patients who died due to respiratory failure.
CT: Computed tomography; BAL: Bronchoalveolar lavage; LB: Lung biopsy
* ILD score according to Fan[41]
Ten age-matched children were selected as the control group (median age: 7.5 ± 2.9 years, m/f: 6/4). These children were considered as healthy, i.e. had no systemic disease, had no suspected or proven pulmonary disease and were free of respiratory tract infections. These children underwent elective tonsillectomy under general anaesthesia. BAL was performed prior to the surgical procedure.
Ten age-matched children with chronic severe asthma (median age: 8.7 ± 1.6 years, m/f: 5/5), from a previous study[42], who were comparable to the ILD group in terms of gender and age were included as disease control group. All parents and/or patients gave their informed consent prior to bronchoscopy and the institutional review board approved the study protocol.
Bronchoalveolar lavage
Bronchoscopy with BAL was performed as described previously[43]. Residual BALF cells were used for flow cytometry. The BALF recovery and the viability of cells did not differ significantly between the patient groups. Cellular profiles are shown in Table 2.
Table 2 Bronchoalveolar lavage cells
ILD-NC ILD-C Control
Total cells × 103/ml 230 (2.1–1124)** 144 (11–268)* 89 (83–97)
Recovery (%) 55 (25–86) 49 (34–75) 54 (35–70)
Neutrophils (%) 10.5 (1–44)* 8.5 (3–30)* 2 (0–3)
Eosinophils (%) 1 (0–6) 1.5 (0–3) 0 (0–1)
Mast cells (%) 2 (0–43) 2 (1–4) 0 (0-0)
Plasma cells (%) 0 (0–4) 0 (0–4) 0 (0-0)
Macrophages (%) 60 (7–97)* 49 (26–77)* 94 (81–92)
Lymphocytes (%) 24 (2–54)** 22 (5–34)** 4 (2–13)
CD4+ T cells (%)† 23 (9–45) 29 (9–82) 23 (15–28)
CD8+ T cells (%)† 29 (6–62) 27 (2–83) 25 (15–31)
CD4/8 ratio 0.7 (0.3–6) 1.1 (0.1–55) 0.7 (0.4–0.9)
results are expressed as medians with ranges shown in parenthesis.
ILD-NC: children with interstitial lung disease without systemic corticosteroid treatment;
ILD-C: children with interstitial lung disease with systemic corticosteroid treatment;
*p < 0.05, **p < 0.01 as compared to the control group, Mann-Whitney-U Test.
Total cells and differential cell count were obtained from cytospin slides, CD4+, CD8+ and CD4/CD8 T cells using flow cytometry.
†CD4+ T cells and CD8+ T cells are shown as the percentage of total lymphocytes in BALF, i.e. cells gated in the lymphocyte population. Neutrophils, eosinophils, mast cells, plasma cells, macrophages and lymphocytes are shown as percentage of total cells in BALF.
Flow cytometry
BALF cells were analyzed by four-colour flow cytometry (FACSCalibur, Becton-Dickinson, Heidelberg, Germany) as described previously[42]. The following antibodies were used: CD4-allophycocyanine (APC) mouse IgG1, CD8-phycocyanine 5 (PC5) mouse IgG1 (Immunotech, Marseille, France), CD69-PE mouse IgG1, CCR5-PE mouse IgG2a, CCR4-PE mouse IgG2a (BD Pharmingen, Heidelberg, Germany), CCR2-PE mouse IgG2b, CXCR3-fluorescein isothiocyanate (FITC) mouse IgG1 and CCR3-FITC rat IgG2a (R&D Systems, Wiesbaden, Germany). Mouse IgG1-FITC, mouse IgG1-PE, mouse IgG2a-PE, mouse IgG2b-PE (Immunotech, Marseille, France) and rat IgG2a-FITC (kindly provided by Dr. E. Kremmer, GSF-Institute of Molecular Immunology, Munich, Germany) were used as isotype controls.
Detection of MCP-1 and cytokines
Levels of MCP-1 and Th1 (IL-2, IFN-γ), Th2 (IL-4, IL-5, IL-10) and pro-inflammatory cytokines (TNF-α, IL-6) were quantified by a multiplex, particle-based assay (Bio-Rad Laboratories, Minneapolis, USA) as described previously[42]. The detection limits for all cytokines were 1.5–2.5 pg/ml (min.) and 1000 pg/ml (max.).
Quantitative RT-PCR
BALF cells were lysed in Trizol LS Reagent (Invitrogen, Life Technologies, Karlsruhe, Germany) and were stored at -20°C until mRNA extraction. Total mRNA was isolated according to the manufacturer's instructions and reverse transcribed into cDNA. Contamination with genomic DNA was excluded by mRNA controls without reverse transcriptase in the cDNA synthesis reaction. The following oligonucleotide primers were used: MCP-1 (5-TGAAGCTCGCACTCTCGCCT-3; 5- GTGGAGTGAGTGTTCAAGTC-3); and GAPDH (5-GAGGTGAAGGTCGGAGTC-3; 5-AAGATGGTGATGGGATTTC-3). Expression levels were determined in duplicates by Real time RT-PCR using SYBR green and the iCycler iQ detection system (Biorad, Hercules, CA, USA) according to the manufacturer's instructions. Threshold cycle (CT) values for genes of interest were normalized to GAPDH and used to calculate the relative mRNA expression.
Statistical analysis
The non-parametric Mann-Whitney U test was applied. Correlations were tested with Spearman's rank correlation test. A probability of p < 0.05 was regarded as significant[44] (SPSS statistical program, version 11.5, SPSS Inc. Chicago, USA).
Results
MCP-1 levels and CCR2+ T cells in BALF
Levels of MCP-1 were significantly higher in children with ILD (n = 25) as compared to the control group at protein and mRNA level (Figure 1A, B). MCP-1 protein and mRNA expression levels correlated positively with each other (r = 0.72, p < 0.01). ILD children with pulmonary fibrosis had significantly higher MCP-1 levels in BALF as compared to children with non-fibrotic ILD (Figure 1C). MCP-1 levels related positively to the stage of disease (Figure 1D). The highest levels of MCP-1 were observed in the three patients who died after respiratory failure (Table 1; P17, P19, P22). Furthermore, MCP-1 levels correlated negatively with restrictive lung function parameters (TLC, FVC) (Figures 2A, B).
Figure 1 MCP-1 levels in children with ILD. MCP-1 levels in bronchoalveolar lavage fluid (BALF) of children with interstitial lung diseases (ILD) and healthy controls are shown at the (A) protein and at the (B) mRNA level. (C) MCP-1 levels in BALF of ILD children with and without pulmonary fibrosis. Pulmonary fibrosis was assessed by computed tomography according to [36,37]. (D) MCP-1 levels in ILD children related to ILD disease severity according to the criteria of Fan [33]. 1 = asymptomatic, no desaturation; 2 = symptomatic but normoxic (> 90%) under all conditions; 3 = symptomatic with desaturation during sleep or exercise; 4 = symptomatic with desaturation at rest; MCP-1 protein levels were quantified in BALF by a multiplex, particle-based assay (Bio-Rad Laboratories, Minneapolis, USA) as described previously [42]. MCP-1 mRNA levels were quantified in BALF cells by Real time RT-PCR using SYBR green and the iCycler iQ detection system (Biorad, Hercules, CA, USA) and were normalized to GAPDH. Median values are shown by horizontal bars. Differences between the patient groups were tested with the Mann-Whitney U test; * p < 0.05, *** p < 0.001; Children with systemic corticosteroid therapy are shown as grey circles. P: Pulmonary alveolar proteinosis; S: Sarcoidosis; † symbolize children who died due to respiratory failure.
Figure 2 Correlation of MCP-1 levels with lung function parameters in children with ILD. MCP-1 levels in bronchoalveolar lavage fluid (BALF) correlated with (A) forced vital capacity (FVC) and (B) total lung capacity (TLC) in children with interstitial lung disease (ILD). FVC and TLC are shown as % of predicted. MCP-1 levels in BALF were quantified by a multiplex, particle-based assay; P: Pulmonary alveolar proteinosis; S: Sarcoidosis;
To test whether increased MCP-1 levels are associated with increased frequencies of CCR2+ T cells, BALF lymphocytes were quantified by flow cytometry. CCR2 was expressed on a higher percentage of CD4+ than CD8+ T cells. The majority of CCR2+ T cells showed an activated phenotype (75% CCR2+CD69+). Children with ILD had significantly higher percentages of CCR2+CD4+and CCR2+CD8+ T cells in BALF as compared to control children (Figure 3A). Similar to MCP-1, percentages of CCR2+CD4+ cells were significantly higher in ILD children with pulmonary fibrosis as compared to children with non-fibrotic ILD (Figure 3B). Again, the highest percentages of CCR2+CD4+ T cells were observed in the three deceased patients and CCR2+CD4+ cells related positively to the stage of ILD (Figure 4). Furthermore, percentages of CCR2+CD4+ T cells correlated negatively with FVC and TLC in ILD patients (Figures 5A, B). Pulmonary levels of MCP-1 correlated positively with CCR2+CD4+ T cells (Figure 5C). No association between MCP-1/CCR2+ cells and immunosuppressive treatment was found in ILD patients.
Figure 3 CCR2+ T cells in children with ILD. (A) Percentages of CCR2+CD4+ and CCR2+CD8+ T cells in in bronchoalveolar lavage fluid (BALF) of children with interstitial lung diseases (ILD) and healthy children. (B) Percentages of CCR2+CD4+ and CCR2+CD8+ T cells in BALF of children with and without pulmonary fibrosis. Percentages of CCR2+CD4+ and CCR2+CD8+ T cells were analyzed in BALF by flow cytometry. Pulmonary fibrosis was assessed by computed tomography according to [36,37]. Median values are shown by horizontal bars. Differences between the patient groups were tested with the Mann-Whitney U test; * p < 0.05; *** p < 0.001; Children with systemic corticosteroid therapy are shown as grey circles. P: Pulmonary alveolar proteinosis; S: Sarcoidosis; † symbolize the children who died due to respiratory failure.
Figure 4 CCR2+ CD4+ T cells and ILD disease severity. Percentages of CCR2+CD4+ T cells in bronchoalveolar lavage fluid (BALF) of children with interstitial lung disease (ILD) related to ILD disease severity. Percentages of CCR2+CD4+ T cells were analyzed in BALF by flow cytometry. ILD disease severity was scored according to the ILD score of Fan(40): 1 = asymptomatic, no desaturation; 2 = symptomatic but normoxic (> 90%) under all conditions; 3 = symptomatic with desaturation during sleep or with exercise; 4 = symptomatic with desaturation at rest; Median values are shown by horizontal bars. Differences between the patient groups were tested with the Mann-Whitney U test; * p < 0.05, ** p < 0.01; Children with systemic corticosteroid therapy are shown as grey circles. P: Pulmonary alveolar proteinosis; S: Sarcoidosis; † symbolize the children who died due to respiratory failure.
Figure 5 CCR2+CD4+ T cells and lung function parameters in children with ILD. Correlation of CCR2+CD4+ T cells in bronchoalveolar lavage fluid (BALF) with (A) forced vital capacity (FVC) and (B) total lung capacity (TLC) in children with interstitial lung diseases (ILD). Correlation of percentages of CCR2+CD4+ T cells with levels of MCP-1 in BALF of children with ILD (C). FVC and TLC are shown as % of predicted. Percentages of CCR2+CD4+ T cells were analyzed in BALF by flow cytometry. P: Pulmonary alveolar proteinosis; S: Sarcoidosis
To verify if increased levels of MCP-1 and percentages of CCR2+ T cells are characteristic for pediatric ILD, we analyzed these markers in ten selected age-matched children with well-characterized allergic asthma who are described in detail in a previous study[42]. Levels of MCP-1 and CCR2+ T cells from asthmatic children were in the range of the control group and did not correlate with each other (data not shown).
To assess the value of CCR2+CD4+ T cells and MCP-1 levels in the longitudinal course, three consecutive therapeutical BALs were analyzed in three patients with PAP (P17, P18, P19) and one patient with cholesterol pneumonitis (P22). Two PAP patients (P17, P19) and the patient with cholesterol pneumonitis worsened in the clinical course continuously (increasing oxygen requirement, increasing dyspnoe) and died from respiratory failure, while one PAP patient remained clinically stable (P18). The deceased PAP patients had continuously rising levels of MCP-1 and increasing percentages of CCR2+CD4+ T cells in the three follow-up BALs (Figures 6A, B; black circles) while the clinically stable patient showed steady levels of MCP-1 and percentages of CCR2+CD4+ T cells (Figures 6A, B; white circles).
Figure 6 Longitudinal analysis of MCP-1 levels and CCR2+CD4+ T cells. Longitudinal analysis of (A) MCP-1 levels and (B) CCR2+CD4+ T cells in three consecutive bronchoalveolar lavage fluids (BALF) of four children with interstitial lung diseases, including two children with pulmonary alveolar proteinosis (P) and one child with cholesterol pneumonitis (CP). The child with cholesterol pneumonitis and one child with pulmonary alveolar proteinosis died by respiratory failure (†), while one child with pulmonary alveolar proteinosis stayed clinically stable. † symbolize the childen who died. MCP-1 levels were quantified in BALF by a multiplex, particle-based assay. Percentages of CCR2+CD4+ T cells were analyzed in BALF by flow cytometry.
Th1- and Th2-lymphocytes and cytokines in BALF
To test whether increased CCR2+ T cells and levels of MCP-1 were paralleled by a pulmonary Th1/Th2-shift, CCR4+ and CCR3+(Th2) and CCR5+an d CXCR3+ (Th1) cells were determined in BALF together with an array of pulmonary Th1/Th2 cytokines.
Children with ILD had significantly higher percentages of CCR4+CD4+ (Th2) cells as compared to control children (Figure 7A). CCR4 was predominantly expressed on CD4+ cells. The majority of CCR4+CD4+ cells had an activated phenotype (68% CCR4+CD69+). CCR3+ (Th2) cells were not detectable in BALF. Percentages of CCR5+ and CXCR3+ T cells (both Th1) were low and did not differ among the patient groups (Figures 7B, C).
Figure 7 Pulmonary CCR4+, CCR5+, and CXCR3+ T cells. Percentages of (A) CCR4+CD4+, CCR4+CD8+, (B) CCR5+CD4+ and CCR5+CD8+ and (C) CXCR3+CD4+ and CXCR3+CD8+ T cells in bronchoalveolar lavage fluid (BALF) are shown in children with interstitial lung diseases (ILD) and healthy controls. Percentages of CCR4+CD4+, CCR4+CD8+, CCR5+CD4+, CCR5+CD8+, CXCR3+CD4+ and CXCR3+CD8+ T cells were analyzed in BALF by flow cytometry. Median values are shown by horizontal bars. Differences between the patient groups were tested with the Mann-Whitney U test; * p < 0.05; ** p < 0.01
Levels of IFN-γ were increased in ILD patients (p < 0.05), whereas the remaining cytokines did not differ among the patient groups (data not shown).
Discussion
The present work demonstrates that BALF levels of MCP-1 are consistently increased in pediatric ILD. This is accompanied by increased frequencies of the corresponding CCR2 + T cells. Levels of MCP-1 and frequencies of CCR2+ T cells were higher in fibrotic than in non-fibrotic forms of ILD and correlated with restrictive lung function parameters and ILD disease severity, indicating a relevance of the MCP-1/CCR2 axis in the pathogenesis of pediatric ILD. Infiltrating T cells are a characteristic feature of pulmonary tissue from ILD patients[45] and T cells in BALF were found to correlate with T cells in pulmonary tissue[46]. In line with previous findings[47,48], T cells were increased in BALF of our children with ILD as compared to control patients, suggesting a contribution of T cells to the pathogenesis of pediatric ILD. Studies in adult patients indicated that MCP-1 plays a role in the pathogenesis of different forms of ILD, including IPF[9,12,13], PAP[11,14], sarcoidosis[12], scleroderma with lung involvement[49] and granulomatous lung diseases[50]. Serum levels of MCP-1 were significantly elevated in adult patients with ILD[9,51] and were closely related to the clinical course[9]. However, as outlined above, ILD in children differs noticeably from ILD in adulthood. Pediatric ILD is extremely rare and little data exist with respect to pathoimmunological mechanisms. Thus, it is very hard to study a large patient group and to find enough children for each ILD subtype. We found elevated levels of MCP-1 and CCR2+ T cells in various etiologies of ILD, which suggests a common pulmonary T cell response for various forms of pediatric ILD.
Thus far, frequencies of BALF CCR2+ T cells in human ILD have not been determined. The parallel increase of MCP-1 and CCR2+ T cells in BALF of ILD children further substantiates the importance of this chemokine and its receptor in the pathogenesis of ILD, as suggested by mouse models. In these models, the relevance of the MCP-1/CCR2 interaction was mainly addressed with respect to pulmonary fibrosis. Our children with pulmonary fibrosis had increased levels of MCP-1 and increased percentages of CCR2+ cells compared to children with non-fibrotic ILD. However, MCP-1 levels and percentages of CCR2+ T cells were elevated both in fibrotic and non-fibrotic ILD children as compared to controls. In addition, MCP-1 and CCR2+ T cells were also elevated in pediatric PAP that usually does not progress to pulmonary fibrosis. In fact, one of the three patients with the highest levels of MCP-1 and CCR2+ T cells had PAP without any indication of fibrosis. Similar observations were made recently for MCP-1 in adult PAP patients[11]. A possible biological relevance of MCP-1 levels and CCR2+ T cells in pediatric ILD is further suggested by their correlation with restrictive lung function parameters and the ILD disease severity score and by the finding that the deceased children with the most severe course of disease exhibited the highest BALF levels of these markers. The possibility that MCP-1 and CCR2+ T cells are a general phenomenon of pediatric lung diseases seems very unlikely, since these markers were present only at low levels in BALF of children with severe allergic asthma. This is in line with findings in an Aspergillus-induced allergic mouse model, where a Th2-mediated lung pathology occured in the absence of MCP-1 or CCR2[52].
To assess the value of CCR2+ T cells and MCP-1 levels in the longitudinal course of children with ILD, three consecutive BALs were performed in four children with ILD including three ILD patients who died and one patient who stayed clinically stable. The three deceased children had high and continuously rising levels of MCP-1 and CCR2+CD4+ T cells, while the stable patient had low levels of MCP-1 and percentages of CCR2+CD4+ T cells. Thus, levels of MCP-1 and percentages of CCR2+CD4+ T cells might reflect the disease progression in pediatric ILD.
Interestingly, immunosuppressive treatment was not associated with altered levels of MCP-1 and CCR2+ T cells in BALF (data not shown). This is in contrast to a study of Suga et al.[9] in adult ILD patients where serum levels of MCP-1 were closely related to the effectiveness of corticosteroid therapy. Given the assumption that MCP-1 and CCR2 are important players in the pathophysiology of ILD in children, the lack of association with corticosteroid therapy might explain, at least in part, why corticosteroids are sometimes unable to control the progression of pediatric ILD.
Several studies indicated that MCP-1 and CCR2 are involved in Th1[53,54] and Th2 immunity [55-58]. Furthermore, it has been suggested that ILD and pulmonary fibrosis are associated with a Th2 immune response[20-22,59-61]. Experiments in mice showed that a lack of MCP-1[62] leads to decreased Th1 responses while MCP-1 over-expression[58] results in increased levels of Th2 cytokines. Th1/Th2 cytokine levels in BALF were low or undetectable in BALF of our children. However, CCR4+CD4+ T cells were moderately but significantly elevated in ILD patients. On the other hand, CCR4+CD4+ T cells are clearly less frequent in ILD compared to allergic asthma[42]. Thus, a strong Th2 response seems unlikely in our ILD patients. Beneath T-cells, MCP-1 attracts CCR2+ monocytes/macrophages[63]. In mouse models, MCP-1 was found to attract monocytes to the inflamed lung, which was accompanied by a concomitant downregulation of pulmonary MCP-1 levels[64]. We found no difference in the percentage of CCR2+ alveolar macrophages in BALF between children with ILD and control children or between fibrotic and non-fibrotic forms of ILD (data not shown). Instead, we found a strong correlation between percentages of CCR2+ T cells and levels of MCP-1 in BALF of ILD patients. Therefore, we assume that pulmonary MCP-1 acts on CCR2+ T cells, which accumulate in the BALF of children with ILD.
Conclusion
In conclusion, CCR2+ T cells and levels of MCP-1 are characteristic components in BALF of children with ILD. A pathophysiological role in pediatric ILD seems likely as their levels relate to restrictive lung function and ILD disease severity. Therefore, pharmacological targeting of the MCP-1/CCR2 axis might represent an additional option for the treatment of ILD in children.
Abbreviations
BAL(F): Bronchoalveolar lavage (fluid)
CC: CC chemokine receptor
CXC: CXC chemokine receptor
FVC: Forced vital capacity
IFN-γ: Interferon-γ
IL-: Interleukin
IPF: Idiopathic pulmonary fibrosis
IPH: Idiopathic pulmonary hemosiderosis
LIP: Lymphocytic interstitial pneumonia
MCP-1: Monocyte chemotactic protein 1 (CCL2)
PAP: Pulmonary alveolar proteinosis
TGF-β: Transforming growth factor β
Th1/Th2: T helper cell 1/2
TLC: Total lung capacity
TNF-α: Tumor necrosis factor-α
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
DH carried out the experimental analyses and wrote the manuscript. MG characterized the study population, performed bronchoalveolar lavage and participated in the study design. TN performed bronchoalveolar lavage and patient characterization. GZ and CP participated in the experimental analyses. DR and DJS participated in the study design and reviewed the manuscript. SKE designed the study, supervised the experimental analyses and wrote the manuscript. All authors read and approved the final manuscript.
Acknowledgements
This work was supported by grants from the Else-Kröner-Fresenius Stiftung, the Friedrich-Baur-Stiftung, by a grant of the University and Science Program of the Ludwig-Maximilians-University (HWP) and by the Clinical Cooperation Groups "Pediatric Immune Regulation" and "Immune Monitoring". We thank Cory M. Hogaboam, Department of Pathology, University of Michigan Medical School, Ann Arbor, for helpful discussions and critical revision of the mansucript.
==== Refs
Clement A Allen J Corrin B Dinwiddie R le Pointe HD Eber E Task force on chronic interstitial lung disease in immunocompetent children European Respiratory Journal 2004 24 686 697 15459150 10.1183/09031936.04.00089803
Bush A Diagnosis of interstitial lung disease Pediatric Pulmonology 1996 22 81 82 8875578
Hartl D Griese M Interstitial lung disease in children – genetic background and associated phenotypes Respiratory Research 2005 8
Yoshimura T Leonard EJ Human Monocyte Chemoattractant Protein-1 (MCP-1) – Secretion by Human-Fibroblasts Journal of Leukocyte Biology 1989 46 331
Yoshimura T Yuhki N Moore SK Appella E Lerman MI Leonard EJ Human Monocyte Chemoattractant Protein-1 (MCP-1) – Full-Length Cdna Cloning, Expression in Mitogen-Stimulated Blood Mononuclear Leukocytes, and Sequence Similarity to Mouse Competence Gene Je FEBS Letters 1989 244 487 493 2465924 10.1016/0014-5793(89)80590-3
Lundien MC Mohammed KA Nasreen N Tepper RS Hardwick JA Sanders KL Induction of MCP-1 expression in airway epithelial cells: Role of CCR2 receptor in airway epithelial injury Journal of Clinical Immunology 2002 22 144 152 12078856 10.1023/A:1015420029430
GharaeeKermani M Denholm EM Phan SH Costimulation of fibroblast collagen and transforming growth factor beta(1) gene expression by monocyte chemoattractant protein-1 via specific receptors Journal of Biological Chemistry 1996 271 17779 17784 8663511 10.1074/jbc.271.30.17779
Hogaboam GM Bone-Larson CL Lipinski S Lukacs NW Chensue SW Strieter RM Differential monocyte chemoattractant protein-1 and chemokine receptor 2 expression by murine lung fibroblasts derived from Th1-and Th2-type pulmonary granuloma models Journal of Immunology 1999 163 2193 2201
Suga M Iyonaga K Ichiyasu H Saita N Yamasaki H Ando M Clinical significance of MCP-1 levels in BALF and serum in patients with interstitial lung diseases European Respiratory Journal 1999 14 376 382 10515417 10.1034/j.1399-3003.1999.14b23.x
Iyonaga K Suga M Ichiyasu H Yamamoto T Hiraga Y Ando M Measurement of serum monocyte chemoattractant protein-1 and its clinical application for estimating the activity of granuloma formation in sarcoidosis Sarcoidosis Vasculitis and Diffuse Lung Diseases 1998 15 165 172
Bonfield TL John N Malur A Barna BP Culver DA Kavuru MS Elevated monocyte chemotactic proteins 1, 2, and 3 in pulmonary alveolar proteinosis are associated with chemokine receptor suppression Clinical Immunology 2005 114 79 85 15596412 10.1016/j.clim.2004.09.004
Car BD Meloni F Luisetti M Semenzato G Gialdronigrassi G Walz A Elevated Il-8 and Mcp-1 in the Bronchoalveolar Lavage Fluid of Patients with Idiopathic Pulmonary Fibrosis and Pulmonary Sarcoidosis American Journal of Respiratory and Critical Care Medicine 1994 149 655 659 8118632
Iyonaga K Takeya M Saita N Sakamoto O Yoshimura T Ando M Monocyte Chemoattractant Protein-1 in Idiopathic Pulmonary Fibrosis and Other Interstitial Lung-Diseases Human Pathology 1994 25 455 463 8200639 10.1016/0046-8177(94)90117-1
Iyonaga K Suga M Yamamoto T Ichiyasu H Miyakawa H Ando M Elevated bronchoalveolar concentrations of MCP-1 in patients with pulmonary alveolar proteinosis European Respiratory Journal 1999 14 383 389 10515418 10.1034/j.1399-3003.1999.14b24.x
Carr MW Roth SJ Luther E Rose SS Springer TA Monocyte Chemoattractant Protein-1 Acts As A T-Lymphocyte Chemoattractant Proceedings of the National Academy of Sciences of the United States of America 1994 91 3652 3656 8170963
Murphy PM Baggiolini M Charo IF Hebert CA Horuk R Matsushima K International union of pharmacology. XXII. Nomenclature for chemokine receptors Pharmacological Reviews 2000 52 145 176 10699158
Maus U von Grote K Kuziel WA Mack M Miller EJ Cihak J The role of CC chemokine receptor 2 in alveolar monocyte and neutrophil immigration in intact mice American Journal of Respiratory and Critical Care Medicine 2002 166 268 273 12153956 10.1164/rccm.2112012
Moore BB Paine R Christensen PJ Moore TA Sitterding S Ngan R Protection from pulmonary fibrosis in the absence of CCR2 signaling Journal of Immunology 2001 167 4368 4377
Wynn TA Fibrotic disease and the Th1/Th2 paradigm Nature Reviews Immunology 2004 4 583 594 15286725 10.1038/nri1412
Lukacs NW Hogaboam C Chensue SW Blease K Kunkel SL Type 1/type 2 cytokine paradigm and the progression of pulmonary fibrosis Chest 2001 120 5S 8S 11451879 10.1378/chest.120.1_suppl.S5
Jakubzick C Choi ES Carpenter KJ Kunkel SL Evanoff H Martinez FJ Human pulmonary fibroblasts exhibit altered interleukin-4 and interleukin-13 receptor subunit expression in idiopathic interstitial pneumonia American Journal of Pathology 2004 164 1989 2001 15161635
Jakubzick C Choi ES Kunkel SL Evanoff H Martinez FJ Puri RK Augmented pulmonary IL-4 and IL-13 receptor subunit expression in idiopathic interstitial pneumonia Journal of Clinical Pathology 2004 57 477 486 15113854 10.1136/jcp.2003.012799
Monroe JG Haldar S Prystowsky MB Lammie P Lymphokine Regulation of Inflammatory Processes – Interleukin-4 Stimulates Fibroblast Proliferation Clinical Immunology and Immunopathology 1988 49 292 298 3262472 10.1016/0090-1229(88)90119-5
Postlethwaite AE Holness MA Katai H Raghow R Human Fibroblasts Synthesize Elevated Levels of Extracellular-Matrix Proteins in Response to Interleukin-4 Journal of Clinical Investigation 1992 90 1479 1485 1401080
Duncan MR Berman B Gamma-Interferon Is the Lymphokine and Beta-Interferon the Monokine Responsible for Inhibition of Fibroblast Collagen Production and Late But Not Early Fibroblast Proliferation Journal of Experimental Medicine 1985 162 516 527 2410528 10.1084/jem.162.2.516
Serpier H Gillery P SalmonEhr V Garnotel R Georges N Kalis B Antagonistic effects of interferon-gamma and interleukin-4 on fibroblast cultures Journal of Investigative Dermatology 1997 109 158 162 9242501 10.1111/1523-1747.ep12319207
Yuan WH Yufit T Li LY Mori Y Chen SJ Varga J Negative modulation of (alpha 1(I) procollagen gene expression in human skin fibroblasts: Transcriptional inhibition by interferon-gamma Journal of Cellular Physiology 1999 179 97 108 10082137 10.1002/(SICI)1097-4652(199904)179:1<97::AID-JCP12>3.0.CO;2-E
Oldroyd SD Thomas GL Gabbiani G El Nahas AM Interferon-gamma inhibits experimental renal fibrosis Kidney International 1999 56 2116 2127 10594787 10.1046/j.1523-1755.1999.00775.x
Buttner C Skupin A Reimann T Rieber EP Unteregger G Geyer P Local production of interleukin-4 during radiation-induced pneumonitis and pulmonary fibrosis in rats: Macrophages as a prominent source of interleukin-4 American Journal of Respiratory Cell and Molecular Biology 1997 17 315 325 9308918
Emura M Nagai S Takeuchi M Kitaichi M Izumi T In vitro Production of B-Cell Growth-Factor and B-Cell Differentiation Factor by Peripheral-Blood Mononuclear-Cells and Bronchoalveolar Lavage Lymphocytes-T from Patients with Idiopathic Pulmonary Fibrosis Clinical and Experimental Immunology 1990 82 133 139 2208788
Wallace WAH Ramage EA Lamb D Howie SEM A Type-2 (Th2-Like) Pattern of Immune-Response Predominates in the Pulmonary Interstitium of Patients with Cryptogenic Fibrosing Alveolitis (CFA) Clinical and Experimental Immunology 1995 101 436 441 7664490
Fan LL Langston C Pediatric interstitial lung disease – Children are not small adults American Journal of Respiratory and Critical Care Medicine 2002 165 1466 1467 12045116 10.1164/rccm.2204012
Fan LL Kozinetz CA Deterding RR Brugman SM Evaluation of a diagnostic approach to pediatric interstitial lung disease Pediatrics 1998 101 82 85 9417156 10.1542/peds.101.1.82
[Anon] Standardization of Spirometry – 1994 Update American Journal of Respiratory and Critical Care Medicine 1995 152 1107 1136 7663792
Agusti C American Thoracic Society/European Respiratory Society International Multidisciplinary Consensus Classification of the Idiopathic Interstitial Pneumonias American Journal of Respiratory and Critical Care Medicine 2002 165
Kazerooni EA Martinez FJ Flint A Jamadar DA Gross BH Spizarny DL Thin-section CT obtained at 10-mm increments versus limited three-level thin-section CT for idiopathic pulmonary fibrosis: Correlation with pathologic scoring American Journal of Roentgenology 1997 169 977 983 9308447
Copley SJ Coren M Nicholson AG Rubens MB Bush A Hansell DM Diagnostic accuracy of thin-section CT and chest radiography of pediatric interstitial lung disease American Journal of Roentgenology 2000 174 549 554 10658741
Brasch F Muller KM Classification of pulmonary alveolar proteinosis in newborns, infants, and children Pathologe 2004 25 299 309 15241624 10.1007/s00292-004-0706-5
[Anon] Idiopathic pulmonary fibrosis: Diagnosis and treatment – International consensus statement American Journal of Respiratory and Critical Care Medicine 2000 161 646 664 10673212
Coren ME Nicholson AG Goldstraw P Bush A Open lung biopsy in the investigation of diffuse lung disease in children American Journal of Respiratory and Critical Care Medicine 1999 159 A779
Fan LL Langston C Chronic Interstitial Lung-Disease in Children Pediatric Pulmonology 1993 16 184 196 8309744
Hartl D Griese M Nicolai T Zissel G Prell C Konstantopoulos N Pulmonary chemokines and their receptors differentiate children with asthma and chronic cough Journal of Allergy and Clinical Immunology 2005 115 728 736 15805991 10.1016/j.jaci.2004.11.049
Griese M Felber J Reiter K Strong P Reid K Belohradsky BH Airway inflammation in children with tracheostomy Pediatric Pulmonology 2004 37 356 361 15022133 10.1002/ppul.10432
Motulsky H Intuitive biostatistics 1995 New York: Oxford University Press
Katzenstein ALA Myers JL Idiopathic pulmonary fibrosis – Clinical relevance of pathologic classification American Journal of Respiratory and Critical Care Medicine 1998 157 1301 1315 9563754
Semenzato G Bortolin M Facco M Tassinari C Sancetta R Agostini C Lung lymphocytes: Origin, biological functions, and laboratory techniques for their study in immune-mediated pulmonary disorders Critical Reviews in Clinical Laboratory Sciences 1996 33 423 455 8922892
Ronchetti R Midulla F Sandstrom T Bjermer L Zebrak J Pawlik J Bronchoalveolar lavage in children with chronic diffuse parenchymal lung disease Pediatric Pulmonology 1999 27 395 402 10380091 10.1002/(SICI)1099-0496(199906)27:6<395::AID-PPUL6>3.0.CO;2-K
Tessier V Chadelat K Baculard A Housset B Clement A BAL in children – A controlled study of differential cytology and cytokine expression profiles by alveolar cells in pediatric sarcoidosis Chest 1996 109 1430 1438 8769489
Luzina IG Atamas SP Wise R Wigley FM Xiao HQ White B Gene expression in bronchoalveolar lavage cells from scleroderma patients American Journal of Respiratory Cell and Molecular Biology 2002 26 549 557 11970906
Oshima M Maeda A Ishioka S Hiyama K Yamakido M Expression of C-C chemokines in bronchoalveolar lavage cells from patients with granulomatous lung diseases Lung 1999 177 229 240 10384061
Ohnishi H Yokoyama A Kondo K Hamada H Abe M Nishimura K Comparative study of KL-6, surfactant protein-A, surfactant protein-D, and monocyte chemoattractant protein-1 as serum markers for interstitial lung diseases American Journal of Respiratory and Critical Care Medicine 2002 165 378 381 11818324
Koth LL Rodriguez MW Bernstein XL Chan S Huang XZ Charo IF Aspergillus antigen induces robust Th2 cytokine production, inflammation, airway hyperreactivity and fibrosis in the absence of MCP-1 or CCR2 Respiratory Research 2004 5
Traynor TR Herring AC Dorf ME Kuziel WA Toews GB Huffnagle GB Differential roles of CC chemokine ligand 2/monocyte chemotactic protein-1 and CCR2 in the development of T1 immunity Journal of Immunology 2002 168 4659 4666
Boring L Gosling J Chensue SW Kunkel SL Farese RV Broxmeyer HE Impaired monocyte migration and reduced type 1 (Th1) cytokine responses in C-C chemokine receptor 2 knockout mice Journal of Clinical Investigation 1997 100 2552 2561 9366570
Chonsue SW Boring L Warmington KS Ruth JR Charo IF Kunkel SL Effect of C-C chemokine receptor 2 (CCR2) knockout on Schistosomal egg antigen-elicited granuloma formation and the regional lymphoid response Faseb Journal 1998 12 A646
Gu L Tseng S Horner RM Tam C Loda M Rollins BJ Control of Th2 polarization by the chemokine monocyte chemoattractant protein-1 Nature 2000 404 407 411 10746730 10.1038/35006097
Kim Y Sung SSJ Kuziel WA Feldman S Fu SM Rose CE Enhanced airway Th2 response after allergen challenge in mice deficient in CC chemokine receptor-2 (CCR2) Journal of Immunology 2001 166 5183 5192
Matsukawa A Lukacs NW Standiford TJ Chensue SW Kunkel SL Adenoviral-mediated overexpression of monocyte chemoattractant protein-1 differentially alters the development of Th1 and Th2 type responses in vivo Journal of Immunology 2000 164 1699 1704
Belperio JA Dy M Murray L Burdick MD Xue YY Strieter RM The role of the Th2CC chemokine ligand CCL17 in pulmonary fibrosis Journal of Immunology 2004 173 4692 4698
Kunkel SL Strieter RM Cytokine Networking in Lung Inflammation Hospital Practice 1990 25 63 2120257
Selman M Plunging into the chaos of the cytokine/chemokine cocktail in pulmonary fibrosis – How many and how important are they? American Journal of Respiratory and Critical Care Medicine 2003 168 730 731 14522809 10.1164/rccm.2307002
Lu B Rutledge BJ Gu L Fiorillo J Lukacs NW Kunkel SL Abnormalities in monocyte recruitment and cytokine expression in monocyte chemoattractant protein 1-deficient mice Journal of Experimental Medicine 1998 187 601 608 9463410 10.1084/jem.187.4.601
Maus U von Grote K Kuziel WA Mack M Miller EJ Cihak J The role of CC chemokine receptor 2 in alveolar monocyte and neutrophil immigration in intact mice American Journal of Respiratory and Critical Care Medicine 2002 166 268 273 12153956 10.1164/rccm.2112012
Maus UA Wellmann S Hampl C Kuziel WA Srivastava M Mack M CCR2-positive monocytes recruited to inflamed lungs downregulate local CCL2 chemokine levels American Journal of Physiology-Lung Cellular and Molecular Physiology 2005 288 L350 L358 15516494 10.1152/ajplung.00061.2004
|
16095529
|
PMC1199626
|
CC BY
|
2021-01-04 16:23:25
|
no
|
Respir Res. 2005 Aug 11; 6(1):93
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-93
|
oa_comm
|
==== Front
Theor Biol Med ModelTheoretical Biology & Medical Modelling1742-4682BioMed Central London 1742-4682-2-271602950810.1186/1742-4682-2-27ResearchStatistical distribution of blood serotonin as a predictor of early autistic brain abnormalities Janušonis Skirmantas [email protected] Yale University School of Medicine, Department of Neurobiology, P.O. Box 208001, New Haven, CT 06520-8001, USA2005 19 7 2005 2 27 27 9 3 2005 19 7 2005 Copyright © 2005 Janušonis; licensee BioMed Central Ltd.2005Janušonis; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
A wide range of abnormalities has been reported in autistic brains, but these abnormalities may be the result of an earlier underlying developmental alteration that may no longer be evident by the time autism is diagnosed. The most consistent biological finding in autistic individuals has been their statistically elevated levels of 5-hydroxytryptamine (5-HT, serotonin) in blood platelets (platelet hyperserotonemia). The early developmental alteration of the autistic brain and the autistic platelet hyperserotonemia may be caused by the same biological factor expressed in the brain and outside the brain, respectively. Unlike the brain, blood platelets are short-lived and continue to be produced throughout the life span, suggesting that this factor may continue to operate outside the brain years after the brain is formed. The statistical distributions of the platelet 5-HT levels in normal and autistic groups have characteristic features and may contain information about the nature of this yet unidentified factor.
Results
The identity of this factor was studied by using a novel, quantitative approach that was applied to published distributions of the platelet 5-HT levels in normal and autistic groups. It was shown that the published data are consistent with the hypothesis that a factor that interferes with brain development in autism may also regulate the release of 5-HT from gut enterochromaffin cells. Numerical analysis revealed that this factor may be non-functional in autistic individuals.
Conclusion
At least some biological factors, the abnormal function of which leads to the development of the autistic brain, may regulate the release of 5-HT from the gut years after birth. If the present model is correct, it will allow future efforts to be focused on a limited number of gene candidates, some of which have not been suspected to be involved in autism (such as the 5-HT4 receptor gene) based on currently available clinical and experimental studies.
==== Body
Background
Our ability to treat and prevent autism is severely limited by our lack of knowledge of what biological abnormality causes this developmental disorder. Since autism is considered primarily a brain disorder, much of the research over the past decades has focused on the autistic brain. Different groups have reported a wide range of anatomical abnormalities in autistic brains, such as reduced numbers of Purkinje cells in the cerebellum [1-3]; an unusually rapid growth of the cerebral cortical volume and head circumference during the first years after birth [4-9]; abnormal cortical minicolumns [10-13]; abnormalities of the limbic system [14-19]; abnormalities of the brainstem [20-22]; and other brain alterations [23-25].
Considering the complexity of brain development and its highly dynamic nature, these abnormalities may be the result of a long, complex chain of events. The original abnormality that caused them may occur early in development [26] and may be no longer obvious by the time autism is diagnosed. For example, an autistic-like loss of Purkinje cells may be caused by a mutation of the toppler gene, which causes severe ataxia in mice and appears to be irrelevant to autism [27]. Post-mortem analysis of younger autistic brains is not an option, because it is usually not clear until age 2 or 3 which brains are autistic and which are not.
Fortunately, evidence suggests that at least one biological factor that causes the development of the autistic brain has a different function outside the central nervous system (CNS), where it continues to operate well into childhood and perhaps even into adulthood. Since the early 1960s, the most consistent biological finding in autistic individuals has been their statistically elevated serotonin (5-hydroxytryptamine, 5-HT) levels in blood platelets, or platelet hyperserotonemia [28-33]. Unlike many of the reported alterations in the brain, this finding has been replicated numerous times by different groups, some of which have used large numbers of subjects. According to Anderson [33], "the platelet hyperserotonemia of autism [...] is generally considered to be one of the more robust and well-replicated findings in biological psychiatry". The main reason why we have not capitalized on this major finding is that we have not been able to understand its origin or its relation to the brain.
It is unlikely that the autistic platelet hyperserotonemia is induced by the brain. The human blood-brain barrier (BBB) becomes mature around one year after birth, if not earlier [34,35], and is virtually impenetrable to 5-HT. Tryptophan, a 5-HT precursor, can cross the BBB, but tryptophan levels do not appear to be altered in autistic individuals [36]. Unlike the anatomy of the mature brain, platelet 5-HT levels should be actively maintained, because the half-life of platelets is only a few days [37,38]. This suggests that the factor that causes the platelet hyperserotonemia continues to be functionally active years after birth.
The statistical distribution of platelet 5-HT levels in normal and autistic groups has certain characteristic features [31], but only recent studies have attempted to describe them in detail [39,40]. These distributions are likely to contain information about the underlying processes controlling platelet 5-HT levels and, therefore, may help identify the factor that causes the platelet hyperserotonemia of autism. This same biological factor may be active during brain development (not necessarily in the same role), but there its identity may be obscured by the final complexity of a several-year-old autistic brain (Fig. 1). In the present study, published distributions of blood 5-HT levels are analyzed by a novel, quantitative approach that may help trace early, experimentally undetectable brain abnormalities leading to autism.
Figure 1 A biological factor that causes autism may have a dual function. A factor that causes autism (shown in red) may be expressed (1) in the CNS, where it plays a role in the early development of the brain, and (2) outside the CNS, where it participates in processes that determine the 5-HT levels in blood platelets. The "central" and "peripheral" 5-HT systems are separated by the blood-brain barrier (BBB) that matures after birth. It is usually not clear until age 2 or 3 whether the brain is autistic (black box). By that time, the factor has altered numerous developmental processes in the brain and may no longer be obvious. This same factor continues to operate years after birth outside the CNS, where it maintains higher than normal 5-HT levels in blood platelets. In contrast to the brain, blood platelets are short-lived and continue to be produced throughout the life span.
Results
Basic model
The origin of the platelet hyperserotonemia of autism cannot be understood unless a certain model of the underlying physiological processes is accepted – whether it is an implicit model that is not clearly stated, a model described in words, or a mathematical model. One advantage of mathematical modeling is that it requires a clear description of all relevant interactions among the components of the system. Its greatest disadvantage is that sometimes clear-cut choices have to be made where experimental data may suggest a few possible alternatives. In this section I introduce a model that is based on what is known about the 5-HT circulation outside the CNS and point out two important but unresolved problems.
In search of a factor that can both cause platelet hyperserotonemia and alter normal brain function, many recent studies have focused on the serotonin transporter (SERT) that is expressed in blood platelets and brain neurons [41]. Despite early promising results [42], different groups have found little or no linkage [43] between SERT polymorphisms and autism in various ethnic groups [40,44-47]. I have recently proposed [48] that the factor that interferes with brain development in autism may also regulate the release of 5-HT from gut enterochromaffin (EC) cells, the main source of blood 5-HT [36,49,50]. First, this hypothesis assumes that EC cells can monitor (directly or by way of gastrointestinal neurons) the 5-HT levels in the surrounding extracellular space and can decrease or increase their 5-HT release accordingly. Similar control mechanisms have long been suspected in the brain, where serotonergic neurons express 5-HT autoreceptors [51,52]. Second, the levels of extracellular 5-HT in the gut wall are assumed to be at equilibrium with the levels of free 5-HT in the arterial blood. While the baseline extracellular levels of 5-HT in the gut wall have not been precisely measured, the estimated levels of free 5-HT in the arterial blood appear to be comparable to the extracellular 5-HT levels in the brain [51,53], which expresses some of the same 5-HT receptors as the gut [51,54-57].
This hypothesis can be cast in a mathematical form. Suppose that EC cells indirectly monitor the levels of free 5-HT that arrives in the gut with the arterial blood, compare these levels with the expected 5-HT levels, and adjust their 5-HT release to a new value (Rn+1), using a pre-set release value (RC) as the reference point. The strength (gain) of this adjustment is controlled by a factor α, which is hypothesized to be different in normal and autistic individuals. After the blood leaves the gut, a large proportion (γ) of the free 5-HT is quickly removed by the liver, lungs and other organs that express SERT and monoamine oxidases (MAOs) [58-62]. The numerical value of γ is likely to vary from individual to individual, because the SERT and MAO genes have a number of polymorphic variants distributed in the population [40,45,46,63-66]. Therefore, γ is considered to be a random variable with a known probability distribution. The model can then be described by the following system of equations:
Fn + 1 = (1 - γ)Fn + Rn + 1, (2)
Where (1 - γ)Fn is the flux of free 5-HT that enters the gut with the arterial blood, FC is the pre-set ("expected") flux, and Fn + 1 is the flux of free 5-HT that exits the gut (α ≥ 0, 0 ≤ γ ≤ 1, FC > 0, RC > 0). In the model, the 5-HT release from EC cells does not include the 5-HT that is used for local signaling and is rapidly removed by local gastrointestinal epithelial and neural cells expressing SERT [54,67,68]. This 5-HT could be included in the model, together with the local clearance rate, if estimates of these parameters were available.
It is thought that little free 5-HT is taken up by blood platelets, before most of it is removed by the liver, lungs and other organs [53,60]. Also, it has been suggested that platelet 5-HT levels may depend on the levels of free 5-HT in the blood almost linearly [53]. Then, at the steady state, Fn + 1 = Fn ≡ F and Rn + 1 = Rn ≡ R for any n, and platelet 5-HT levels are
where K > 0 is a constant.
Note that ser(α, γ) is a decreasing function of γ. Also, at the steady state,
R = γF. (4)
It should be emphasized that the mathematical simplicity of equations (1) and (2) in no way implies that the biological regulation of 5-HT release in the gut is simple. The human gut is a remarkably complex organ that uses a wide range of neurotransmitters and that may have at least as many neurons as the spinal cord [50]. Nevertheless, recent studies suggest that complex biological systems, such as brain neurons, can be "actively linear" [69], meaning that sophisticated biological mechanisms may act on intrinsically non-linear physical processes to produce quantitative relationships that are mathematically linear.
The dependence of platelet 5-HT levels on α and γ is plotted in Figure 2, where the numerical values of FC and RC are taken from previously published experimental and theoretical studies [48,53,70], and where the regulation of the 5-HT release from EC cells is assumed to be less than fully functional in autistic individuals (note the low α value). A key feature of this dependence is that, in normal individuals, platelet 5-HT levels remain low with any γ, whereas in autistic individuals these levels may be normal or higher than normal depending on the individual's γ. This dependence captures one of the most puzzling properties of the autistic distribution of platelet 5-HT levels, which always overlaps with the control (normal) distribution, but always includes individuals whose 5-HT levels are higher than normal [31]. It may also explain why the SERT and MAO genes may appear to be linked with autism but may not actually cause it. As shown in Figure 2, a low γ is a necessary but not sufficient condition for the platelet hyperserotonemia to occur. Given a low γ, the platelet hyperserotonemia will occur only in those individuals whose regulation of the 5-HT release from EC cells is compromised (i.e., they are autistic and have a low α). It follows then that γ acts only as a modifier of platelet 5-HT levels, and that the statistical distribution of γ may be the same in normal and autistic populations. Assuming an individual's γ value is determined, at least in part, by his/her variants of the SERT and MAO genes expressed in the liver, lungs and other organs, normal and autistic populations may have similar distributions of SERT and MAO polymorphisms. This assumption is supported by recent studies [40,45-47,63,64].
Figure 2 Platelet levels as a function of α and γ. Platelet 5-HT levels, ser(α, γ), plotted as a function of α (the factor regulating 5-HT release from EC cells) and γ (the rate of 5-HT clearance by the liver, lungs, and other organs). This relationship is described by equation (3), where K is a constant. Note that if α is normal (high), platelet 5-HT levels stay low with any γ, but if α is autistic (low), individuals with a low γ become hyperserotonemic. The black circles mark the points whose coordinates are independent of α and are γ* = RC/(RC + FC) and ser(α, γ*) = KFC. Note in equations (1) and (2) that R = RC if and only if γ = γ*, so the distribution of γ is likely to contain γ*. This guarantees that the distributions of the 5-HT levels in normal autistic groups will always overlap, as observed in clinical studies. For illustrative purposes, the normal and autistic values of α were arbitrarily set at 0.20 and 0.02, respectively. These are realistic values, as follows in the text. The other parameter values were taken from published studies [48, 53, 70] and were FC = 210 ng/min and RC = 3000 ng/min.
Two potentially contentious decisions were made in the model. First, the exact levels of free 5-HT in the blood remain a debated issue. While a number of studies have found "low" but consistently measurable levels of free 5-HT in the human blood [53,70,71], Chen et al. [72] have suggested that the concentration of free 5-HT in the blood may be negligible, since these researchers have detected virtually no 5-HT in the whole blood of SERT-deficient mice whose blood platelets cannot take up 5-HT. Second, the model assumes that virtually all of the 5-HT stored in blood platelets is taken up by them after the lungs, liver, and other organs have cleared a large proportion of the 5-HT released by the gut. While evidence exists this may be the case [53,60], not all researchers agree. One could conceivably take into account both of these views by setting
ser(α, γ) ≡ K1 F + K2(1 - γ)F
or, in a more general form,
where K1, K2 ≥ 0 are constants and K(ω) is a function. However, this would require more detailed information about the dynamics of the 5-HT uptake by platelets, which is not currently available [31].
Distributions generated by the model
While the model (Fig. 2) appears to capture some of the key characteristics of the reported platelet 5-HT levels, it remains unclear whether it would produce similar results if α and γ took on other numerical values. The regulation of the 5-HT release in EC cells is poorly understood and no experimental estimates for the parameter α are available. Is it actually lower in autistic individuals? Likewise, how reasonable is it to suppose that the distribution of γ is the same in normal and autistic groups? Importantly, would the model produce consistent numerical values of parameters if different experimental studies were used?
To answer these questions, one may consider the basic framework of the model to be correct, but make no a priori assumptions about the values of the parameters (with the exception of those that are experimentally known) or about their differences in normal and autistic individuals. Then the unknown parameters of the model may be allowed to vary in the numerical space until the statistical distributions of 5-HT levels produced by the model closely match those reported in actual clinical studies. In order to be able to do this, one first has to find the theoretical statistical distributions of platelet 5-HT levels produced by the model.
The exact population distribution of γ is unknown, but its mean value is likely to be close to one [60]. Since SERT gene polymorphisms may occur with comparable frequencies [73], the statistical distribution of γ in a population can be approximated by a continuous uniform distribution on the interval [a, b] with the probability density function
It can be shown from equations (3) and (5) that the probability density function of platelet 5-HT levels then is
The theoretical population mean μser(α, a, b) and variance (α, a, b) of platelet 5-HT levels follow immediately:
and
where U ≡ FC - RCα.
The standard deviation of platelet 5-HT levels in the population then is
Distributions reported in clinical studies
Mean values of normal and autistic blood 5-HT levels have been reported and discussed in numerous publications [28-33]. In contrast, the precise statistical distributions of the platelet 5-HT levels in normal and autistic groups, such as their histograms (which roughly approximate their theoretical probability density functions), have so far attracted little attention. Only a few recent reports have presented more detail about the shape of these distributions. These reports are used in the following analysis:
(i) Mulder et al. [39] is recent and perhaps the most reliable report to date. It has used a relatively large sample of subjects whose platelet 5-HT levels are presented in histograms. The authors of this report are well-established researchers of blood 5-HT and autism. One of the co-authors, G.M. Anderson, has had numerous publications on the subject over the past several decades.
(ii) Coutinho et al. [40] have studied a large sample of subjects and presented their 5-HT levels in histograms, also explicitly listing their minimum and maximum values. However, their reported mean 5-HT levels are somewhat low, and the autistic 5-HT levels are higher than, but not significantly different from, the normal 5-HT levels.
(iii) McBride et al. [74] is a detailed report on the means and standard variations of platelet 5-HT levels in ethnically different groups, but the data are not presented in histogram form. Here, the minimum and maximum values of the distributions are recovered from their Figure 1, and the pooled means of the pre-pubertal children are recalculated from their Table 2.
Table 2 Predicted and observed ranges, means (<ser>), and standard deviations (SD) of platelet 5-HT levels, ser(α, γ). The distribution of γ was assumed to be continuously uniform; the theoretical SD values given in the table can be further improved by assuming that γ has a beta distribution or a normal distribution (see the text). Note that, strictly speaking, the model's <ser > and SD are precise theoretical expectations and standard deviations and, therefore, the notation μser (α, a, b) and σser (α, a, b) would be more accurate (but less convenient here).
Mulder et al. [39] (nmol/109 platelets) Coutinho et al. [40] (ng/109 platelets) McBride et al. [74] (ng/ml)
Model Observed Model Observed Model Observed
Minnormal 1.42 0.67 0 66 75 85
Maxnormal 5.57 5.67 598 676 417 449
<ser>normal 3.66 3.58 320 260 252 230
SDnormal 1.19 1.08 172 137 99 -
Minautistic 1.37 2.33 0 50 73 120
Maxautistic 8.18 8.33 925 1125 546 567
<ser>autistic 4.58 4.51 414 304 294 287
SDautistic 1.96 1.61 265 207 136 -
It is important to note that these reports are the only ones presently available and, therefore, no selection bias was introduced by choosing them for the present study.
Finding α and [a, b] from clinical data
In order to be able to compare the model's predictions with actual clinical reports, the numerical output of the model has to be scaled to the units of the used experimental studies. This scaling can be done by adjusting the parameter K in equation (3). The studies have reported the following means of the blood 5-HT levels in their normal groups: 3.58 nmol/109 platelets [39], 260 ng/109 platelets [40], and 230 ng/ml [74]. The last number was obtained by pooling the reported pre-pubertal means of the three ethnic groups. Assuming the flux of free 5-HT to the gut is around 210 ng/min in normal individuals [48,53,70], it follows from equation (3) that
where <...> denotes experimentally obtained means. Now we can calculate the approximate K values for each of the studies by dividing their reported mean 5-HT levels by the approximate flux of free 5-HT to the gut. This yields the following K values for the reports of Mulder et al. [39], Coutinho et al. [40] and McBride et al. [74], respectively: 0.0170 (nmol min ng-1 10-9 platelets), 1.2381 (min 10-9 platelets), and 1.0952 (min ml-1).
Next, we try to find such numerical values of [a, b], αnormal, and αautistic, that they minimize the difference between the predicted and observed levels of blood 5-HT. Suppose that the observed levels of blood 5-HT vary from MinOBS to MaxOBS and that the observed mean of blood 5-HT is <ser>OBS. The following error function can then be constructed:
where
and i = normal, autistic.
Note that, compared with the mismatch between the predicted and observed ranges of the distributions, the mismatch between the predicted and observed means is penalized "twice as much", because observed means are likely to be more accurate than observed minimal and maximal values.
This error function was numerically minimized by using the standard Nelder-Mead (downhill simplex) and differential evolution methods [75] implemented in Mathematica's NMinimize function (Wolfram Research, Inc.). Since the values of RC and FC may be approximated from published studies but are not necessarily accurate, RC was centered at 3000 ng/min based on a published estimate [53] and was allowed to vary ± 33%, whereas the value of FC was centered at 210 ng/min based on published estimates [48,53,70] and was allowed to vary ± 50% (more variation was allowed for FC because less is known about its actual value). No constraints were set for the interval [a, b] (i.e., 0 ≤ a <b ≤ 1). The variables αnormal and αautistic were allowed to vary from 0 to 5 and no a priori assumptions were made about their relative values (i.e., both αnormal >αautistic and αnormal ≤ αautistic were allowed). It can be shown that the system (equations (1) and (2)) is stable if 0≤α<FC(2 - γ)/[RC(1 - γ)]. Since the system should be stable for any γ ∈[a, b] and [a, b] is likely to contain the point γ ≈ 0.99 [60] or γ ≈ 0.93 [48], choosing α between 0 and 5 allows the optimization procedure to use virtually any value of α where the system maintains stability.
The numerical values of the model's parameters (αnormal, αautistic, [a, b], FC, and RC) that minimized the error function are given in Table 1. Note that all three clinical studies yielded similar sets of values. Most importantly, the minimization algorithms yielded the best match between the model and the clinical reports when αautistic was virtually zero.
Table 1 Estimates of FC, RC, a, b, αnormal, and αautistic, obtained by numerical minimization of the error function.
Data source K FC RC a b αnormal αautistic
Mulder et al. [39] 0.0170 105 2000 0.8060 0.9612 0.1510 0.0000
Coutinho et al. [40] 1.2381 105 2000 0.7280 1.0000 0.0981 0.0000
McBride et al. [74] 1.0952 105 2000 0.8006 0.9678 0.0895 0.0000
By plugging these obtained values of the parameters into equations (12), (13), (14) and (9), one can obtain the values of 5-HT levels predicted by the model and compare them with the actual observed levels. As shown in Table 2, the predicted values closely match the values observed in Mulder et al. [39] and McBride et al. [74]. The largest mismatch was between the predicted and observed minimal values. The model predicted slightly higher mean 5-HT levels for Coutinho et al. [40] than were actually observed; interestingly, Coutinho et al. [40] have in fact reported unusually low platelet 5-HT levels.
Distribution of γ can be approximated by beta and normal distributions
One advantage of choosing the uniform distribution to represent γ is that it simplifies calculations and allows finding the exact formulae for means and standard deviations. However, the model tends to overestimate the standard deviations of platelet 5-HT levels (Table 2), because in the uniform distribution even extreme γ values occur with same probability as all others. Instead of approximating the distribution of γ as uniform, one may want a distribution of which the probability density function drops off more smoothly near the minimal and maximal values. This can be achieved by replacing the uniform distribution of γ with the beta distribution, the uniform distribution being its special case [76]. The following deals with mathematical technicalities of this replacement. Non-mathematically inclined readers may skip them and go immediately to Figures 4 and 5 referred to at the end of this section.
Figure 4 Model replicates published data. A, B, The model's simulation of Mulder et al.'s sampling [39], assuming γ has the beta distribution on the interval [0.8060, 0.9612] with both shape parameters equal to 1.5. The platelet 5-HT levels were calculated by using equation (3), with the values of K, FC, RC, αnormal and αautistic taken from Table 1. C, D, The actual data from Mulder et al. [39] (reprinted by permission from Lippincott Williams & Wilkins, modified). In the simulated and actual sampling, 60 normal and 33 autistic subjects were used. Note that the exact appearance of the histograms will vary from sampling to sampling due to the small number of cases in each bin.
Figure 5 Model predicts the shape of the normal and autistic distributions of platelet 5-HT levels. Histograms obtained by simulating a sampling of a very large number of normal and autistic individuals (a million subjects in each group). The distribution of γ was assumed to be (A, B) the beta distribution on the interval [0.8060; 0.9612] with both shape parameters equal to 1.5 (see the text); or (C, D) the normal (Gaussian) distribution with mean 0.8836 (the midpoint of the interval [0.8060; 0.9612]) and standard deviation 0.04 (see the text). The platelet 5-HT levels were calculated by using equation (3), with the values of K, FC, RC, αnormal and αautistic taken from Table 1. In a very large sampling, the number of cases in each histogram bin closely approximates the number of cases predicted by the exact probability distribution functions. The Chi-square test confirmed that the normal and autistic distributions predicted by the model may underlie the distributions reported by Mulder et al. (2004). The following goodness-of-fit results were obtained: = 12.38 (P = 0.26) and = 11.29 (P = 0.19) for the normal and autistic groups, respectively, if γ had the beta distribution; and = 13.36 (P = 0.27) and = 12.21 (P = 0.14) for the normal and autistic groups, respectively, if γ had the normal distribution (bins were pooled if theoretical bins had fewer than 3 cases). It is important that both the normal and autistic distributions had the same underlying distribution of γ and that only one parameter, α, was needed to switch from the normal distribution to the autistic distribution. Also, compare the histograms in C and D, based on the data of Mulder et al. [39], with those in Figure 1 of Coutinho et al. [40].
Note that if the obtained parameter values (Table 1) are plugged into equation (3), the normal and autistic platelet 5-HT levels turn out to depend on γ almost linearly (Fig. 3). This allows "warping" the uniform distribution of γ into a symmetric beta distribution on the same interval, with little effect on the theoretical mean values of ser(α, γ). Suppose that γ has a symmetric beta distribution on [a, b], whose shape is determined by the parameters m and n, such that m = n (if m = n = 1, the beta distribution becomes the uniform distribution). We can use a Taylor series to formally linearize ser(α, γ) around γ0 = (a + b)/2 as ser(α, γ) ≈ ser(α, γ0) - λ (γ - γ0) ≡ serL(α, γ),
Figure 3 Platelet levels plotted with the parameter values derived from published studies. Platelet 5-HT levels as functions of γ for the data of Mulder et al. [39], Coutinho et al. [40] and McBride et al. [74]. Equation (3) and the estimated parameter values from Table 1 were used. The arrowheads mark the predicted intervals of the γ distributions (Table 1). For comparison, the Y-axes were scaled proportionally to the K values of the three studies (Table 1).
Then, keeping in mind that γ has a beta distribution, the standard deviation of serL(α, γ) becomes
Since the values of λ, a, and b have already been estimated (Table 1), it is now possible to obtain the m values that yield such standard deviations of the linearized ser(α, γ) that they precisely match those reported in the clinical studies (Table 2). The following m values were obtained for the normal and autistic groups, respectively: 1.2940 and 1.7028 for the data of Mulder et al. [39]; and 1.8308 and 1.8748 for the data of Coutinho et al. [40]. Pooled standard variations were unavailable in McBride et al. [74]. We have earlier assumed that normal and autistic groups have the same γ distribution. Therefore, the actual m values can be approximated by 1.50 for Mulder et al. [39] and 1.85 for Coutinho et al. [40].
Likewise, γ can be assumed to have a normal distribution with mean (a + b)/2 and standard deviation σ. Then the standard deviation of serL(α, γ) becomes
σserL(α, a, b, σ) = λσ, (17)
where λ is the same as in equation (15), and we obtain the following σ values for the normal and autistic groups, respectively: 0.0410 and 0.0370 for the data of Mulder et al. [39]; and 0.0630 and 0.0624 for the data of Coutinho et al. [40]. Therefore the actual σ values can be approximated by 0.04 for Mulder et al. [39] and 0.06 for Coutinho et al. [40].
The model now easily generates "normal" and "autistic" samples of platelet 5-HT levels that closely match the actual reported data (Fig. 4). Most importantly, the switch from the normal distribution to the autistic distribution requires changing only one parameter, α.
It is not known what normal and autistic distributions would look like if one could sample a very large number of subjects. The model can predict the shape of these distributions by simulating such large sampling (Fig. 5).
Is the 5-HT synthesis rate altered in autism?
One of the most important questions in autism research is whether the rate of 5-HT synthesis is altered in the brain and gut of autistic individuals. If 5-HT synthesis is altered in the autistic brain, as some studies have suggested [77-79], this potentially may have a great impact on brain development [80,81] (but caution should be exercised in predicting the extent of these alterations [82]).
The brain 5-HT and the gut 5-HT are synthesized by two different tryptophan hydroxylases [49] that, at least in humans, have different properties and are regulated differently [83]. While the biological factor underlying the parameter α of the model is hypothesized to play a role in the developing brain (Fig. 1), the model makes no assumptions about its exact function in the brain. In the brain, it may not regulate 5-HT release from serotonergic neurons and may have a different function (see, for example, Figure 4 of [48]). Therefore, this section focuses only on the 5-HT synthesis and release in the gut.
It is important to note that the model says nothing about the rate of 5-HT synthesis in the gut and rather deals with the rate of 5-HT release from the gut. However, most clinical and experimental studies make no such distinction and, therefore, their relevance to the model is discussed assuming higher 5-HT synthesis rates do lead to higher 5-HT release rates.
It follows from equations (3) and (4) that, at the steady state,
and that this relationship is independent of γ. This means that if one were to sample any group of individuals and could measure their platelet 5-HT levels and gut 5-HT release rates precisely, the correlation coefficient between these two variables would always be minus one, irrespective of the distribution of γ. In other words, equation (18) predicts that individuals with higher platelet 5-HT levels should have lower 5-HT release rates.
How can lower 5-HT release rates lead to higher platelet 5-HT levels? Note that, in the model, both the platelet 5-HT levels and the 5-HT release rate are dynamically linked through the 5-HT clearance rate, γ. As γ grows lower, less 5-HT is removed from the system and more of 5-HT is accumulated in blood platelets. At the same time, these higher 5-HT levels drive down the 5-HT release rate in the gut, as required by equation (1).
Still, it appears that the results of clinical studies are inconsistent with equation (18). Three important findings should be noted:
(i) Minderaa et al. [36] have found no significant correlation between whole blood 5-HT levels and 5-HT synthesis in the gut, measured as the production of urinary 5-HIAA [36]. Similar results have been obtained by Launay et al. [84] and other groups (reviewed in [31]).
(ii) Croonenberghs et al. [85] have shown that the 5-HT synthesis in the gut of autistic individuals may be higher than that in normal individuals, at least when subjects are administered 5-hydroxytryptophan (5-HTP), an immediate precursor of 5-HT.
(iii) Carcinoid tumors, derived from gut EC cells, may result in excessive synthesis and release of 5-HT, which in turn may lead to elevated platelet 5-HT levels [86].
A more careful analysis reveals that these findings are not only consistent with the model, but that the model can reconcile some of the apparent contradictions among them:
(i) It follows from the model that the measured correlation between platelet 5-HT levels and 5-HT release rates should be close to zero in autistic groups, even though equation (18) holds.
In fact, we can rewrite equation (18) as
Now consider two random variables, η and ξ, that are linearly dependent such that
η = wξ + q, (20)
where w and q are constants. It follows from equation (20) that the correlation between them is either -1 or 1, depending on the sign of w.
Denote the means of these variables μη and μξ, respectively, and their standard deviations ση and σξ, respectively. Suppose next that the errors of measurement of η and ξ are independent random variables εη and εξ, such that their expected values are zero and standard deviations are δη and δξ, respectively. Note that experimentally we can measure only η* = η + εη and ξ* = ξ + εξ. The expected values of η* and ξ* are the same as those of η and ξ. However, the theoretical correlation coefficient between η* and ξ* now becomes
If the standard deviations of the errors of measurement are small, we obtain ρ(η*, ξ*) ≈ ± 1, as expected from equation (20).
Now we return to equation (19). Any experimental measurement of R (5-HT release) and ser(α, γ) (platelet 5-HT levels) will contain a measurement error. Denoting these measured values ser*(α, γ) and R*, one obtains from equations (19), (20), and (21) that the correlation coefficient between R* and ser*(α, γ) is
where
w = -(αRC)/(KFC), (23)
σser > 0 is the standard deviation of ser(α, γ), and δR > 0 and δser > 0 are the standard deviations of the errors of measurement of R and ser(α, γ), respectively. The estimated values of K, FC, RC, and α can be obtained from Table 1 and the values of σser from Table 2 or from the original published data.
Consider now an autistic group whose α → 0 (Table 1). Then, from equation (23), w → 0, and it follows from equation (22) that .
(ii) Croonenberghs et al. [85] have recently shown that oral administration of 5-hydroxytryptophan (5-HTP) leads to higher platelet 5-HT levels in autistic patients, and the authors have suggested that the 5-HT synthesis rate may be higher in the gut of autistic subjects compared with normal subjects.
Suppose that the administered 5-HTP is converted to 5-HT at the same rate in both normal and autistic groups. It is likely that the exogenous influx of 5-HTP results in a comparable exogenous influx of 5-HT, because the rate-limiting step in the synthesis of 5-HT is not the 5-HTP conversion to 5-HT, but rather the tryptophan conversion to 5-HTP [87].
Notice that the system is not in its steady state during the experiment and, therefore, we have to use equations (1) and (2), which now should contain the exogenous source of 5-HT. It is straightforward to see that the system then becomes
Fn + 1 = (1 - γ)Fn + Rn + 1 + REX, (25)
where REX is the exogenous flux of 5-HT.
Solving equations (24) and (25) step-by-step essentially replicates the major finding of Croonenberghs et al. [85] (Fig. 6). However, the model predicts that the higher blood 5-HT levels in autistic subjects are not due to a higher 5-HT synthesis rate, but rather to the failure of their gut to decrease the release of endogenous 5-HT in response to the high concentration of 5-HT caused by the administration of 5-HTP.
Figure 6 An exogenous source of 5-HT elevates platelet 5-HT levels in an autistic group more than in a normal group. For the simulation, the initial values of platelet 5-HT levels (ser(α, γ)) and 5-HT release rate (R) were set at zero and the system developed according to equations (1) and (2). After the system reached its steady state, an exogenous 5-HT source was "turned on" (+REX) and the system developed according to equations (24) and (25). After 5 steps, the exogenous 5-HT source was "turned off" (-REX) and the system developed according to equations (1) and (2) until it returned to its steady state. Each point is the mean of 10,000 simulated individuals whose γ had the beta distribution on the interval [0.8060, 0.9612] (see Table 1) with both shape parameters equal to 1.5 (see the text). Individual plots (not shown) looked essentially the same as the mean plots. The ratio between the autistic and normal platelet 5-HT levels (A) at step 7 (at the steady state) is 1.25 and the same ratio at step 13 is 1.35. The numerical values of the parameters were K = 0.0170, FC = 105, RC = 2000, αnormal = 0.1510, αautistic = 0.0000 (Table 1) and REX = 4000. Compare these plots with Figure 1 of Croonenberghs et al. [85].
(iii) In the case of carcinoid tumors, abnormally large amounts of 5-HT may be released into the blood. It is likely that the normal mechanisms regulating 5-HT release are compromised or absent in carcinoid tumors. Then instead of equations (1) and (2) one can consider only one equation (2), which can be rewritten as
Fn + 1 = (1 - γ)Fn + RCARCINOID, (26)
where RCARCINOID is large and relatively constant. Then, at the steady state,
F = RCARCINOID/γ
and
It is obvious that in this abnormal case higher 5-HT release rates will lead to higher platelet 5-HT levels, as reported by Kema et al. [86].
Discussion
The presented model is based on the hypothesis that at least one factor that interferes with normal brain development in autism also participates in the regulation of 5-HT release from enterochromaffin cells. When applied to the data of three published studies, the model predicts that this factor is virtually non-functional in autistic individuals (Table 1).
Before the biological nature of this factor is discussed, it should be noted that the parameter values obtained for each of the three published studies were virtually the same (Table 1). This underlying consistency of the data is not trivial, since Mulder et al. [39] have suggested that their autistic distribution may be bimodal and thus qualitatively different from the control (normal) distribution, whereas Coutinho et al. [40] have reported a clearly unimodal autistic distribution that so overlapped with the control distribution that their means were not statistically significant. It should also be noted that initially γ was allowed to vary from zero to one, but the numerical optimization based on the published data narrowed this range down to approximately 0.8 – 1.0 (Table 1). This agrees well with actual experimental data. An early study has approximated the dog's γ as 0.99 and shown that the 5-HT clearance by the lungs varies from 0.80 to 0.98 [60]. The mean human γ may be somewhat smaller, because the rate of 5-HT release by gut enterochromaffin cells has been predicted to be around 3000 ng/min [53] and the arterial flow of free 5-HT has been estimated to be around 210 ng/min [48,53,70]. This suggests that, in humans, approximately 93% of free 5-HT is cleared in one circulation and, therefore, the value of γ is close to 0.93. The model predicted similar γ distributions in normal and autistic groups, supporting the hypothesis that the frequencies of SERT and MAO polymorphisms in normal and autistic groups may be the same.
The most significant result is that the factor that regulates 5-HT release from EC cells (represented by the parameter α) appears to be virtually non-functional in autistic individuals (Table 1). What is the biological nature of α? Evidence suggests that EC cells may express 5-HT3, 5-HT4 and 5-HT1A receptors [55,88-90] and that they may also express 5-HT2 receptors [89]. Some of these receptors appear to be involved in the autoregulation of 5-HT release [89,90]. While one report has failed to find 5-HT3 and 5-HT4 receptor mRNAs in cultured EC cells [91], the regulation of 5-HT release from EC cells may also be indirect, by way of enteric neurons. These neurons are known to express various 5-HT receptors [54,55,92,93] and can control 5-HT release from EC cells by acting on their cholinergic and other receptors [88,94-96].
The model is based on a negative feedback loop. It has been shown that such negative feedback may be mediated by 5-HT4 receptors expressed by EC cells and that this negative feedback appears to dominate over the positive feedback mediated by 5-HT3 receptors [89,90]. A recent study has suggested that under normal circumstances (as opposed to conditions such as carcinoid tumors) the concentration of endogenous 5-HT may not be high enough to activate 5-HT4 receptors and alter the 5-HT release from EC cells [89]. At least superficially, this mirrors recent findings in the brain, where 5-HT1A and 5-HT1B receptors, long assumed to act as autoreceptors, may not actually be activated by extracellular 5-HT unless its concentration reaches excessive levels [51]. Since precise measurements of 5-HT release in the gut and the brain are difficult, it is more likely that these receptors do control 5-HT release under normal circumstances, but that their effect on 5-HT release is more subtle than we expect. The model's small value of α appears to predict such subtle regulation.
Can 5-HT4 receptors be involved in autism? One agonist used to study the effects of 5-HT4 receptors on the 5-HT release from EC cells has been 5-methoxytryptamine (5-MT) [89,90], which has high affinity for these receptors [57]. While 5-MT has been reported to inhibit the 5-HT release from EC cells, subcutaneous 5-MT injections in pregnant rats produces pups with autistic-like symptoms [97] and subcutaneous 5-MT injections in pregnant mice may lead to an autistic-like disruption of cortical columns in the pups [11,81]. Normal brain development may be altered if brain 5-HT4 receptors are compromised, because these receptors appear to be expressed in the marginal zone of the adult human brain [98] and, therefore, may also be expressed in Cajal-Retzius cells of the developing brain. It has been recently shown that an abnormal serotonergic input to Cajal-Retzius cells during development may lead to autistic-like cortical abnormalities [81]. Interestingly, the expression of the 5-HT4 receptor is very low in the cerebral cortex of the guinea pig [99], suggesting that this receptor may play a specific role in the primate brain. Generally, we are only beginning to understand the role of the 5-HT4 in brain development, because the human 5-HT4 receptor gene consists of at least 38 exons and at least eight C-terminal splice variants of the human 5-HT4 receptor have been described [57].
Other 5-HT receptors, as well as other mechanisms, may be involved both in the regulation of 5-HT release from the gut and in brain development. For example, 5-HT1A and 5-HT2 receptors have been implicated in autism [31,100-102]. As already discussed, these receptors can also regulate the 5-HT release from EC cells. Moreover, 5-MT is a rather non-specific 5-HT receptor agonist [103] and appears to be co-localized with 5-HT in most brain neurons [104]. Therefore, some of its effects may be produced by its acting on a few types of 5-HT receptors at the same time, both in the gut and the brain.
The model assumes that the 5-HT clearance rate (γ) and the gain of 5-HT release (α) are independent. Generally, the expression of neurotransmitter receptors or their sensitivity can dynamically change depending on the availability of the neurotransmitter. For example, gut 5-HT3 receptors undergo structural and functional changes in SERT-knockout mice [105] and 5-HT1A receptors in the human brain have different affinities in individuals with different SERT polymorphic variants [106]. These and other related findings are likely to become indispensable for understanding the platelet hyperserotonemia of autism; unfortunately, too little information is currently available for quantitative modeling of these relationships.
Intriguingly, α may be represented by biological mechanisms other than 5-HT receptors. For example, adenosine and ATP may modulate the 5-HT release from human EC cells [107,108] and ATP also activates microglia in the brain [109]. A study, called by some researchers "the most important postmortem study of autism to date" [110], has found an abnormal activation of microglia in autistic brains [111].
It should be noted in conclusion that the mathematical framework of the model allows it to be modified so that it no longer depends on free 5-HT in the blood. In fact, one could conceivably build a model where 5-HT is released by EC cells, cleared by SERT-expressing cells locally, and where the remaining extracellular 5-HT acts on the mechanisms controlling 5-HT release from EC cells, without leaving the gut. Assuming γ now denotes the local clearance and α is the gain of the 5-HT release, one again may arrive at a system of equations similar to equations (1) and (2).
Conclusion
The origin of autism is as much a conceptual problem as it is experimental. The theoretical approach introduced here brings together information on the "central" and "peripheral" 5-HT and offers new insights into early abnormalities of the developing autistic brain that may otherwise escape direct experimental detection.
Methods
All symbolic and numerical calculations were done in Mathematica 5.0.0, 5.0.1, 5.1.0, or 5.1.1 (Wolfram Research, Inc.). Where the numerical minimization of the error function produced different sets of numerical values in different releases of Mathematica, the values that yielded the smallest error were used (for the purpose of this study, Mathematica 5.1.1 was superior to the earlier releases). The figures were generated in Mathematica and prepared for publication in Adobe Illustrator 10 or CS (Adobe Systems, Inc.).
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
SJ conceived of and carried out the presented study.
Acknowledgements
I thank Dr. P. Rakic and the National Alliance for Autism Research (NAAR) for their financial support, the anonymous reviewers for their valuable suggestions, and Dr. G.M. Anderson, Dr. A.E. Ayoub and Michael Fischer for their comments on the revised manuscript. I also thank Vaiva, my inspiration.
==== Refs
Williams RS Hauser SL Purpura DP DeLong GR Swisher CN Autism and mental retardation: neuropathologic studies performed in four retarded persons with autistic behavior Arch Neurol 1980 37 749 753 7447762
Ritvo ER Freeman BJ Scheibel AB Duong T Robinson H Guthrie D Ritvo A Lower Purkinje cell counts in the cerebella of four autistic subjects: initial findings of the UCLA-NSAC Autopsy Research Report Am J Psychiatry 1986 143 862 866 3717426
Kern JK Purkinje cell vulnerability and autism: a possible etiological connection Brain Dev 2003 25 377 382 12907269 10.1016/S0387-7604(03)00056-1
Woodhouse W Bailey A Rutter M Bolton P Baird G Le CA Head circumference in autism and other pervasive developmental disorders J Child Psychol Psychiatry 1996 37 665 671 8894947
Courchesne E Karns CM Davis HR Ziccardi R Carper RA Tigue ZD Chisum HJ Moses P Pierce K Lord C Lincoln AJ Pizzo S Schreibman L Haas RH Akshoomoff NA Courchesne RY Unusual brain growth patterns in early life in patients with autistic disorder: an MRI study Neurology 2001 57 245 254 11468308
Aylward EH Minshew NJ Field K Sparks BF Singh N Effects of age on brain volume and head circumference in autism Neurology 2002 59 175 183 12136053
Courchesne E Carper R Akshoomoff N Evidence of brain overgrowth in the first year of life in autism JAMA 2003 290 337 344 12865374 10.1001/jama.290.3.337
Carper RA Moses P Tigue ZD Courchesne E Cerebral lobes in autism: early hyperplasia and abnormal age effects Neuroimage 2002 16 1038 1051 12202091 10.1006/nimg.2002.1099
Carper RA Courchesne E Inverse correlation between frontal lobe and cerebellum sizes in children with autism Brain 2000 123 ( Pt 4) 836 844 10734014 10.1093/brain/123.4.836
Casanova MF Buxhoeveden DP Brown C Clinical and macroscopic correlates of minicolumnar pathology in autism J Child Neurol 2002 17 692 695 12503647
Casanova MF Buxhoeveden DP Switala AE Roy E Minicolumnar pathology in autism Neurology 2002 58 428 432 11839843
Casanova MF Buxhoeveden D Gomez J Disruption in the inhibitory architecture of the cell minicolumn: implications for autisim Neuroscientist 2003 9 496 507 14678582 10.1177/1073858403253552
Casanova MF White matter volume increase and minicolumns in autism Ann Neurol 2004 56 453 15349878 10.1002/ana.20196
Raymond GV Bauman ML Kemper TL Hippocampus in autism: a Golgi analysis Acta Neuropathol (Berl) 1996 91 117 119 8773156 10.1007/s004010050401
Blatt GJ Fitzgerald CM Guptill JT Booker AB Kemper TL Bauman ML Density and distribution of hippocampal neurotransmitter receptors in autism: an autoradiographic study J Autism Dev Disord 2001 31 537 543 11814263 10.1023/A:1013238809666
Kemper TL Bauman ML Neuropathology of infantile autism Mol Psychiatry 2002 7 Suppl 2 S12 S13 12142934 10.1038/sj.mp.4001165
Amaral DG Bauman MD Schumann CM The amygdala and autism: implications from non-human primate studies Genes Brain Behav 2003 2 295 302 14606694 10.1034/j.1601-183X.2003.00043.x
Bauman M Kemper TL Histoanatomic observations of the brain in early infantile autism Neurology 1985 35 866 874 4000488
Schumann CM Hamstra J Goodlin-Jones BL Lotspeich LJ Kwon H Buonocore MH Lammers CR Reiss AL Amaral DG The amygdala is enlarged in children but not adolescents with autism; the hippocampus is enlarged at all ages J Neurosci 2004 24 6392 6401 15254095 10.1523/JNEUROSCI.1297-04.2004
Rodier PM Ingram JL Tisdale B Nelson S Romano J Embryological origin for autism: developmental anomalies of the cranial nerve motor nuclei J Comp Neurol 1996 370 247 261 8808733 10.1002/(SICI)1096-9861(19960624)370:2<247::AID-CNE8>3.0.CO;2-2
Rodier PM Converging evidence for brain stem injury in autism Dev Psychopathol 2002 14 537 557 12349873 10.1017/S0954579402003085
Rodier PM The early origins of autism Sci Am 2000 282 56 63 10710787
Belmonte MK Allen G Beckel-Mitchener A Boulanger LM Carper RA Webb SJ Autism and abnormal development of brain connectivity J Neurosci 2004 24 9228 9231 15496656 10.1523/JNEUROSCI.3340-04.2004
Palmen SJ Van EH Hof PR Schmitz C Neuropathological findings in autism Brain 2004 15329353
Courchesne E Redcay E Kennedy DP The autistic brain: birth through adulthood Curr Opin Neurol 2004 17 489 496 15247547 10.1097/01.wco.0000137542.14610.b4
Chugani DC Serotonin in autism and pediatric epilepsies Ment Retard Dev Disabil Res Rev 2004 10 112 116 15362166 10.1002/mrdd.20021
Duchala CS Shick HE Garcia J Deweese DM Sun X Stewart VJ Macklin WB The toppler mouse: A novel mutant exhibiting loss of Purkinje cells J Comp Neurol 2004 476 113 129 15248193 10.1002/cne.20206
Schain RJ Freedman DX Studies on 5-hydroxyindole metabolism in autistic and other mentally retarded children J Pediatr 1961 58 315 320 13747230
Hanley HG Stahl SM Freedman DX Hyperserotonemia and amine metabolites in autistic and retarded children Arch Gen Psychiatry 1977 34 521 531 860890
Anderson GM Freedman DX Cohen DJ Volkmar FR Hoder EL McPhedran P Minderaa RB Hansen CR Young JG Whole blood serotonin in autistic and normal subjects J Child Psychol Psychiatry 1987 28 885 900 3436995
Anderson GM Horne WC Chatterjee D Cohen DJ The hyperserotonemia of autism Ann N Y Acad Sci 1990 600 331 340 2252319
Cook EHJ Brief report: pathophysiology of autism: neurochemistry J Autism Dev Disord 1996 26 221 225 8744489 10.1007/BF02172016
Anderson GM Genetics of childhood disorders: XLV. Autism, part 4: serotonin in autism J Am Acad Child Adolesc Psychiatry 2002 41 1513 1516 12447040 10.1097/00004583-200212000-00025
Wenzel D Felgenhauer K The development of the blood-CSF barrier after birth Neuropadiatrie 1976 7 175 181 60727
Virgintino D Errede M Robertson D Capobianco C Girolamo F Vimercati A Bertossi M Roncali L Immunolocalization of tight junction proteins in the adult and developing human brain Histochem Cell Biol 2004 122 51 59 15221411 10.1007/s00418-004-0665-1
Minderaa RB Anderson GM Volkmar FR Akkerhuis GW Cohen DJ Urinary 5-hydroxyindoleacetic acid and whole blood serotonin and tryptophan in autistic and normal subjects Biol Psychiatry 1987 22 933 940 2440483 10.1016/0006-3223(87)90002-3
Baker GR Sullam PM Levin J A simple, fluorescent method to internally label platelets suitable for physiological measurements Am J Hematol 1997 56 17 25 9298862 10.1002/(SICI)1096-8652(199709)56:1<17::AID-AJH4>3.0.CO;2-5
Heyssel RM Determination of human platelet survival utilizing C-14-labeled serotonin J Clin Invest 1961 40 2134 2142 13907152
Mulder EJ Anderson GM Kema IP De BA Van Lang ND Den Boer JA Minderaa RB Platelet Serotonin Levels in Pervasive Developmental Disorders and Mental Retardation: Diagnostic Group Differences, Within-Group Distribution, and Behavioral Correlates J Am Acad Child Adolesc Psychiatry 2004 43 491 499 15187810 10.1097/00004583-200404000-00016
Coutinho AM Oliveira G Morgadinho T Fesel C Macedo TR Bento C Marques C Ataide A Miguel T Borges L Vicente AM Variants of the serotonin transporter gene (SLC6A4) significantly contribute to hyperserotonemia in autism Mol Psychiatry 2004 9 264 271 15094787 10.1038/sj.mp.4001409
Lesch KP Wolozin BL Murphy DL Reiderer P Primary structure of the human platelet serotonin uptake site: identity with the brain serotonin transporter J Neurochem 1993 60 2319 2322 7684072
Cook EHJ Courchesne R Lord C Cox NJ Yan S Lincoln A Haas R Courchesne E Leventhal BL Evidence of linkage between the serotonin transporter and autistic disorder Mol Psychiatry 1997 2 247 250 9152989 10.1038/sj.mp.4000266
Spielman RS Ewens WJ The TDT and other family-based tests for linkage disequilibrium and association Am J Hum Genet 1996 59 983 989 8900224
McCauley JL Olson LM Dowd M Amin T Steele A Blakely RD Folstein SE Haines JL Sutcliffe JS Linkage and association analysis at the serotonin transporter (SLC6A4) locus in a rigid-compulsive subset of autism Am J Med Genet 2004 127B 104 112 10.1002/ajmg.b.20151
Persico AM Militerni R Bravaccio C Schneider C Melmed R Conciatori M Damiani V Baldi A Keller F Lack of association between serotonin transporter gene promoter variants and autistic disorder in two ethnically distinct samples Am J Med Genet 2000 96 123 127 10686565 10.1002/(SICI)1096-8628(20000207)96:1<123::AID-AJMG24>3.0.CO;2-N
Betancur C Corbex M Spielewoy C Philippe A Laplanche JL Launay JM Gillberg C Mouren-Simeoni MC Hamon M Giros B Nosten-Bertrand M Leboyer M Serotonin transporter gene polymorphisms and hyperserotonemia in autistic disorder Mol Psychiatry 2002 7 67 71 11803447 10.1038/sj.mp.4001923
Wu S Guo Y Jia M Ruan Y Shuang M Liu J Gong X Zhang Y Yang J Yang X Zhang D Lack of evidence for association between the serotonin transporter gene (SLC6A4) polymorphisms and autism in the Chinese trios Neurosci Lett 2005 381 1 5 15882779 10.1016/j.neulet.2005.01.073
Janusonis S Serotonergic paradoxes of autism replicated in a simple mathematical model Med Hypotheses 2005 64 742 750 15694691 10.1016/j.mehy.2004.09.020
Walther DJ Peter JU Bashammakh S Hortnagl H Voits M Fink H Bader M Synthesis of serotonin by a second tryptophan hydroxylase isoform Science 2003 299 76 12511643 10.1126/science.1078197
Gershon MD Review article: serotonin receptors and transporters -- roles in normal and abnormal gastrointestinal motility Aliment Pharmacol Ther 2004 20 Suppl 7 3 14 15521849 10.1111/j.1365-2036.2004.02180.x
Adell A Celada P Abellan MT Artigas F Origin and functional role of the extracellular serotonin in the midbrain raphe nuclei Brain Res Brain Res Rev 2002 39 154 180 12423765 10.1016/S0165-0173(02)00182-0
Jacobs BL Azmitia EC Structure and function of the brain serotonin system Physiol Rev 1992 72 165 229 1731370
Anderson GM Stevenson JM Cohen DJ Steady-state model for plasma free and platelet serotonin in man Life Sci 1987 41 1777 1785 3657383 10.1016/0024-3205(87)90696-5
Gershon MD Plasticity in serotonin control mechanisms in the gut Curr Opin Pharmacol 2003 3 600 607 14644011 10.1016/j.coph.2003.07.005
Kirchgessner AL Liu MT Raymond JR Gershon MD Identification of cells that express 5-hydroxytryptamine1A receptors in the nervous systems of the bowel and pancreas J Comp Neurol 1996 364 439 455 8820876 10.1002/(SICI)1096-9861(19960115)364:3<439::AID-CNE5>3.0.CO;2-5
Costall B Naylor RJ 5-HT3 receptors Curr Drug Targets CNS Neurol Disord 2004 3 27 37 14965242 10.2174/1568007043482624
Bockaert J Claeysen S Compan V Dumuis A 5-HT4 receptors Curr Drug Targets CNS Neurol Disord 2004 3 39 51 14965243 10.2174/1568007043482615
Ramamoorthy S Bauman AL Moore KR Han H Yang-Feng T Chang AS Ganapathy V Blakely RD Antidepressant- and cocaine-sensitive human serotonin transporter: molecular cloning, expression, and chromosomal localization Proc Natl Acad Sci U S A 1993 90 2542 2546 7681602
Billett EE Monoamine oxidase (MAO) in human peripheral tissues Neurotoxicology 2004 25 139 148 14697888 10.1016/S0161-813X(03)00094-9
Thomas DP Vane JR 5-hydroxytryptamine in the circulation of the dog Nature 1967 216 335 338 6053807
Shiue GG Choi SR Fang P Hou C Acton PD Cardi C Saffer JR Greenberg JH Karp JS Kung HF Shiue CY N,N-dimethyl-2-(2-amino-4-(18)F-fluorophenylthio)-benzylamine (4-(18)F-ADAM): an improved PET radioligand for serotonin transporters J Nucl Med 2003 44 1890 1897 14660713
Lin KJ Ye XX Yen TC Wey SP Tzen KY Ting G Hwang JJ Biodistribution study of [(123)I] ADAM in mice: correlation with whole body autoradiography Nucl Med Biol 2002 29 643 650 12234588 10.1016/S0969-8051(02)00323-2
Cohen IL Liu X Schutz C White BN Jenkins EC Brown WT Holden JJ Association of autism severity with a monoamine oxidase A functional polymorphism Clin Genet 2003 64 190 197 12919132 10.1034/j.1399-0004.2003.00115.x
Jones MB Palmour RM Zwaigenbaum L Szatmari P Modifier effects in autism at the MAO-A and DBH loci Am J Med Genet 2004 126B 58 65 10.1002/ajmg.b.20172
Filic V Vladic A Stefulj J Cicin-Sain L Balija M Sucic Z Jernej B Monoamine oxidases A and B gene polymorphisms in migraine patients J Neurol Sci 2005 228 149 153 15694196 10.1016/j.jns.2004.11.045
Soeby K Larsen SA Olsen L Rasmussen HB Werge T Serotonin transporter: Evolution and impact of polymorphic transcriptional regulation Am J Med Genet B Neuropsychiatr Genet 2005 15858819
Chen JX Pan H Rothman TP Wade PR Gershon MD Guinea pig 5-HT transporter: cloning, expression, distribution, and function in intestinal sensory reception Am J Physiol 1998 275 G433 G448 9724254
Grundy D Schemann M Serotonin in the gut: pretty when it gets down to the nitty gritty Neurogastroenterol Motil 2004 16 507 509 15500506 10.1111/j.1365-2982.2004.00573.x
Cash S Yuste R Linear summation of excitatory inputs by CA1 pyramidal neurons Neuron 1999 22 383 394 10069343 10.1016/S0896-6273(00)81098-3
Beck O Wallen NH Broijersen A Larsson PT Hjemdahl P On the accurate determination of serotonin in human plasma Biochem Biophys Res Commun 1993 196 260 266 8216301 10.1006/bbrc.1993.2243
Anderson GM Feibel FC Cohen DJ Determination of serotonin in whole blood, platelet-rich plasma, platelet-poor plasma and plasma ultrafiltrate Life Sci 1987 40 1063 1070 3821372 10.1016/0024-3205(87)90568-6
Chen JJ Li Z Pan H Murphy DL Tamir H Koepsell H Gershon MD Maintenance of serotonin in the intestinal mucosa and ganglia of mice that lack the high-affinity serotonin transporter: Abnormal intestinal motility and the expression of cation transporters J Neurosci 2001 21 6348 6361 11487658
Anderson GM Gutknecht L Cohen DJ Brailly-Tabard S Cohen JH Ferrari P Roubertoux PL Tordjman S Serotonin transporter promoter variants in autism: functional effects and relationship to platelet hyperserotonemia Mol Psychiatry 2002 7 831 836 12232775 10.1038/sj.mp.4001099
McBride PA Anderson GM Hertzig ME Snow ME Thompson SM Khait VD Shapiro T Cohen DJ Effects of diagnosis, race, and puberty on platelet serotonin levels in autism and mental retardation J Am Acad Child Adolesc Psychiatry 1998 37 767 776 9666633 10.1097/00004583-199807000-00018
Press WH Numerical recipes in C: the art of scientific computing 1992 2nd ed Cambridge, Cambridge University Press
Rohatgi VK Statistical inference 2003 Mineola, N.Y, Dover Publications
Chandana SR Behen ME Juhasz C Muzik O Rothermel RD Mangner TJ Chakraborty PK Chugani HT Chugani DC Significance of abnormalities in developmental trajectory and asymmetry of cortical serotonin synthesis in autism Int J Dev Neurosci 2005 23 171 182 15749243 10.1016/j.ijdevneu.2004.08.002
Chugani DC Muzik O Behen M Rothermel R Janisse JJ Lee J Chugani HT Developmental changes in brain serotonin synthesis capacity in autistic and nonautistic children Ann Neurol 1999 45 287 295 10072042 10.1002/1531-8249(199903)45:3<287::AID-ANA3>3.0.CO;2-9
Coon H Dunn D Lainhart J Miller J Hamil C Battaglia A Tancredi R Leppert MF Weiss R McMahon W Possible association between autism and variants in the brain-expressed tryptophan hydroxylase gene (TPH2) Am J Med Genet B Neuropsychiatr Genet 2005 135 42 46 15768392 10.1002/ajmg.b.30168
Whitaker-Azmitia PM Serotonin and brain development: role in human developmental diseases Brain Res Bull 2001 56 479 485 11750793 10.1016/S0361-9230(01)00615-3
Janusonis S Gluncic V Rakic P Early serotonergic projections to Cajal-Retzius cells: relevance for cortical development J Neurosci 2004 24 1652 1659 14973240 10.1523/JNEUROSCI.4651-03.2004
Hendricks TJ Fyodorov DV Wegman LJ Lelutiu NB Pehek EA Yamamoto B Silver J Weeber EJ Sweatt JD Deneris ES Pet-1 ETS gene plays a critical role in 5-HT neuron development and is required for normal anxiety-like and aggressive behavior Neuron 2003 37 233 247 12546819 10.1016/S0896-6273(02)01167-4
McKinney J Knappskog PM Haavik J Different properties of the central and peripheral forms of human tryptophan hydroxylase J Neurochem 2005 92 311 320 15663479 10.1111/j.1471-4159.2004.02850.x
Launay JM Ferrari P Haimart M Bursztejn C Tabuteau F Braconnier A Pasques-Bondoux D Luong C Dreux C Serotonin metabolism and other biochemical parameters in infantile autism. A controlled study of 22 autistic children Neuropsychobiology 1988 20 1 11 2466221
Croonenberghs J Verkerk R Scharpe S Deboutte D Maes M Serotonergic disturbances in autistic disorder: L-5-hydroxytryptophan administration to autistic youngsters increases the blood concentrations of serotonin in patients but not in controls Life Sci 2005 76 2171 2183 15733932 10.1016/j.lfs.2004.06.032
Kema IP de Vries EG Slooff MJ Biesma B Muskiet FA Serotonin, catecholamines, histamine, and their metabolites in urine, platelets, and tumor tissue of patients with carcinoid tumors Clin Chem 1994 40 86 95 7507008
Siegel GJ Basic neurochemistry: molecular, cellular, and medical aspects 1999 6th ed Philadelphia, Lippincott Williams & Wilkins
Racke K Reimann A Schworer H Kilbinger H Regulation of 5-HT release from enterochromaffin cells Behav Brain Res 1996 73 83 87 8788482 10.1016/0166-4328(96)00075-7
Schworer H Ramadori G Autoreceptors can modulate 5-hydroxytryptamine release from porcine and human small intestine in vitro Naunyn Schmiedebergs Arch Pharmacol 1998 357 548 552 9650808
Gebauer A Merger M Kilbinger H Modulation by 5-HT3 and 5-HT4 receptors of the release of 5-hydroxytryptamine from the guinea-pig small intestine Naunyn Schmiedebergs Arch Pharmacol 1993 347 137 140 8474534 10.1007/BF00169258
Schafermeyer A Gratzl M Rad R Dossumbekova A Sachs G Prinz C Isolation and receptor profiling of ileal enterochromaffin cells Acta Physiol Scand 2004 182 53 62 15329057 10.1111/j.1365-201X.2004.01299.x
Wade PR Tamir H Kirchgessner AL Gershon MD Analysis of the role of 5-HT in the enteric nervous system using anti-idiotopic antibodies to 5-HT receptors Am J Physiol 1994 266 G403 G416 8166280
Kirchgessner AL Tamir H Gershon MD Identification and stimulation by serotonin of intrinsic sensory neurons of the submucosal plexus of the guinea pig gut: activity-induced expression of Fos immunoreactivity J Neurosci 1992 12 235 248 1729436
Ahlman H Larsson I Lundberg J Dahlstrom A Donahue PE Bhargava HN Neural control of serotonin release from duodenal enterochromaffin cells of the cat Surg Forum 1978 29 404 407 401207
Kuemmerle JF Kraus H Kellum JM Serotonin release is mediated by muscarinic receptors on duodenal mucosal cells J Surg Res 1987 43 139 142 3306152
Racke K Schworer H Nicotinic and muscarinic modulation of 5-hydroxytryptamine (5-HT) release from porcine and canine small intestine Clin Investig 1992 70 190 200 1381648
Kahne D Tudorica A Borella A Shapiro L Johnstone F Huang W Whitaker-Azmitia PM Behavioral and magnetic resonance spectroscopic studies in the rat hyperserotonemic model of autism Physiol Behav 2002 75 403 410 11897268 10.1016/S0031-9384(01)00673-4
Varnas K Halldin C Pike VW Hall H Distribution of 5-HT4 receptors in the postmortem human brain--an autoradiographic study using [125I]SB 207710 Eur Neuropsychopharmacol 2003 13 228 234 12888181 10.1016/S0924-977X(03)00009-9
Vilaro MT Cortes R Mengod G Serotonin 5-HT(4) receptors and their mRNAs in rat and guinea pig brain: Distribution and effects of neurotoxic lesions J Comp Neurol 2005 484 418 439 15770652 10.1002/cne.20447
McBride PA Anderson GM Hertzig ME Sweeney JA Kream J Cohen DJ Mann JJ Serotonergic responsivity in male young adults with autistic disorder. Results of a pilot study Arch Gen Psychiatry 1989 46 213 221 2919950
Goldberg J Szatmari P Zwaigenbaum L Diminished serotonin receptor function: A potential endophenotype of autism International Meeting for Autism Research 2004 S3.2.3
Chugani DC Role of altered brain serotonin mechanisms in autism Mol Psychiatry 2002 7 Suppl 2 S16 S17 12142936 10.1038/sj.mp.4001167
Yu AM Idle JR Byrd LG Krausz KW Kupfer A Gonzalez FJ Regeneration of serotonin from 5-methoxytryptamine by polymorphic human CYP2D6 Pharmacogenetics 2003 13 173 181 12618595 10.1097/00008571-200303000-00007
Patel S Dulluc J Geffard M Comparison of serotonin and 5-methoxytryptamine immunoreactivity in rat raphe nuclei Histochemistry 1986 85 259 263 3528081 10.1007/BF00494813
Liu MT Rayport S Jiang Y Murphy DL Gershon MD Expression and function of 5-HT3 receptors in the enteric neurons of mice lacking the serotonin transporter Am J Physiol Gastrointest Liver Physiol 2002 283 G1398 G1411 12388212
David SP Murthy NV Rabiner EA Munafo MR Johnstone EC Jacob R Walton RT Grasby PM A functional genetic variation of the serotonin (5-HT) transporter affects 5-HT1A receptor binding in humans J Neurosci 2005 25 2586 2590 15758168 10.1523/JNEUROSCI.3769-04.2005
Christofi FL Kim M Wunderlich JE Xue J Suntres Z Cardounel A Javed NH Yu JG Grants I Cooke HJ Endogenous adenosine differentially modulates 5-hydroxytryptamine release from a human enterochromaffin cell model Gastroenterology 2004 127 188 202 15236185 10.1053/j.gastro.2004.04.070
Bertrand PP ATP and sensory transduction in the enteric nervous system Neuroscientist 2003 9 243 260 12934708 10.1177/1073858403253768
Davalos D Grutzendler J Yang G Kim JV Zuo Y Jung S Littman DR Dustin ML Gan WB ATP mediates rapid microglial response to local brain injury in vivo Nat Neurosci 2005 8 752 758 15895084 10.1038/nn1472
Courchesne E Pierce K Why the frontal cortex in autism might be talking only to itself: local over-connectivity but long-distance disconnection Curr Opin Neurobiol 2005 15 225 230 15831407 10.1016/j.conb.2005.03.001
Vargas DL Nascimbene C Krishnan C Zimmerman AW Pardo CA Neuroglial activation and neuroinflammation in the brain of patients with autism Ann Neurol 2005 57 67 81 15546155 10.1002/ana.20315
|
16029508
|
PMC1199627
|
CC BY
|
2021-01-04 16:39:26
|
no
|
Theor Biol Med Model. 2005 Jul 19; 2:27
|
utf-8
|
Theor Biol Med Model
| 2,005 |
10.1186/1742-4682-2-27
|
oa_comm
|
==== Front
Virol JVirology Journal1743-422XBioMed Central London 1743-422X-2-601608079710.1186/1743-422X-2-60ResearchHBx M130K and V131I (T-A) mutations in HBV genotype F during a follow-up study in chronic carriers León Bernal [email protected] Lizeth [email protected] Minor [email protected] Ronald B [email protected] Federico [email protected] Libia [email protected] Kirsten [email protected] International Center for Medical Research and Training, Louisiana State University ICMRT-LSU, San José, Costa Rica2 Pathology Department, San Juan de Dios Hospital, CCSS, Costa Rica3 Molecular Biology Center, Universidad of Costa Rica4 Virology Department, Microbiology School, Universidad of Costa Rica5 Microbiology, Immunology & Parasitology Department, School of Medicine, Louisiana State University, USA2005 4 8 2005 2 60 60 5 4 2005 4 8 2005 Copyright © 2005 León et al; licensee BioMed Central Ltd.2005León et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Around 400 million people worldwide are chronically infected with Hepatitis B virus (HBV). An estimated 10% of these chronic patients develop progressive liver damage including cirrhosis and Hepatocellular Carcinoma (HCC). The HBx gene encodes a protein of 154 amino acids which is a transactivator and has been associated with HBV pathogenesis. A change in the amino acid sequences at positions 130 and 131 in the HBV-X protein (M130K and V131I) produced by T-A point mutations at the nucleic acids level has been associated with severe liver damage and HCC in patients from China and Africa. Further, such changes have been proposed as a prognostic marker for progressive liver damage and HCC. The purpose of this study was to determine if T-A mutations are present in HBV chronic carriers with genotype F (the major genotype in Costa Rica) and further, if these mutations are associated with HBV disease progression in Costa Rica HBV patients from 1972 to 1985.
Results
Serum samples from 50 HBV positive individuals were amplified and directly sequenced, 48 belonged to genotype F, 1 from genotype D and another was classified as D or E.
T-;A mutations were absent in 17 acute patients who recovered, but was present in 12 of 29 chronic carrier samples (42.8%), in one sample the T-A mutations were detected as early as 29 days after clinical onset of disease. In 17 carriers with available liver biopsies, T-;A mutations were found in 8 sera of 13 (61.5%) classified as moderate or severe, and none in 4 biopsies with mild liver damage. However, it was not possible to demonstrate a statistical association between the presence of T-A mutations and moderate/severe liver damage, using a Fischer exact test, 1 tail, p = 0.05.
In 4 patients HCC was diagnosed, and 2 of them presented the T-A mutations in their sera.
Conclusion
T-A mutations were found in HBV genotype F in chronic carriers but not in patients who recovered from acute infection. These mutations could be developing early during infection although the possibility of infection with the mutant virus could not be excluded.
More studies are necessary to establish if the T-A mutation can be used as a prognostic marker for severity of liver disease in patients infected with HBV.
==== Body
Background
The hepatitis B virus (HBV) is a small double stranded DNA virus that produces a chronic infection in 2–10% of adults and in approximately 90% of infected infants. Approximately 10% of these chronic patients develop progressive liver damage including cirrhosis and Hepatocellular Carcinoma (HCC)[1]. The mechanism by which HBV progression to liver cirrhosis and/or HCC occurs is not clear, however many studies suggest that the X protein (HBx) is related to this process. HBx has been associated with a variety of biological functions. As a transcriptional transactivator, it can regulate transcription of a wide diversity of viral and cellular promoters [2,3]. HBx overlaps with regions of crucial importance for viral replication such as: the direct repeat sequences DR1 and DR2, the preC/C gene promoter and the enhancer II region. There are controversial results about the consequence of mutations in this region and its relationship with pathogenesis. A study carried out in Korea determined that mutations in the core promoter have little effect on viral load and the HBeAg status [4]. In contrast, another study points out that changes in HBx especially in the core promoter region may alter HBV gene expression [5]. Among other alterations observed in the HBx gene are deletions and one of the most common is the 8 bp deletion between nucleotides 1763–1770 [6], which has been described to decrease the virus replication [7,8]. These deletions in HBx as in other HBV genes have also been related to development of cirrhosis in long term renal transplant patients [9].
Natural mutations in the HBx gene have been related to progression to chronic disease as a consequence of the rescission of anti proliferative and apoptotic effects, which might produce uncontrolled growth and contribute to multistep hepatocarcinogenesis [10].
A double point mutation with a transversion nucleotide from adenine to thymine at nucleotide 1762, K130M with a transition from adenine to guanine at position 1764 V131I (T-A mutations), has been found more frequently in patients with hepatic tumors than in asymptomatic chronic patients from China [11,12] and Africa [13]. In East Asia where genotype C is the most common genotype, it has been reported that the T-A mutation occurs more frequently in relation to this genotype [14]. HBV is classified worldwide into eight genotypes designated A to H, with a specific geographical distribution [15-17].
Genotype F has been described as the HBV genotype of the Amerindians. In Central America a study determined 79% of samples belong to genotype F [18] and in Costa Rica genotype F is the most common, while the overall prevalence of HBV is considered low (0.5 – 1%).
From 1972 to 1985 a study on the natural history of HBV was done in San Ramón and Palmares, two adjacent Costa Rican counties [19]. In this study 488 cases of HBV were diagnosed, 80% with an age range between 5 and 40 years. In the group ≤ 5 years old 33% became chronic carriers and in the group > 5 years only 4.7% did. The 77.7% cases were primary HBV infections and the rest were due to household contacts. The purpose of this study was to analyze the presence of T-A mutations in the HBx gene for this population; the time which at they occur and if they are related to hepatic injure. Furthermore, the presence of other mutations in this gene were also observed
Results
PCR detection rate
Of the 77 selected samples, 18 were from group A, 14 from group B and 45 from group C; overall, 50 samples (64.9%) could be amplified and sequenced. Of these fifty, 17 (94.4%) were from group A (recovered patients), 12 (85.7%) from group B (paired samples – known onset), and 21(46.6%) from group C (chronic patient with unknown onset). The sensitivity of the nested PCR was 8000 copies/ml.
T-A mutations were present in chronic HBV carriers but not in acute recovered patients
Table 1 shows the mutation rate of T-A in HBx for M130K and V131I amongst the three study groups. The T-A mutations were not present in any of the 17 sequences from group A, where the average days in which samples were taken was 17 days ranging from 3–33 days. Of 8 chronic patients in group B, the T-A mutations were identified in 5 (62.5%) of the sequenced samples and V131I alone was detected in two. In one of the patients, T-A mutations were detected at day 29 after clinical onset. Four patients were not considered in the distribution of T-A mutations, since the follow-up samples could not be amplified. From group C the T-A mutations were detected in 7 of the 21 sequenced samples, and V131I alone in 3 samples.
Table 1 Distribution of the T-A mutations leading to (K130M and V131I) in the study groups.
GROUP MUTATIONS
T-A mutations V131I alone
#/n (%) #/n (%)
A 0/17 - -
B 5/8 (62.5) 2*/8
C** 7/20 (31.8) 3/20
* One of these two sample had V131I mutation in the first sample and TGA mutations later after a five year interval. One sample presented a deletion in that position
** One sample presented a deletion in the T-A position
Biopsy results and T-A mutations distribution
Of the 29 chronic carrier samples from groups B and C sequenced during the chronic phase, 18 patients had a liver biopsy characterized using the Knodell Index (KI). Five (26%) patients had a KI ≤ 2 points, (mild liver lesions with fatty deposits), 9 (47%) had KI between 3 and 4 points (moderate lesions) and 4 (21%) had a KI > 4 points (severe lesions). These are shown in fig. 1a, 1b and 1c respectively.
Figure 1 (A) – Persistent chronic hepatitis, Knodell index ≤ 2. Photomicrograph of liver showing chronic hepatitis with minimal activity. Hepatocytes showing regenerative features are seen, with minimal inflammation and scattered ground- glass hepatocytes. Cobblestone arrangement (diffuse regeneration) with Hadziyannis cells and without necrosis or fibrosis. (H&E 250×). (B) – Mild lobular chronic hepatitis, Knodell index 3–4. Photomicrograph of liver showing chronic hepatitis with mild activity. Spotty hepatocyte necrosis is seen in a lobular pattern with focal lymphocytic infiltration. Lesions are characterized by focal necrosis, conserved sinusoidal and trabecular patterns, lobular, portal, and focal lymphocytic infiltrated. (H&E 400×). (C) – Moderate lobular chronic hepatitis, Knodell index > 4. Photomicrograph of liver showing chronic hepatitis with moderate activity. There is portal chronic inflammation, focal interface hepatitis and periportal fibrous septa. Portal chronic swollen periportal apoptosis, post-necrosis fibrous interportal bridges. Nodular regeneration (pre-cirrhosis). (H&E 250×)
Figure 2 Sample deletions treated with Ssp I restriction enzymes. Recognition site of the enzyme SspI in the sequences with 8 bp deletion (left). In the right, samples with presumed deletions were run in a 3% agarose gel. Each pair of lines have the same sample treated with and without the Ssp I enzyme. An HIV sample having the AATATT site was used as positive control in lanes 1 and 2, sample 1430 (616 bp) lanes 3 and 4 (negative control), 1000 bp ladder marker lane 5, sample 6290 lanes 6 and 7, sample 467 lanes 8 and 9 sample 6516 lines 10 and 11, sample 6541 lanes 12 and 13. The samples 467 and 6516 treated with SspI presented two bands of 507 and 109 bp, lanes 8 and 10 (arrows) confirming the deletion. Details in sequence are:
Of the 5 carriers with biopsy classified as KI ≤ 2, one sample had an 8 bp deletion that included the T-A mutations site and another sample the V131I mutation alone. In the group with KI > 2 points (moderate/severe) T-A mutations were present in 8 (61.5%) of the sequenced samples (Table 2).
Table 2 Correlation between Knodell Index (KI) and HBx-T-A mutations.
MUTATIONS
T-A mutations V131I alone
Results KI #/n (%) #/n (%)
≤ 2 0/4 - 1/4 -
> 2 8/13 * (61.5) -
* One sample presented a deletion in the T-A position Fisher exact test, 1 tail, M130K p = 0.05, V131I p = 0.24
Table 3 reveals HBV carrier biopsies with KI > 2, age of the carrier at time of biopsy and sample collection, TSGO/TSGP levels and HBeAg/anti-HBe status.
According with statistics of the Costa Rican National Tumor Registry (NTR), four patients included in this study died from HCC during the last 2 decades and two of these had the presence of T-A mutations.
Table 3 Characterisation of samples with biopsies considered moderate and severe and patients who died from HCC.
Sex/Group Patient ident Sample Id/Time of sample collection after onset or study initiation Age at time of sample collection Patient age at time of biopsy collection Knodell Index K130M/V131I HBeAg/Anti-HBe TSGO/TSGP
M/B 950-08 865M/2 y 4 -/- +/- 50/32
6217M/12 y 14 15 2+1 = 3 +/+ -/-
M/B 671-10 445E/19 d 8 -/- +/- 74/83
5252M/9Y 17 18 2+2 = 4 +/+ -/+ 32/18
M/C 266-03 3751M/8 m 35 -/- -/+ 32/13
6400M/6 y 40 40 3+3 = 6 +/+ -/- 36/13
M/B 496-04 6461E/2 y 12 -/+ -/- 45/40
4904M/7 y 17 19 2+1 = 3 +/+ -/+ 28/25
M/B 969-04 467H/23 d 24 deletion -/- 500/550
6604M/7 y 31 32 HCC 5+2 = 7 +/+ -/- ND/18
M/C 1232-06 1481M/10 m 12 -/+ -/+ 36/21
6290M/9 y 22 23 2+1= 3 -/- -/+
M/C 65-35 6891M/11 y +/+ -/+ 55/50
20 20 3+0 = 3
M/C 158-01 6151M/11 y +/+ -/+ 36/16
33 35 1+2 = 3
M/C 921-07 6403M/10 y +/+ -/- 28/21
31 32 3+1 = 4
M/C 673-04 6572M/8 y -/- -/- ND/9
25 26 3+1 = 4
M/C 5-02 6067M/9 y 2+2 = 4 -/- -/+ 32/16
53 54
M/C 671-06 6593M/15 y -/- -/- ND/13
26 26 4+3 = 7
M/C 671-09 5251M/9y 2+4 = 6 -/- -/+ 28/18
16 19
M/B 1688-16 3254H/3 d 3 HCC -/- -/- 475/225
16 6653M/3 y 6 NB +/+ -/- ND/36
M/B 1400-01 575H/2 d 65 HCC -/- -/- 610/1200
5433M/4 Y 69 T -/- +/- 55/55
6825H/7 y 72 -/- +/- ND/55
M/C 1205-15 5399M/7 y 35 HCC -/- -/+ 32/28
ND = Not done, HCC = Hepatocellular carcinoma, T = tumor tissue only.
8 bp delections represent 8 % of the total samples
Four samples of the 50 samples (1 from group B and 3 from group C) presented 8 bp deletions at positions 389 to 397 nt of the HBx gene; the core promoter region, corresponding to 1763–1770 nt of the complete genome. Fig 2 shows the sequence and the band patterns of samples; 6541 (group C), 6290 (group C), 6516 (group C) and 467 (group B). To confirm that these deletions were not a PCR artifact, the samples were further digested by SspI. Of the four samples presenting the deletion only 2 were corroborated by SspI, both samples (467 and 6516) were re-amplified from the PCR1 product.
Mutations observed in HBV acute infected patients that recovered versus chronic carriers
The percentage of the most frequent polymorphism found in the study as well as the consensus sequences of each of the population selected for statistical analysis are shown in Table 4. Group A presented more amino acid or nucleotide variability than the other groups, however, in acute phase samples from group B, 50% of these had common mutations at position 12 (T12A).
Table 4 Major sequence polymorphisms found in the groups studied.
Amino acid-Position-mutation Frequency (%) Consensus sequences
L5M 28 Group A: 17 recovered patients
Q8K 22
T12A 36
S29P 38
S31P 30
S33P 30
V37I 33
P40S 30 Group B: 7 acute-chronic patients
MAARLCCQLDP-RDVLCLRPVGAESRGRSLSGSLGAVPPPSPSAVPADDGSHLSLRGLPV CSFSSAGPCALRFTSARRMETTVNAPRSLPTVLHKRTLGLSGRSMTWIEDYIKDCVFKDW EELGEEIRLKVFVLGGCRHKLVCSPAPCNFFTSA*
D48N 25
R87W 30
R103W 33
T106P 25 Group B: 32 chronic patients
MAARLCCQLDPTRDVLCLRPVGAESRGRSLSGSLGAVPPPSPSAVPADDGSHLSLRGLPV CSFSSAGPCALRFTSARRMETTVNAPRSLPTVLHKRTLGLSGRSMTWIEDYIKDCVFKDW EELGEEIRL- - FVLGGCRHKLVCSPAPCNFFTSA*
D110E 33
K130M 24
V131I 27
Deleted nt 8 Consensus Deletion group 8 bp
MAARLCCQLDPTRDVLCLRPVGAESRGRSLSGSLGAVPPPSPSAVPADDGSHLSLRGLPV CSFSSAGPCALRFTSARRMETTVNAPRSLPTVLHKRTLGLSGRSMTWIEDYIKDCVFKDW EELGEEIRLNIRRL*
390-397
end codon
135 aa
Hyphens in the consensus sequence represent the amino acid polymorphism associated with the left column. The predicted consensus amino acid sequence was obtained with Bioedit Software from the nucleotide sequence of the sample study.
Samples Genotype
Of the total 50 samples sequenced ; 48 belonged to genotype F, one sample belonged to genotype D subtype adw, and the other to subtype ay, which was classified as genotype E by a web-based genotyping tool and as D by phylogenetic tree analysis (data not shown).
Discusion
T-A mutations were not found in any of 17 samples from HBV patients who had recovered; a similar result had been obtained in a study with self-limited acute hepatitis [20]. However, another study showed T-A mutations during the acute phase in one out of 11 from genotype A, none of the 5 patients from genotype B and 4 out of 27 from genotype C [21]. The significance of this finding needs to be further studied.
T-A mutations were found in 12 (41.3%) of 29 samples from chronic carriers. In one carrier the mutations were detected 29 days after onset, with the probability that this carrier could have been directly infected with HBV containing the T-A mutations. In the 23 acute phase samples, T-A mutations were not detected and therefore the possibility to have an initial infection with T-A in other populations appears to be low. However, Kobayashi et al, has shown in their study a higher prevalence of the T-A mutations in chronic patients during the acute phase than in acute self limited HBV infection in patients infected with genotypes C, A and B [21].
In chronic carriers, with a liver biopsy classified as moderate or severe, T-A mutations were present in 61.5% (8/13) and none in 4 biopsies classified as mild. However this result was not statistically significant based on the Fisher exact test, 1 tail, p = 0.05, probably due to the small sample size in the groups. Other studies have shown a better correlation between the presence of T-A mutations and patients with fulminant hepatitis, severe exacerbation [20] or liver cirrhosis [22] especially with genotypes A or C when compared with asymptomatic carriers [12-14]. In agreement with the literature T-A mutations seem to appear more frequently in genotypes C [23,24] and A [13] than D [25,26] or B [27].
In this study, V131I also occurred alone in 5 samples (17%) of the 29 chronic patients; this event has been commonly reported by others [6,14,25,27,28]; nevertheless M130K alone is very unusual. It has been described in 1 of 12 fulminant hepatitis patients [20] and in 1 genotype B strain [27]. In one of the paired samples from this study and in another from reference [6], the V131I mutation appears in time before the methionine change at position 130.
In a Korean study T-A mutations were found in 32% (13/41) of HBV carriers, and a triple mutation G1714A, C1718T, A1721G was found in 27% (11/41) patients [4]. In our study wild type (wt) HBV strain nucleotide were found in the 1714 and 1718 positions, but the mutation A1721G was found in genotype F samples and not in two samples with other genotypes. Again, T-A mutations are common in all genotypes while other mutations seem to be more related to specific genotypes.
No association could be established between the presence of T-A mutations and HBeAg status (Table 3), similar to other published data [4,24]. Of the four samples with the 8 bp deletion only (467 and 6516), two were re-amplified from the PCR1 product and corroborated by enzyme restriction digestion, which demonstrates that the deletion was not a PCR artifact. This 8 bp deletion in the T-A site has been reported previously [6,8,9,27,29] and it has been associated with a low viral load [7,8,29]. Different clones isolated from several patients showed a heterogeneous population of strains including T-A mutations, wt strains as well as the 8 bp deletion. This could be a possible reason why we observed different results in amplified samples of the initial PCR products with an 8 deletion than in the reanalyzed two samples where the deletion was not detected.
An interesting fact is that these deletions alter the X open reading frame, changing K130N and introducing an isoleucine in the 131 site and a stop codon in the position135.
The polymorphic differences observed between the sequence of acute HBV recovered patients and chronic carriers are related to the genetic diversity of strain more than the study group classification (A,B,C). All sequences isolated in this study belong to genotype F with the exception of 2. Using blast searches sequences from genotype F can be divided in AY090455 – 1889 NIC sequences similar to those which are related to South American sequences and the AY090456- 1980HCR sequences similar to those which are related to Central America sequences. The polymorphism observed in the nucleotides as well as the amino acids in these groups may be due to a variability present in the group related to the South American sequences.
Many efforts have been made in order to clarify the role of viral variants in the pathogenesis of HBV infection; and still there is no final consensus. T-A mutations have been proposed as possible prognostic markers for liver disease progression [14] however, more studies are needed to elucidate the role of the T-A mutations and its relation to HBV diversity and disease outcome.
Conclusion
According to our results, T-A mutations were frequently observed in HBV chronic carriers, but were not found in acute recovered patients.
T-A mutations are frequent in all genotypes while other mutations seem to be more related to specific genotypes.
T-A mutations may appear early during HBV infection although the possibility of initial infection cannot be excluded.
Methods
Study population
Samples were obtained from a study of HBV in San Ramón and Palmares, Costa Rica areas outside of the capital city, San José, between 1972–1985 [19]. Based on serological markers and history of clinical onset, three groups were established: Group A, included 18 samples from acute cases who recovered from the infection; they presented initially as HBsAg positive, anti-IgM HBc positive and had elevated ALT levels. A patient was catalogued as a chronic carrier if HBsAg was present more than 6 months after the onset of disease. Group B, included 14 paired samples from chronic patients with known onset; with at least 3 years difference between the samples. Group C included 45 chronic patients with unknown date of onset. Twenty-nine patients had liver biopsy results, 4 from group B and 25 from group C.
The samples from all groups were negative by anti HAV IgM or anti- HCV [31] and were kept frozen.
This project was approved by the Ethical Committee of the Universidad of Costa Rica.
Biopsy classification Pathology
The inflammatory activity of Knodell in Chronic Persistent Hepatitis (CPH) between 1 and 2 points, is represented by a uniform and diffuse cobblestone arrangement of swollen hepatocytes, with compressed sinusoids; some of which show Hadziyannis cells containing abundant HBsAg.
Lobular Chronic Hepatitis (LCH) is between 2 and 6 points with an intact lobular architecture, perivenular cell swelling, focal hepatocytolysis and a variable degree of inflammatory activity [32]. Further, these lesions are characterized by focal necrosis, abnormal hepatocytes and scattered passive fibrous interportal bridges.
In this study the Knodell Index (KI) was used as follows: ≤ 2 points was considered mild liver lesion, 3 and 4 moderate and > 4 as severe liver damage.
PCR Methods
Primers were chosen from conserved regions of the following HBV genotypes sequence obtained from GenBank. Genotype A subtype adw2 (AF297625) and (AF373066), genotype B (AF121243), genotype C subtype adr (AB033550), subtype adw (AB033557), genotype D subtype ayw (AF280817), genotype E (AB032431), genotype F (AB036919), genotype G (AB064310) and (AF160501).
Outer primers selected were: sense (1182–1200) 5'GTTTGCTGACGCAACCCCC3' and the antisense 5'CAATGTCCATGCCCCAAAGC3' (1891–1910). The expected amplified product size was 728 bp. Inner primers: sense 5'GATCCATACTGCGGAACTCC3' (1263–1282) and antisense 5'AGCTTGGAGGCTTGAACAGT3' (1859–1878).
Genomic DNA was extracted from 200 μl of serum using the QIAamp DNA mini Kits (Qiagen® U.S.A.) according to manufacturer's instructions.
Nested PCR was performed using a thermocycler (Perkin-Elmer).
For the first PCR, 10 μl of the extracted product were added to a total of 50 μl of reaction volume containing 2.5 units of Taq (Promega® 5 units/μl), 3.5 mM of MgCl2, 0.092 nmoles/μl of primers final concentration, 0.4 mmolar/μl of each dNTP. This amplification was performed at 94°C for 3 min followed by 40 cycles at 94°C for 1 min, 50°C for 1 min and 72°C for 1 min, with a final extension of 4 min to 72°C.
For the nested PCR, 5 μl of product from the first PCR were added to 50 μl of reaction, with a final concentration of MgCl2, 2.5 mM and 0.080 nM of primers. Cycling conditions for the second round were 94°C for 3 min, 40 cycles to 94°C for 0.40 min, 55°C for 0.40 min and 72°C for 1.30 min. The final extension was 72°C for 4 min.
Nested products with a size of 616 bp were corroborated by 2% agarose gel electrophoresis stained with ethidium bromide.
Dilutions of 1:10 of a commercial CPG® DNA plasmid with 105copies/μl of the total HBV genome were prepared and used as control as well as to determine the limit detection (sensitivity) of the PCR system.
Sequencing conditions
Nested PCR product (616 bp) was run on 1% agarose gels and the expected band was cut and purified by a Qiagen column system following manufacturer's instructions.
An Open Gene™ sequencer system (Visible Genetics) was used. For sequencing the following primers were labeled with cy 5.0 and cy 5.5 dyes: Sense 5' 5cy55 GTTTYGCTCGCAGCMGGTC3' y = c/t, m = c/a (1292–1310) and antisense 5'-5cy5 CTTGAACGATRGGACATGAAC3' R = a/g (1848–1868).
Primers were diluted to a concentration of 3 pM in TE buffer. All reagents were used according to manufacturer's instructions. The first denaturation step was at 94°C for 2:30 min followed by 35 cycles of 0:30 min at 94°C, 0:30 min at 50°C, 1 min at 70°C and a final extension step at 72°C for 7 min. Finally, 1.5 μl of each sample was run in a polyacrylamide gel at 1500 volts for 90 min.
A consensus sequence of the genotype F strain (NCBI AB036919, AB036905, X75658) was used as our wild type sequence.
Genotype sequencing
The HBx gene sequences were compared with homologue sequences obtained from the GeneBank data base using the BLAST program [33]. The genotype was determined using a web- based genotyping tool for viral sequences [34]. The subtype of some of the samples was determined previously by specific antibodies available in our laboratory.
Restriction Enzyme digestion
In order to corroborate an 8 bp deletion observed in some sequences, a restriction enzyme SspI was used (New England, BioLabs INC,). As a positive control a sample from HIV having the same recognition site was used and a HBx sample with the wild type sequence was employed as a negative control. Ten μl of each purified product from the nested PCR were dispensed into two different vials of 200 μl. In one vial 1 μl of SspI enzyme (5000 units/ml), 2 μl of enzyme buffer (New England, BioLabs INC,) and 7 μl of water were added; while in the other vial the enzyme was omitted. All samples were heated at 37°C for 90 minutes and run in a 3% agarose gel. Results were visualized with ethidium bromide.
Statistical analysis
The Fisher's exact test was used to evaluate the relationship between two discrete and dichotomy variables. The t test, for independent samples, was used to analyze continuous variables when it was necessary. A new dichotomy variable for hepatic damage was built into biopsy results and using data from the Costa Rican National Tumor Registry (NTR); by division into "mild damage" and "moderate/severe damage". The relative risk (RR) was calculated with a 95% confidence interval. All analyzes were done with the JMP 4 software version 4.0.4 A BUSINESS UNIT OF SAS Copyright © 1989 – 2001 SAS Institute Inc. (all rights reserved) and Epiinfo software CDC.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
BL, FA, KV experimental design planning research
BL, MV laboratory: molecular and pathology work, respectively
BL, FA statistical analysis
BL, KV editing
LH, LT, RBL contributed to manuscript content and editing of drafts
Acknowledgements
This research was supported by Ministerio de Ciencia y Tecnología (Ministry of Science and technology), Consejo Nacional para Investigaciones Científicas y Tecnológicas (National Council for Science Research and Technology) and Organización Panamericana de la Salud (Health Panamerican Organization) grant.
The authors thank all the persons that kindly collaborated in the revision of the manuscript, particularly to Dr. Joseph Schwarzman, Professor of Pathology, Dartmouth-Hitchcock Medical Center, Lebanon, NH, USA for his appropriate comments and to Ms. Virginia Larrad for editorial assistance.
==== Refs
Beasley RP Hwang LY Lin CC Chien CS Hepatocelullar carcinoma and hepatitis B virus: a prospective study of 22707 men in Taiwan Lancet 1981 2 1129 33 6118576 10.1016/S0140-6736(81)90585-7
Bergametti F Sitterlin D Transy C Turnover of Hepatitis B Virus X protein is regulated by a Damaged DNA-Binding Complex J Virol 2002 76 6495 6501 12050362 10.1128/JVI.76.13.6495-6501.2002
Lee H Yun Y HBx protein of Hepatitis B virus activates Jak1-STAT signaling J Biol Chem 1998 273 25510 25515 9738022 10.1074/jbc.273.39.25510
Chun YK Kim JY Woo HJ Oh SM Kang I Ha J Kim SS No significant correlation exist between core promoter mutations, viral replication and liver damage in chronic hepatitis B infection Hepatology 2000 32 1154 1162 11050069 10.1053/jhep.2000.19623
Buckwold VE Xu Z Chen M Yen TS Ou JH Effects of a naturally occurring mutation in the hepatitis B virus basal core promoter on precore gene expression and viral replication J Virol 1996 70 5845 5851 8709203
Horikita M Itoh S Yamamoto K Shibayama T Tsuda F Okamoto H Differences in the entire nucleotide sequence between hepatitis B virus genomes from carriers positive for antibody to hepatitis B e antigen with and without active disease J Med Virol 1994 44 96 103 7798892
Sallam TA Tong CY Two distinct types of hepatitis B virus core promoter variants in Yemeni blood donors J Med Virol 2002 68 328 34 12226818 10.1002/jmv.10207
Schlager F Schaefer S Metzler M Gratzki N Lampert F Gerlich WH Repp R Quantitative DNA fragment analysis for detecting low amounts of hepatitis B virus deletion mutants in highly viremic carriers Hepatology 2000 32 1096 105 11050061 10.1053/jhep.2000.19323
Preikschat P Gunther S Reinhold S Will H Budde K Neumayer H Kruger DH Meisel H Complex HBV populations with mutations in core promoter, C gene and Pre S region are associated with development of cirrhosis in long-term renal transplant recipients Hepatology 2002 35 466 477 11826424 10.1053/jhep.2002.30698
Sirma H Giannini C Poussin K Paterlini P Kremsdorf D Brechot C Hepatitis B Virus X mutants, present inhepatocellular carcinoma tissue abrogate both the antiproliferative and transactivation effects of HBx Oncogene 1999 26 4848 59 10490818 10.1038/sj.onc.1202867
Hsia CC Yuwen H Tabor E Hot spot mutations in hepatitis B virus X gene in hepatocelullar carcinoma Lancet 1996 348 625 626 8774611 10.1016/S0140-6736(05)64851-9
Fang ZL Ling R Wang SS Nong J Huang CS Harrison TJ HBV core promoter mutations prevail in patients with hepatocellular carcinoma in Guangxi, China J Med Virol 1998 56 18 24 9700628 10.1002/(SICI)1096-9071(199809)56:1<18::AID-JMV4>3.0.CO;2-Q
Batista M Kramvis A Kew M High prevalence of 1672T 1764G mutations in the basic core promoter of Hepatitis B virus isolated from black Africans with hepatocellular carcinoma compared with asymptomatic carriers Hepatology 1999 29 946 953 10051502 10.1002/hep.510290336
Lindh M Hannoun C Dhillon AP Norkrans G Horal P Core Promoter Mutations and Genotypes in Relation to Viral Replication and Liver Damage in East Asian Hepatitis B Virus Carriers J Inf Dis 1999 179 775 82 10068571 10.1086/314688
Norder H Hammas B Lee SD Courouce AM Mushahwar IK Magnius L Genetic relatedness of hepatitis B viral strains of diverse geographical origin and natural variations in the primary structure of the surface antigen J Gen Virol 1993 74 1341 1348 8336122
Norder H Courouce AM Magnius LO Complete genomes, phylogenetic relatedness, and structural proteins of six strains of the hepatitis B virus, four of which represent two new genotypes Virology 1994 218 214 223 8615024 10.1006/viro.1996.0181
Arauz-Ruiz P Norder H Robertson BH Magnius L Genotype H: a new American genotype of hepatitis B virus revealed in Central America J Gen Virol 2002 83 2059 2073 12124470
Arauz-Ruiz P Norder H Visoná K Magnius L Genotype F prevails in HBV infected patients of Hispanic origin in Central America and may carry the precore stop mutant J Med Virol 1997 51 305 312 9093945 10.1002/(SICI)1096-9071(199704)51:4<305::AID-JMV8>3.0.CO;2-9
Visoná K Eduarte C Zamora E Salazar L Estudio epidemiológico de las hepatitis virales en San Ramón y Palmares de 1972–1985 Acta Médica Costarricense 1989 33 69 77
Honda A Yokusaka O Suzuki K Saisho H Detection of mutations in hepatitis B virus enhancer 2/core promoter and x protein regions in patients with fatal hepatitis B virus infection J Med Virol 2000 62 167 176 11002245 10.1002/1096-9071(200010)62:2<167::AID-JMV7>3.0.CO;2-Z
Kobayashi M Arase Y Ikeda K Tsubota A Suzuki Y Saitoh S Kobayashi M Suzuki F Akutas N Hosaka T Someya T Matsuda M Sato J Miyakawa Y Kumada H Wild type precore and core promoter sequences in patients with acute self limited or chronic Hepatitis B Scand Journal Gastroenterol 2004 1 53 59
Cho SW Shin YJ Hahm KB Jin JH Kim YS Kim HJ Analysis of the precore and core promoter DNA Sequence in liver tissues from patients with Hepatocellular carcinoma J Korean Med 1999 14 424 30
Fang ZL Yang J Ge X Zhuang H Gong J Li R Ling R Harrison TJ Core promoter mutations (A1762 T and G1764 A) and viral genotype in chronic hepatitis B and hepatocellular carcinoma in Guangxi, China J Med Virol 2002 68 33 40 12210428 10.1002/jmv.10167
Ni YH Chang MH Hsu HY Tsuei DJ Longitudinal study on mutation profilies of core promoter and precore regions of the hepatitis B virus genome in children Ped Res 2004 56 396 399
Vernard V Corsaro D Kajzer C Bronowicki J Faou A Hepatitis B virus X gene variability in French-born patients with chronic hepatitis and hepatocellular carcinoma J Med Virol 2000 62 177 184 11002246 10.1002/1096-9071(200010)62:2<177::AID-JMV8>3.0.CO;2-V
Hannoun C Horal P Lindh M Long -term mutation rates in the hepatitis B virus genome J Gen Virol 2000 81 75 83 10640544
Bläckberg J Kidd-Ljunggren K Mutations within the hepatitis B virus genome among hepatitis B patients with Hepatocellular carcinoma J Med Virol 2003 71 18 23 12858404 10.1002/jmv.10458
Gandhe SS Chadha MS Walimbe AW Arankalle VA Hepatitis B virus: prevalence of precore/core promoter mutants in different clinical categories of indian patients Viral Hepatitis 2003 10 367 382 10.1046/j.1365-2893.2003.00445.x
Li KS Yamashiro T Sumie A Terao H Mifune K Nishizono A Hepatitis B virus harboring nucleotide deletions in the core promoter region and genotype B correlate with low viral replication activity in anti-HBe positive carriers J Clin Virol 2001 23 97 106 11595588 10.1016/S1386-6532(01)00212-8
Kuang SY Jackson PE Wang JB Lu PX Munoz A Qian GS Kensler TW Groopman JD Specific mutations of hepatitis B virus in plasma predict liver cancer development Proc Natl Acad Sci U S A 2004 101 3575 80 14990795 10.1073/pnas.0308232100
Palacios A Taylor L Haue L Luftig RB Visona KA Development of low cost peptide-based anti-Hepatitis C virus screening and confirmatory assays: Comparison with commercially available tests J Med Virol 1999 58 221 226 10447416 10.1002/(SICI)1096-9071(199907)58:3<221::AID-JMV6>3.0.CO;2-S
Peters RL Okuda K, Peters RL Hepatocellular carcinoma 1976 Chapter 8 John Wiley
Altschul SF Gish W Miller W Myers E Lipman D Basic local alignment search tool J Mol Biol 1990 215 403 410 2231712 10.1006/jmbi.1990.9999
Rozanov M Plikat U Chappey C Kochergin A Tatusova TA Web-based genotyping resource for viral sequences Nucleic Acids Research 2004 32 w654 659 15215470
|
16080797
|
PMC1199628
|
CC BY
|
2021-01-04 16:39:00
|
no
|
Virol J. 2005 Aug 4; 2:60
|
utf-8
|
Virol J
| 2,005 |
10.1186/1743-422X-2-60
|
oa_comm
|
==== Front
World J Surg OncolWorld Journal of Surgical Oncology1477-7819BioMed Central London 1477-7819-3-451602273910.1186/1477-7819-3-45ReviewDietary fat and risk of breast cancer Binukumar Bhaskarapillai [email protected] Aleyamma [email protected] Division of Epidemiology and Clinical Research, Regional Cancer Centre, Thiruvananthapuram – 695011 Kerala, India2005 18 7 2005 3 45 45 29 11 2004 18 7 2005 Copyright © 2005 Binukumar and Mathew; licensee BioMed Central Ltd.2005Binukumar and Mathew; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Breast cancer is one of the major public health problems among women worldwide. A number of epidemiological studies have been carried out to find the role of dietary fat and the risk of breast cancer. The main objective of the present communication is to summarize the evidence from various case-control and cohort studies on the consumption of fat and its subtypes and their effect on the development of breast cancer.
Methods
A Pubmed search for literature on the consumption of dietary fat and risk of breast cancer published from January 1990 through December 2003 was carried out.
Results
Increased consumption of total fat and saturated fat were found to be positively associated with the development of breast cancer. Even though an equivocal association was observed for the consumption of total monounsaturated fatty acids (MUFA) and the risk of breast cancer, there exists an inverse association in the case of oleic acid, the most abundant MUFA. A moderate inverse association between consumption of n-3 fatty acids and breast cancer risk and a moderate positive association between n-6 fatty acids and breast cancer risk were observed.
Conclusion
Even though all epidemiological studies do not provide a strong positive association between the consumption of certain types of dietary fat and breast cancer risk, at least a moderate association does seem to exist and this has a number of implications in view of the fact that breast cancer is an increasing public health concern.
==== Body
Background
Breast cancer is the most common malignancy amongst women worldwide, constituting 10% of all cancers. It is estimated that more than 1.1 million new breast cancer cases occur worldwide annually representing over twenty percent of malignancies in women [1]. Incidence rates of breast cancer are approximately 90–130 per 100,000 women in developed countries and those in developing countries are approximately ten to sixty per 100,000 women [1]. Several studies have been carried out to identify the risk factors for developing breast cancer. Studies of reproductive factors suggest that nulliparity and late age at first childbirth are the most consistent risk factors associated with breast cancer [2]. Certain dietary factors such as a higher intake of fat and meat also seem to increase the risk of breast cancer [3,4]. Fat is a macronutrient and is considered to be a major source of calories or energy. Although some fat in the diet is necessary, too much of fat can lead to heart diseases, cancers, obesity and other health problems. Numerous studies in women, using different study designs and in different geographical areas have been carried out in order to establish the relationship of dietary fat to breast cancer risk.
The objective of the present communication is to summarize evidence from various case-control and cohort studies on the consumption of dietary fat and its sub-types [saturated, monounsaturated (MUFA) and polyunsaturated (PUFA) fatty acids] and their effect on the development of breast cancer. The major sources of saturated fats are meat, poultry, dairy products, eggs and some plant foods such as coconut, coconut oil and palm oil [5]. High concentrations of MUFA are found in vegetable oils and their traces are also found in meat. Oleic acid, the most abundant MUFA, is found in animal and vegetable oils. Oleic acid is the major component of olive oil. PUFAs are classified into two families – n-3 PUFAs and n-6 PUFAs. PUFAs are mostly found in plants, fish and seafoods. Rich sources of n-3 PUFA include mackerel, salmon, and albacore tuna [6]. Types of n-3 fatty acids include eicosapentaenoic acid (EPA; 20:5 n-3), docosahexaenoic acid (DHA; 22:6 n-3), α-linolenic acid (18:3 n-3) and stearidonic acid (18:4 n-3). EPA and DHA are present in fatty fish although it can be obtained from enriched eggs [7]. N-6 fatty acids are found in high concentration in corn oil, safflower oil, soybean oil, sunflower oil and cottonseed oil. Types of n-6 fatty acids include linoleic acid (18:2 n-6), arachidonic acid (20:4 n-6), docosapentaenoic acid (DPA; 22:5 n-6) and γ-linolenic acid (18:3 n-6) [8].
Methods
A Pubmed search of the literature was carried out covering studies conducted over a period of fourteen years (from 1990 to 2003) using keywords "dietary fat" and "risk" and "breast cancer". The information on study design, country, data collection method, study period, sample size, study subjects, confounding variables, partition values used for comparisons, odds ratio/relative risk and trend p-value (if available) were obtained from the articles. Only case-control and cohort studies conducted among women were considered for the present review. Reports from symposia, genetic studies, survival/mortality studies of breast cancer patients and all studies associated with breast cancer other than dietary fat in women were excluded from the present review.
Total dietary fat
Case-control studies
Of the seventeen case-control studies [6,9-24] that investigated the relation of total dietary fat with risk of breast cancer, seven studies reported a significant positive association between total fat intake and risk of breast cancer [10,11,17-19,22,24]. The odds ratios (OR) of these studies ranged from 1.45 to 8.47 in the highest category of total fat consumption. Significant dose-response relationships between the increased consumption of total fat and breast cancer risk were observed in four studies [11,18,22,24]. Increased risk with border line significance was found for the intake of fat meat in one study [OR = 1.18 (95% CI: 1.0–1.5) in the highest quartile of fat meat intake] [17]. Non-significant increased risks were reported in six studies [6,14,16,20,21,23]. Significant reduced risk with a dose-response relationship was observed in one study [12]. Three studies have shown no association [9,13,15] (Additional file 1).
Cohort studies
Of the twenty cohort studies, which investigated the relationship between total dietary fat intake and breast cancer risk, ten were based exclusively on post-menopausal women [25-34]. Five studies showed significant positive associations with dose-response relationships between total fat and breast cancer risk [25,26,31,34,35]. The relative risks for these studies ranged from 1.15 to 3.47 in the highest level of total fat consumption. Non-significant positive associations between total fat and the risk of breast cancer were observed in nine studies [30,32,33,36-41]. Six studies found no association between the fat intake and the risk of breast cancer [27-29,42-44] (Additional file 2).
Components of fat
Saturated fat
Case-control studies
Four studies out of the seventeen reviewed here found significant positive associations with dose-response relationships between saturated fat and breast cancer [9,18,19,45]. The odds ratios for these studies ranged from 1.90 to 2.38. Nine studies showed non-significant increase in associations of saturated fat with the development of breast cancer [12,14,16,21,23,46-48,66]. Four studies reported no association between saturated fat and breast cancer risk [6,15,49,50] (Additional file 1).
Cohort studies
Among the seventeen cohort studies, three reported positive associations with significant dose-response relationships between the consumption of saturated fat and the risk of breast cancer [26,36,41]. The relative risks for these studies ranged from 1.13 to 1.47 in the highest level of saturated fat consumption. Non-significant increased associations for breast cancer were found in eight studies [30-33,37,38,40,51]. Six studies found no associations between the consumption of saturated fat and the risk of breast cancer [27,28,34,42,43,52] (Additional file 2).
Monounsaturated fat -Oleic acid
Case-control studies
Two studies of increased consumption of oleic acid (the most abundant MUFA) [12,57] and four studies of increased consumption of olive oil (a major component of oleic acid) [48,54-56] reported reduced risks of breast cancer. No association was observed in one study for the consumption of oleic acid [53] (Additional file 1).
Cohort studies
Some cohort studies found that increased consumption of oleic acid reduced the risk of breast cancer but the associations were non-significant [32,33,43]. However, other studies have shown no association [27,40] (Additional file 2).
Total monounsaturated fat
Case-control studies
Contrary to the studies on oleic acid consumption, three of the twelve case-control studies reported positive associations with a dose-response relationship between increased consumption of total MUFA and the risk of breast cancer [16,18,23]. The odds ratios for these studies ranged from 1.04 to 1.89 in the highest level of MUFA consumption. Non-significant increased associations for higher consumption of MUFA were found in three studies [6,21,23]. The rest of the studies reviewed reported no association between MUFA consumption and subsequent risk of breast cancer [14,15,46,48-50] (Additional file 1).
Cohort studies
Of the twelve studies that have been reviewed, three showed significant positive associations with dose-response relationships for increased intake of MUFA and the risk of breast cancer [34,37,38]. The relative risks for these studies ranged from 1.72 to 2.01 in the highest level of MUFA consumption. Five studies reported non-significant increased breast cancer risk [28,30,31,36,51]. Three studies showed inverse associations with significant dose-response relationships between the higher intake of MUFA and the risk of breast cancer [33,42,52]. The relative risks of these studies varied from 0.61 to 0.94. No significant association between the consumption of MUFA and breast cancer was found in one study [41] (Additional file 2).
Total polyunsaturated fat
Case-control studies
Six studies out of fourteen reported a significantly reduced breast cancer risk with increased PUFA consumption [12,15,45,48,57,66]. The odds ratios for these studies ranged from 0.39 to 0.70 in the highest category of PUFA consumption. Decreased breast cancer risk with significant dose response patterns was reported for high consumption of PUFA in three studies [12,15,57]. No associations were found between PUFA intake and risk of breast cancer in eight studies [6,14,16,18,21,23,46,49] (Additional file 1).
Cohort studies
Contrary to the results based on case-control studies, three out of eleven cohort studies reported significant positive associations with dose-response relationship between the increased consumption of PUFA and the risk of breast cancer [30,34,52]. The relative risks of these studies ranged from 1.49 to 3.02. Four studies reported non-significant increased risks associated with increased PUFA consumption [28,31,36,38]. Four studies reported no association between the consumption of PUFA and risk of breast cancer [33,41,42,51] (Additional file 2).
n-3 fatty acids
Case-control studies
One study of n-3 fatty acids [58] and another study of DHA reported significantly reduced risks for breast cancer [59] (Additional file 1). A significantly reduced breast cancer risk with a dose-response relationship was observed in one study for the increased consumption of DHA (n-3 fatty acid) [59]. Two studies of consumption of alpha-linolenic acid showed inverse association with a significant dose-response relationship [50,59]. One study of EPA (n-3 fatty acid) showed an inverse association [58]. No association was observed in other studies for the increased consumption of n-3 fatty acids [6,16,53,55] with the development of breast cancer (Additional file 1).
Cohort studies
Significantly reduced risk of breast cancer was observed in one study for increased consumption of n-3 fatty acids (RR = 0.72 in the highest quartile of intake) [28]. Contrary to the above result, a Swedish cohort study among post-menopausal women reported a significantly increased breast cancer risk with dose-response relationship [34]. The other studies showed no associations for breast cancer incidence on the consumption of n-3 fatty acids [36,51], DHA [33] and EPA [33,51] (Additional file 2).
n-6 fatty acids
Case-control studies
One study from France of linoleic acid (n-6 fatty acid) consumption reported a highly significant increased breast cancer risk (OR = 2.31 in the highest category of consumption of linoleic acid) [59]. Two studies of n-6 fatty acids reported non-significant increased breast cancer risks [55,58]. No associations were observed for the consumption of n-6 fatty acids [50,53,59] with the development of breast cancer (Additional file 1).
Cohort studies
One study conducted in Sweden among postmenopausal women reported a significantly increased breast cancer risk with dose-response relationship on the consumption of n-6 fatty acids [OR = 3.02 (95% CI: 1.78–5.13) in the highest quintile] [34]. Non-significant increased breast cancer risks were observed with increased consumption of n-6 fatty acids [28] and linoleic acid (n-6 fatty acid) [31,33]. A few studies found no risk of breast cancer with the consumption of linoleic acid [27,32,40,43] and arachidonic acid (n-6 fatty acid) [33] (Additional file 2).
Discussion
Most of the case-control studies discussed here have shown increased breast cancer risk with increased total fat consumption. However, results of cohort studies did not offer the same degree of consistent support.
Case-control and cohort studies share some strengths and limitations. The primary advantages are the ability to measure diet and potential confounding factors relatively accurately and uniformly at the individual level in both the types of study designs. Also, the subjects in most of the studies were generally drawn from a relatively uniform population. Despite these advantages, there remains a number of limitations common to both case-control and cohort studies.
As fat intake is estimated by food frequency questionnaire (FFQ) in most of the studies discussed in the present review, there remains a certain amount of measurement error in such procedures. Non-differential random measurement error (i.e. independent of case or non-case status) in estimated fat intake might bias estimated risks toward the null and concurrent measurement error in potential confounding variables (e.g., other sources of energy intake) could bias results in either direction. Only a few studies mentioned about validated FFQ [6,14,24,28,33,53,66]. Non-validated questionnaire for collecting the dietary information might also affect the reliability of study results. Another important limitation identified by the present review is that most of the studies are generally conducted in a relatively homogenous population; hence the range of fat intake and the power of the study to detect a true association is restricted. Due to the lack of a definitive biologic mechanism and the lack of information on the relevant timing of exposure, it is possible that case-control and cohort studies are measuring diet at the wrong point in time; this again might attenuate the observed risk in any particular study.
The main contrast between case-control and cohort studies lies in their differing potential for bias and in the resources required. Selection bias would occur if participation in a case-control study was correlated with fat intake and if cases and controls have different participation rates. In many of the case-control studies discussed here, particularly those conducted in highly industrialized populations such as the United States, there is often a substantial refusal to participate, particularly among control subjects, and it is possible that there is preferential participation by health-conscious individuals, who might have a relatively low fat intake. Such a phenomenon might introduce an artificial positive association between fat intake and breast cancer risk. Recall bias might occur since cases are interviewed after diagnosis and might report their past diet differently to control subjects. In particular, they could report their diets more accurately through better motivation, or conversely, they could report an over inflated fat consumption if aware of the postulated association between fat intake and breast cancer risk.
Since there are many limitations associated with case-control studies, the results from the cohort studies can be considered more authentic. Cohort studies assessing the relationship between total dietary fat and risk of breast cancer have been used since the late 1980's as an alternative to the inconsistent findings of the case-control studies. These are normally considered free of the most common biases that potentially affect case-control studies such as selection and information bias.
Even though most of the cohort studies failed to find an association between total dietary fat intake and breast cancer risk, some studies reported a significant positive association with dose-response relationship [25,26,31,34,35]. The failure of cohort studies to show a consistent and a strong relationship may be attributed to the difficulty in collecting accurate dietary information. The methodological limitations associated with the study designs and inaccuracies in the measurement of fat may have obscured any relationship between fat and breast cancer that might exist.
When we consider the components of fat, saturated fat, the unhealthiest out of all the types of fat, when consumed in large amounts increases the risk of breast cancer. In the present review, several case-control studies have shown positive association with breast cancer risk [9,18,19,45]. However, there are only a few cohort studies [26,36,41] reported statistically significant increased breast cancer risk with increased saturated fat consumption.
Monounsaturated fats (MUFA) are considered as 'non-essential' because they can be synthesized within our bodies, from foods and oils such as oleic acid and olive oil. Even though a few studies showed a positive correlation between MUFA and breast cancer [16,18,34,37], most of the studies reviewed here have shown that oleic acid including olive oil has an inverse relationship with breast cancer [12,16,18,33,34,37]. Many previous epidemiological studies have pointed out the preventive properties of olive oil due to its high content of oleic acid [56,60,61].
In the present review on polyunsaturated fatty acids, we focused mainly on n-6 and n-3 families. Various studies have analyzed the possible relation of (n-3)/(n-6) PUFA ratio and the possible role that it may play in the risk of breast cancer [6,58]. Most of the studies reviewed have shown that a higher (n-3)/(n-6) PUFA ratio may reduce the risk of breast cancer [6,58,59,62]. Some studies, however, found results which suggests that high intake of monounsaturated fatty acids, n-3 fatty acids and n-6 fatty acids were related to lower risk [63] and the same study, which primarily investigated the role of recall bias, found no relation between PUFA and breast cancer risk.
Boyd and others [4] carried out a meta analysis of all published studies up to July 2003. They found that 14 cohort and 31 case-control studies gave similar results and suggested that total and saturated fat intake are positively associated with increased risk of breast cancer. Saadatian-Elahi and others [64] analyzed 3 cohort and 7 case-control studies including 2,031 cases and 2,334 controls and concluded that n-3 PUFA had a protective effect and total MUFA and oleic acid levels were significantly associated with an increase of breast cancer risk. In a pooled analysis, Hunter and others [65] gathered information about 4,980 cases from studies including 337,819 women and found that there was no evidence of association between total dietary fat intake and the risk of breast cancer. The meta-analytical studies arrived at a conclusion similar to the present review and suggest that certain dietary fats are associated with breast cancer risk. Only Hunter et al, [65] have differed in that they found no association. Prentice and Sheppard [83] have, in fact, concluded that 50% reduction of dietary fat from current USA consumption levels would result in a 2.5 fold reduced risk of breast cancer among postmenopausal women.
Previously published review papers reported increased risk of breast cancer with the increased consumption of total dietary fat [68,69,71-75]; saturated fat [69,76-79] and moderately decreased association with oleic acid and monounsaturated fatty acids [67,70,80,81]. These results are consistent with our present review except for an equivocal association observed in the case of monounsaturated fatty acids. A few reviews have shown no association of various types of fat with the etiology of breast cancer [76,82]. Previous epidemiologic evidence provides little support for any important relation between intake of either linoleic acid (n-6 fatty acid) or n-3 fatty acids and risk of breast cancer [67] as against a moderate inverse association between the increased consumption of n-3 fatty acids and the risk of breast cancer and a moderate positive association between the increased consumption of n-6 fatty acids and the risk of breast cancer.
Conclusion
Considering the totality of epidemiologic evidence, it is reasonable to conclude that a positive association exists between total dietary fat intake and breast cancer risk, even though all epidemiological studies do not provide a strong association between consumption of dietary fat and breast cancer risk. Also, increased consumption of saturated fat is positively associated with the development of breast cancer. There exists an inverse association of breast cancer risk with the consumption of oleic acid. A moderate inverse association between breast cancer risk and the consumption of n-3 fatty acids and a moderate positive association between n-6 fatty acids and breast cancer risk were also observed. These have lot of implications in view of the fact that breast cancer is a growing public health problem.
Competing interests
The present review paper will be quite useful for surgeons, clinicians, epidemiologists and other public health professionals particularly working in the filed of cancer.
Authors' contributions
BB – collected the literature, summarized the results in the form of tables and wrote the paper
AM: provided technical guidance for writing the paper and edited the paper
Funding sources
Nil
Supplementary Material
Additional File 1
Table showing case control studies and risk of breast cancer.
Click here for file
Additional File 2
Table showing cohort studies and risk of breast cancer.
Click here for file
==== Refs
Parkin DM Bray F Ferlay J Pisani P Estimating the world cancer burden: Globocon 2000 Int J Cancer 2001 94 153 156 11668491 10.1002/ijc.1440
McPherson K Steel CM Dixon JM Breast cancer- epidemiology, risk factors, and genetics BMJ 2000 321 624 628 10977847 10.1136/bmj.321.7261.624
WCRF (World Cancer Research Fund) American Institute for Cancer Research Expert Panel; Food, Nutrition, and Prevention of Cancer: A Global Perspective Am Inst Cancer Res 1997 Washington, DC
Boyd NF Stone J Vogt KN Connelly BS Martin LJ Minkin S Dietary fat and breast cancer risk revisited: a meta-analysis of the published literature Br J Cancer 2004 111 584 591
Saturated fat last accessed on July 10, 2005
Goodstine SL Zheng T Holford TR Ward BA Carter D Owens PH Mayne ST Dietary (n-3)/(n-6) fatty acid ratio: possible relationship to premenopausal but not to post-menopausal breast cancer risk in U.S. women J Nutr 2003 133 1409 1414 12730430
Horrocks LA Yeo YK Health benefits of docosahexanoic acid Pharmacol Res 1999 40 211 225 10479465 10.1006/phrs.1999.0495
Lee D Thomas G Fats, fish oil and omega 3 fatty acids accessed on July 10, 2005
Do MH Lee SS Jung PJ Lee MH Intake of dietary fat and vitamin in relation to breast cancer risk in Korean women: a case-control study J Korean Med Sci 2003 18 534 540 12923330
D'Avanzo B Negri E Gramenzi A Franceschi S Parazzini F Boyle P La Vecchia C Fats in seasoning and breast cancer risk: an Italian case-control study Eur J Cancer 1991 27 420 423 1828170
Ewertz M Gill C Dietary factors and breast cancer risk in Denmark Int J Cancer 1990 46 779 784 2228305
Franceschi S Favero A Decarli A Negri E La Vecchia C Ferraroni M Russo A Salvini S Amadori D Conti E Montella M Giacosa A Intake of macronutrients and the risk of breast cancer, Lancet 1996 347 1351 1356 8637339 10.1016/S0140-6736(96)91008-9
Graham S Hellmann R Marshall J Freudenheim J Vena J Swanson M Zielezny M Nemoto T Stubbe N Raimondo T Nutritional epidemiology of postmenopausal breast cancer in western New York Am J Epidemiol 1991 134 552 566 1951261
Hermann S Linseisen J Chang-Claude J Nutrition and breast cancer risk by age 50: a population based case-control study in Germany Nutr Cancer 2002 44 23 34 12672638 10.1207/S15327914NC441_4
Lee HP Gourley L Duffy SW Esteve J Lee J Day NE Dietary effects on breast-cancer risk in Singapore Lancet 1991 337 1197 1200 1673746 10.1016/0140-6736(91)92867-2
London SJ Sacks FM Stampfer MJ Henderson IC Maclure M Tomita A Wood WC Remine S Robert NJ Dmochowski JR Fatty acidcomposition of the subcutaneous adipose tissue and risk of proliferative benign breast disease and breast cancer J Natl Cancer Inst 1993 85 785 793 8487323
Potischman N Weiss HA Swanson CA Coates RJ Gammon MD Malone KE Brogan D Stanford JL Hoover RN Brinton LA Diet during adolescence and risk of breast cancer among young women J Natl Cancer Inst 1998 90 226 233 9462680 10.1093/jnci/90.3.226
Richardson S Gerber M Cenee S The role of fat, animal protein and some vitamin consumption in breast cancer: A case-control study in Southern France Int J Cancer 1991 48 1 9 2019449
Ronco A De Stefani E Mendilaharsu M Denno-Pellegrini H Meat, fat and risk of breast cancer: a case-control study from Uruguay Int J Cancer 1996 65 328 331 8575853 10.1002/(SICI)1097-0215(19960126)65:3<328::AID-IJC9>3.0.CO;2-1
Sala E Warren R Duffy S Welch A Luben R Day N High risk mammographic parenchymal patterns and diet: a case-control study Br J Cancer 2000 83 121 126 10883679 10.1054/bjoc.2000.1151
Santiago E Gonzalez MJ Matos MI Perez CM Association between dietary fat and breast cancer in Puerto Rican postmenopausal women attending a breast cancer clinic P R Health Sci J 1998 17 235 241 9883469
Van't Veer P Kok FJ Brants HA Ockhuizen T Sturmans F Hermus RJ Dietary fat and the risk of breast cancer Int J Epidemiol 1990 19 12 18 2351506
Yuan JM Wang QS Ross RK Henderson BE Yu MC Diet and breast cancer in Shanghai and Tianjin, China Br J Cancer 1995 71 1353 1358 7779738
Wakai K Dillon DS Ohno Y Prihartono J Budiningsih S Ramli M Darwis I Tjindarbumi D Tjahjadi G Soetrisno E Roostini ES Sakamoto G Herman S Cornain S Fat intake and breast cancer risk in an area where fat intake is low: a case-control study in Indonesia Int J Epidemiol 2000 29 20 28 10750599 10.1093/ije/29.1.20
Barrett-Connor E Friedlander NJ Dietary fat, calories, and the risk of breast cancer in postmenopausal women: a prospective population-based study J Am Coll Nutr 1993 12 390 399 8409101
Bingham SA Luben R Welch A Wareham N Khaw KT Day N Are imprecise methods obscuring a relation between fat and breast cancer Lancet 2003 362 212 214 12885485 10.1016/S0140-6736(03)13913-X
Byrne C Rockett H Holmes MD Dietary fat, fat subtypes, and breast cancer risk: lack of an association among postmenopausal women with no history of benign breast disease Cancer Epidemiol Biomarkers Prev 2002 11 261 265 11895875
Gago-Dominiguez M Yuan JM Sun CL Lee HP Yu MC Opposing effects of dietary n-3 and n-6 fatty acids on mammary carcinogenesis: The Singapore Chinese Health Study Br J Cancer 2003 89 1686 1692 14583770 10.1038/sj.bjc.6601340
Graham S Zielezny M Marshall J Priore R Freudenheim J Brasure J Haughey B Nasca P Zdeb M Diet in the epidemiology of postmenopausal breast cancer in the New York State Cohort Am J Epidemiol 1992 136 1327 1337 1336931
Kushi LH Sellers TA Potter JD Nelson CL Munger RG Kaye SA Folsom AR Dietary fat and post-menopausal breast cancer J Natl Cancer Inst 1992 84 1092 1099 1619683
Seiri S Krogh V Muti P Micheli A Pala V Crosignani P Berrino F Fat and protein intake and subsequent breast cancer risk in post-menopausal women Nutr Cancer 2002 42 10 17 12235640 10.1207/S15327914NC421_2
Velie E Kulldorff M Schairer C Block G Albanes D Schatzkin A Dietary fat, fat subtypes, and breast cancer in postmenopausal women: a prospective cohort study J Natl Cancer Inst 2000 92 833 839 10814679 10.1093/jnci/92.10.833
Voorrips LE Brants HA Kardinaal AFM van den Brant PA Goldbohm RA Intake of conjugated linoleic acid, fat and other fatty acids in relation to post-menopausal breast cancer: the Netherlands cohort study on diet and cancer Am J Clin Nutr 2002 76 873 882 12324303
Wirfalt E Mattisson I Gullberg B Johansson U Olsson H Berglund G Post-menopausal breast cancer is associated with high intakes of ω-6 fatty acids (Sweden) Cancer Causes Control 2003 13 883 893 12588084 10.1023/A:1021922917489
Howe GR Friedenreich CM Jain M Miller AB A cohort study of fat intake and risk of breast cancer J Natl Cancer Inst 1991 83 336 340 1995917
Cho E Spieglman D Hunter DJ Chen WY Stampfer MJ Colditz GA Willett WC Premenopausal fat intake and risk of breast cancer J Natl Cancer Inst 2003 95 1079 1085 12865454
Gaard M Tretli S Loken EB Dietary fat and the risk of breast cancer: A prospective study of 25,892 Norwegian women, Int J Cancer 1995 63 13 17 7558440
Knekt P Albanes D Seppanen R Aromaa A Jarvinen R Hyvonen L Teppo L Pukkala E Dietary fat and risk of breast cancer Am J Clin Nutr 1990 52 903 908 2239766
Rohan TE Jain M Miller AB A case-cohort study of diet and risk of benign proliferative epithelial disorders of the breast (Canada) Cancer Causes Control 1998 9 19 27 9486460 10.1023/A:1008841118358
Toniolo P Riboli E Shore RE Pasternack BS Consumption of meat, animal products, protein, and fat and risk of breast cancer: a prospective cohort study in New York Epidemiology 1994 5 391 397 7918807
van den Brandt PA van't Veer P Goldbohm RA Dorant E Volovics A Hermus RJ Sturmans F A prospective cohort study on dietary fat and the risk of postmenopausal breast cancer Cancer Res 1993 53 75 82 8416752
Holmes MD Hunter DJ Colditz GA Stampfer MJ Hankinson SE Speizer FE Rosner B Willett WC Association of dietary intake of fat and fatty acids with risk of breast cancer JAMA 1999 281 914 920 10078488 10.1001/jama.281.10.914
Horn-Ross PL Hoggatt KJ West DW Krone MR Stewart SL Anton-Culver H Bernstein L Deapen D Peel D Pinder R Reynolds P Ross RK Wright W Ziogas A Recent diet and breast cancer risk: the California Teachers Study (USA) Cancer Causes Control 2002 13 407 415 12146845 10.1023/A:1015786030864
Willett WC Hunter DJ Stampfer MJ Colditz G Manson JE Spiegelman D Rosner B Hennekens CH Speizer FE Dietary fat and fiber in relation to risk of breast cancer. An 8-year follow-up JAMA 1992 268 2037 2044 1328696 10.1001/jama.268.15.2037
Favero A Parpinel M Montella M Energy sources and risk of cancer of the breast and colon-rectum in Italy Adv Exp Med Biol 1999 472 51 55 10736615
Cade J Thomas E Vail A Case-control study of breast cancer in south east England: nutritional factors J Epidemiol Community Health 1998 52 105 110 9578857
Decarli A Favero A La Vecchia C Russo A Ferraroni M Negri E Franceschi S Macronutrients, energy intake, and breast cancer risk: implications from different models Epidemiology 1997 8 425 428 9209858 10.1097/00001648-199707000-00013
La Vecchia C Favero A Franceschi S Monounsaturated and other types of fat, and the risk of breast cancer Eur J Cancer Prev 1998 7 461 464 9926294
Katsouyanni K Trichopoulou A Stuver S Garas Y Kritselis A Kyriakou G Stoikidou M Boyle P Trichopoulos D The association of fat and other macronutrients with breast cancer: a case-control study from Greece Br J Cancer 1994 70 537 541 8080743
Klein V Chajes V Germain E Schulgen G Pinault M Malvy D Lefrancq T Fignon A Le Floch O Lhuillery C Bougnoux P Low alpha-linolenic acid content of adipose breast tissue is associated with an increased risk of breast cancer Eur J Cancer 2000 36 335 340 10708934 10.1016/S0959-8049(99)00254-3
Saadatian-Elahi M Toniolo P Ferrari P Goudable J Akhmedkhanov A Zeleniuch-Jacquotte A Riboli E Serum fatty acids and risk of breast cancer in a nested case-control study of the New York University Women's Health Study Cancer Epidemiol Biomarkers Prev 2002 11 1353 1360 12433711
Wolk A Bergstrom R Hunter D Willett W Ljung H Holmberg L Bergkvist L Bruce A Adami HO A prospective study of association of monounsaturated fat and other types of fat with risk of breast cancer Arch Intern Med 1998 158 41 45 9437377 10.1001/archinte.158.1.41
Nkondjock A Shatenstein B Ghadirian P A case-control study of breast cancer and dietary intake of individual fatty acids and antioxidants in Montreal, Canada Breast 2003 12 131 138 10.1016/S0960-9776(02)00284-9
Martin-Moreno JM Willett WC Gorgojo L Banegas JR Rodriguez-Artalejo F Fernandez-Rodriguez JC Maisonneuve P Boyle P Dietary fat, olive oil intake and breast cancer risk Int J Cancer 1994 58 774 780 7927867
Simonsen NR Fernandez-Crehuet Navajas J Martin-Moreno JM Strain JJ Huttunen JK Martin BC Thamm M Kardinaal AF van't Veer P Kok FJ Kohlmeier L Tissue stores of individual monounsaturated fatty acids and breast cancer: the EURAMIC study. European Community Multicenter Study on Antioxidants, Myocardial Infarction, and Breast Cancer Am J Clin Nutr 1998 68 134 141 9665107
Trichopoulou A Katsouyanni K Stuver S Tzala L Gnardellis C Rimm E Trichopooulos D Consumption of olive oil and specific food groups in relation to breat cancer risk in Greece J Natl Cancer Inst 1995 87 110 116 7503842
Bonilla-Fernandez P Lopez-Cervantes M Torres-Sanchez LE Tortolero-Luna G Lopez-Carrillo L Nutritional factors and breast cancer in Mexico Nutr Cancer 2003 45 148 155 12881007 10.1207/S15327914NC4502_02
Bagga D Anders KH Wang HJ Glaspy JA Long-chain n-3 to n-6 polyunsaturated fatty acid ratios in breast adipose tissue from women with and without breast cancer Nutr Cancer 2002 42 180 185 12416257 10.1207/S15327914NC422_5
Maillard V Bougnoux P Ferrari P Jourdan ML Pinault M Lavillonniere F Body G Le Floch O Chajes V N-3 and n-6 fatty acids in breast adipose tissue and relative risk of breast cancer in a case-control study in Tours, France Int J Cancer 2002 98 78 83 11857389 10.1002/ijc.10130
Owen RW Giacosa A Hull WE Haubner R Wurtele G Spiegelhalder B Bartsch H Olive-oil consumption and health: the possible role of antioxidants Lancet Oncol 2000 1 107 112 11905662 10.1016/S1470-2045(00)00015-2
Wahrburg U Assmann G Properties of olive oil Lancet 2001 357 1626 11408959 10.1016/S0140-6736(00)04774-7
Kohlmeier L Biomarkers of fatty acid exposure and breast cancer risk Am J Clin Nutr 1997 66 1548S 1556S 9394714
Mannisto S Pietinen P Virtanen M Kataja V Uusitupa M Diet and the risk of breast cancer in a case-control study: does the threat of disease have an influence on recall bias? J Clin Epidemiol 1999 52 429 439 10360338 10.1016/S0895-4356(99)00010-4
Saadatian-Elahi M Norat T Goudable J Riboli E Biomarkers of dietary fat intake and the risk of breast cancer: a meta-analysis Int J Cancer 2004 111 584 591 15239137 10.1002/ijc.20284
Hunter DJ Spiegelman D Adami HO Beeson L van den Brandt PA Folsom AR Fraser GE Goldbohm RA Graham S Howe GR Cohort studies of fat intake and the risk of breast cancer- a pooled analysis N Engl J Med 1996 334 356 361 8538706 10.1056/NEJM199602083340603
Franceschi S Favero A The role of energy and fat in cancers of the breast and colon-rectum in a southern European population Ann Oncol 1999 10 61 63 10676554 10.1023/A:1008384302955
Willett WC Fat, energy and breast cancer J Nutr 1997 127 921S 923S 9164264
Key TJ Allen NE Spencer EA Travis RC Nutrition and breast cancer Breast 2003 12 412 416 14659114 10.1016/S0960-9776(03)00145-0
Prentice RL Future possibilities in the prevention of breast cancer: fat and fiber and breast cancer research Breast Cancer Res 2000 2 268 276 11250720 10.1186/bcr68
Hunter DJ Willett WC Nutrition and breast cancer Cancer Causes Control 1996 7 56 68 8850435 10.1007/BF00115638
Zock PL Dietary fats and cancer Curr Opin Lipidol 2001 12 5 10 11176196 10.1097/00041433-200102000-00002
Law M Dietary fat and adult diseases and the implications for childhood nutrition: an epidemiologic approach Am J Clin Nutr 2000 72 1291S 1296S 11063471
Snyderwine EG Diet and mammary gland carcinogenesis Recent Results Cancer Res 1998 152 3 10 9928542
Santiago E Gonzalez MJ Matos MI Perez CM Dietary fat and breast cancer: a brief update on current knowledge P R Health Sci J 1998 17 273 279 9883473
Wynder EL Cohen LA Muscat JE Winters B Dwyer JT Blackburn G Breast cancer: weighing the evidence for a promoting role of dietary fat J Natl Cancer Inst 1997 89 766 775 9182974 10.1093/jnci/89.11.766
Richter WO Fatty acids and breast cancer – is there a relationship? Eur J Med Res 2003 8 373 380 12915333
Rock CL Demark-Wahnefried W Nutrition and survival after the diagnosis of breast cancer: a review of the evidence J Clin Oncol 2002 20 3302 3316 12149305 10.1200/JCO.2002.03.008
La Guardia M Giammanco M Breast cancer and obesity Panminerva Med 2001 43 123 133 11449184
Howe GR Dietary fat and breast cancer risks. An epidemiologic perspective Cancer 1994 74 1078 1084 8039142
Willett WC Dietary fat and breast cancer Toxicol Sci 1999 52 127 146 10630601
Kushi L Giovannucci E Dietary fat and cancer Am J Med 2002 113 63S 70S 12566141 10.1016/S0002-9343(01)00994-9
Feldman EB Breast cancer risk and intake of fat Nutr Rev 1999 57 353 356 10628187
Prentice RL Sheppard L Dietary fat and cancer: Consistency of epidemiologic data and disease prevention that may follow from a practical reduction in fat consumption Cancer Causes Control 1990 1 81 97 2102280 10.1007/BF00053187
|
16022739
|
PMC1199629
|
CC BY
|
2021-01-04 16:39:05
|
no
|
World J Surg Oncol. 2005 Jul 18; 3:45
|
utf-8
|
World J Surg Oncol
| 2,005 |
10.1186/1477-7819-3-45
|
oa_comm
|
==== Front
World J Surg OncolWorld Journal of Surgical Oncology1477-7819BioMed Central London 1477-7819-3-451602273910.1186/1477-7819-3-45ReviewDietary fat and risk of breast cancer Binukumar Bhaskarapillai [email protected] Aleyamma [email protected] Division of Epidemiology and Clinical Research, Regional Cancer Centre, Thiruvananthapuram – 695011 Kerala, India2005 18 7 2005 3 45 45 29 11 2004 18 7 2005 Copyright © 2005 Binukumar and Mathew; licensee BioMed Central Ltd.2005Binukumar and Mathew; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Breast cancer is one of the major public health problems among women worldwide. A number of epidemiological studies have been carried out to find the role of dietary fat and the risk of breast cancer. The main objective of the present communication is to summarize the evidence from various case-control and cohort studies on the consumption of fat and its subtypes and their effect on the development of breast cancer.
Methods
A Pubmed search for literature on the consumption of dietary fat and risk of breast cancer published from January 1990 through December 2003 was carried out.
Results
Increased consumption of total fat and saturated fat were found to be positively associated with the development of breast cancer. Even though an equivocal association was observed for the consumption of total monounsaturated fatty acids (MUFA) and the risk of breast cancer, there exists an inverse association in the case of oleic acid, the most abundant MUFA. A moderate inverse association between consumption of n-3 fatty acids and breast cancer risk and a moderate positive association between n-6 fatty acids and breast cancer risk were observed.
Conclusion
Even though all epidemiological studies do not provide a strong positive association between the consumption of certain types of dietary fat and breast cancer risk, at least a moderate association does seem to exist and this has a number of implications in view of the fact that breast cancer is an increasing public health concern.
==== Body
Background
Breast cancer is the most common malignancy amongst women worldwide, constituting 10% of all cancers. It is estimated that more than 1.1 million new breast cancer cases occur worldwide annually representing over twenty percent of malignancies in women [1]. Incidence rates of breast cancer are approximately 90–130 per 100,000 women in developed countries and those in developing countries are approximately ten to sixty per 100,000 women [1]. Several studies have been carried out to identify the risk factors for developing breast cancer. Studies of reproductive factors suggest that nulliparity and late age at first childbirth are the most consistent risk factors associated with breast cancer [2]. Certain dietary factors such as a higher intake of fat and meat also seem to increase the risk of breast cancer [3,4]. Fat is a macronutrient and is considered to be a major source of calories or energy. Although some fat in the diet is necessary, too much of fat can lead to heart diseases, cancers, obesity and other health problems. Numerous studies in women, using different study designs and in different geographical areas have been carried out in order to establish the relationship of dietary fat to breast cancer risk.
The objective of the present communication is to summarize evidence from various case-control and cohort studies on the consumption of dietary fat and its sub-types [saturated, monounsaturated (MUFA) and polyunsaturated (PUFA) fatty acids] and their effect on the development of breast cancer. The major sources of saturated fats are meat, poultry, dairy products, eggs and some plant foods such as coconut, coconut oil and palm oil [5]. High concentrations of MUFA are found in vegetable oils and their traces are also found in meat. Oleic acid, the most abundant MUFA, is found in animal and vegetable oils. Oleic acid is the major component of olive oil. PUFAs are classified into two families – n-3 PUFAs and n-6 PUFAs. PUFAs are mostly found in plants, fish and seafoods. Rich sources of n-3 PUFA include mackerel, salmon, and albacore tuna [6]. Types of n-3 fatty acids include eicosapentaenoic acid (EPA; 20:5 n-3), docosahexaenoic acid (DHA; 22:6 n-3), α-linolenic acid (18:3 n-3) and stearidonic acid (18:4 n-3). EPA and DHA are present in fatty fish although it can be obtained from enriched eggs [7]. N-6 fatty acids are found in high concentration in corn oil, safflower oil, soybean oil, sunflower oil and cottonseed oil. Types of n-6 fatty acids include linoleic acid (18:2 n-6), arachidonic acid (20:4 n-6), docosapentaenoic acid (DPA; 22:5 n-6) and γ-linolenic acid (18:3 n-6) [8].
Methods
A Pubmed search of the literature was carried out covering studies conducted over a period of fourteen years (from 1990 to 2003) using keywords "dietary fat" and "risk" and "breast cancer". The information on study design, country, data collection method, study period, sample size, study subjects, confounding variables, partition values used for comparisons, odds ratio/relative risk and trend p-value (if available) were obtained from the articles. Only case-control and cohort studies conducted among women were considered for the present review. Reports from symposia, genetic studies, survival/mortality studies of breast cancer patients and all studies associated with breast cancer other than dietary fat in women were excluded from the present review.
Total dietary fat
Case-control studies
Of the seventeen case-control studies [6,9-24] that investigated the relation of total dietary fat with risk of breast cancer, seven studies reported a significant positive association between total fat intake and risk of breast cancer [10,11,17-19,22,24]. The odds ratios (OR) of these studies ranged from 1.45 to 8.47 in the highest category of total fat consumption. Significant dose-response relationships between the increased consumption of total fat and breast cancer risk were observed in four studies [11,18,22,24]. Increased risk with border line significance was found for the intake of fat meat in one study [OR = 1.18 (95% CI: 1.0–1.5) in the highest quartile of fat meat intake] [17]. Non-significant increased risks were reported in six studies [6,14,16,20,21,23]. Significant reduced risk with a dose-response relationship was observed in one study [12]. Three studies have shown no association [9,13,15] (Additional file 1).
Cohort studies
Of the twenty cohort studies, which investigated the relationship between total dietary fat intake and breast cancer risk, ten were based exclusively on post-menopausal women [25-34]. Five studies showed significant positive associations with dose-response relationships between total fat and breast cancer risk [25,26,31,34,35]. The relative risks for these studies ranged from 1.15 to 3.47 in the highest level of total fat consumption. Non-significant positive associations between total fat and the risk of breast cancer were observed in nine studies [30,32,33,36-41]. Six studies found no association between the fat intake and the risk of breast cancer [27-29,42-44] (Additional file 2).
Components of fat
Saturated fat
Case-control studies
Four studies out of the seventeen reviewed here found significant positive associations with dose-response relationships between saturated fat and breast cancer [9,18,19,45]. The odds ratios for these studies ranged from 1.90 to 2.38. Nine studies showed non-significant increase in associations of saturated fat with the development of breast cancer [12,14,16,21,23,46-48,66]. Four studies reported no association between saturated fat and breast cancer risk [6,15,49,50] (Additional file 1).
Cohort studies
Among the seventeen cohort studies, three reported positive associations with significant dose-response relationships between the consumption of saturated fat and the risk of breast cancer [26,36,41]. The relative risks for these studies ranged from 1.13 to 1.47 in the highest level of saturated fat consumption. Non-significant increased associations for breast cancer were found in eight studies [30-33,37,38,40,51]. Six studies found no associations between the consumption of saturated fat and the risk of breast cancer [27,28,34,42,43,52] (Additional file 2).
Monounsaturated fat -Oleic acid
Case-control studies
Two studies of increased consumption of oleic acid (the most abundant MUFA) [12,57] and four studies of increased consumption of olive oil (a major component of oleic acid) [48,54-56] reported reduced risks of breast cancer. No association was observed in one study for the consumption of oleic acid [53] (Additional file 1).
Cohort studies
Some cohort studies found that increased consumption of oleic acid reduced the risk of breast cancer but the associations were non-significant [32,33,43]. However, other studies have shown no association [27,40] (Additional file 2).
Total monounsaturated fat
Case-control studies
Contrary to the studies on oleic acid consumption, three of the twelve case-control studies reported positive associations with a dose-response relationship between increased consumption of total MUFA and the risk of breast cancer [16,18,23]. The odds ratios for these studies ranged from 1.04 to 1.89 in the highest level of MUFA consumption. Non-significant increased associations for higher consumption of MUFA were found in three studies [6,21,23]. The rest of the studies reviewed reported no association between MUFA consumption and subsequent risk of breast cancer [14,15,46,48-50] (Additional file 1).
Cohort studies
Of the twelve studies that have been reviewed, three showed significant positive associations with dose-response relationships for increased intake of MUFA and the risk of breast cancer [34,37,38]. The relative risks for these studies ranged from 1.72 to 2.01 in the highest level of MUFA consumption. Five studies reported non-significant increased breast cancer risk [28,30,31,36,51]. Three studies showed inverse associations with significant dose-response relationships between the higher intake of MUFA and the risk of breast cancer [33,42,52]. The relative risks of these studies varied from 0.61 to 0.94. No significant association between the consumption of MUFA and breast cancer was found in one study [41] (Additional file 2).
Total polyunsaturated fat
Case-control studies
Six studies out of fourteen reported a significantly reduced breast cancer risk with increased PUFA consumption [12,15,45,48,57,66]. The odds ratios for these studies ranged from 0.39 to 0.70 in the highest category of PUFA consumption. Decreased breast cancer risk with significant dose response patterns was reported for high consumption of PUFA in three studies [12,15,57]. No associations were found between PUFA intake and risk of breast cancer in eight studies [6,14,16,18,21,23,46,49] (Additional file 1).
Cohort studies
Contrary to the results based on case-control studies, three out of eleven cohort studies reported significant positive associations with dose-response relationship between the increased consumption of PUFA and the risk of breast cancer [30,34,52]. The relative risks of these studies ranged from 1.49 to 3.02. Four studies reported non-significant increased risks associated with increased PUFA consumption [28,31,36,38]. Four studies reported no association between the consumption of PUFA and risk of breast cancer [33,41,42,51] (Additional file 2).
n-3 fatty acids
Case-control studies
One study of n-3 fatty acids [58] and another study of DHA reported significantly reduced risks for breast cancer [59] (Additional file 1). A significantly reduced breast cancer risk with a dose-response relationship was observed in one study for the increased consumption of DHA (n-3 fatty acid) [59]. Two studies of consumption of alpha-linolenic acid showed inverse association with a significant dose-response relationship [50,59]. One study of EPA (n-3 fatty acid) showed an inverse association [58]. No association was observed in other studies for the increased consumption of n-3 fatty acids [6,16,53,55] with the development of breast cancer (Additional file 1).
Cohort studies
Significantly reduced risk of breast cancer was observed in one study for increased consumption of n-3 fatty acids (RR = 0.72 in the highest quartile of intake) [28]. Contrary to the above result, a Swedish cohort study among post-menopausal women reported a significantly increased breast cancer risk with dose-response relationship [34]. The other studies showed no associations for breast cancer incidence on the consumption of n-3 fatty acids [36,51], DHA [33] and EPA [33,51] (Additional file 2).
n-6 fatty acids
Case-control studies
One study from France of linoleic acid (n-6 fatty acid) consumption reported a highly significant increased breast cancer risk (OR = 2.31 in the highest category of consumption of linoleic acid) [59]. Two studies of n-6 fatty acids reported non-significant increased breast cancer risks [55,58]. No associations were observed for the consumption of n-6 fatty acids [50,53,59] with the development of breast cancer (Additional file 1).
Cohort studies
One study conducted in Sweden among postmenopausal women reported a significantly increased breast cancer risk with dose-response relationship on the consumption of n-6 fatty acids [OR = 3.02 (95% CI: 1.78–5.13) in the highest quintile] [34]. Non-significant increased breast cancer risks were observed with increased consumption of n-6 fatty acids [28] and linoleic acid (n-6 fatty acid) [31,33]. A few studies found no risk of breast cancer with the consumption of linoleic acid [27,32,40,43] and arachidonic acid (n-6 fatty acid) [33] (Additional file 2).
Discussion
Most of the case-control studies discussed here have shown increased breast cancer risk with increased total fat consumption. However, results of cohort studies did not offer the same degree of consistent support.
Case-control and cohort studies share some strengths and limitations. The primary advantages are the ability to measure diet and potential confounding factors relatively accurately and uniformly at the individual level in both the types of study designs. Also, the subjects in most of the studies were generally drawn from a relatively uniform population. Despite these advantages, there remains a number of limitations common to both case-control and cohort studies.
As fat intake is estimated by food frequency questionnaire (FFQ) in most of the studies discussed in the present review, there remains a certain amount of measurement error in such procedures. Non-differential random measurement error (i.e. independent of case or non-case status) in estimated fat intake might bias estimated risks toward the null and concurrent measurement error in potential confounding variables (e.g., other sources of energy intake) could bias results in either direction. Only a few studies mentioned about validated FFQ [6,14,24,28,33,53,66]. Non-validated questionnaire for collecting the dietary information might also affect the reliability of study results. Another important limitation identified by the present review is that most of the studies are generally conducted in a relatively homogenous population; hence the range of fat intake and the power of the study to detect a true association is restricted. Due to the lack of a definitive biologic mechanism and the lack of information on the relevant timing of exposure, it is possible that case-control and cohort studies are measuring diet at the wrong point in time; this again might attenuate the observed risk in any particular study.
The main contrast between case-control and cohort studies lies in their differing potential for bias and in the resources required. Selection bias would occur if participation in a case-control study was correlated with fat intake and if cases and controls have different participation rates. In many of the case-control studies discussed here, particularly those conducted in highly industrialized populations such as the United States, there is often a substantial refusal to participate, particularly among control subjects, and it is possible that there is preferential participation by health-conscious individuals, who might have a relatively low fat intake. Such a phenomenon might introduce an artificial positive association between fat intake and breast cancer risk. Recall bias might occur since cases are interviewed after diagnosis and might report their past diet differently to control subjects. In particular, they could report their diets more accurately through better motivation, or conversely, they could report an over inflated fat consumption if aware of the postulated association between fat intake and breast cancer risk.
Since there are many limitations associated with case-control studies, the results from the cohort studies can be considered more authentic. Cohort studies assessing the relationship between total dietary fat and risk of breast cancer have been used since the late 1980's as an alternative to the inconsistent findings of the case-control studies. These are normally considered free of the most common biases that potentially affect case-control studies such as selection and information bias.
Even though most of the cohort studies failed to find an association between total dietary fat intake and breast cancer risk, some studies reported a significant positive association with dose-response relationship [25,26,31,34,35]. The failure of cohort studies to show a consistent and a strong relationship may be attributed to the difficulty in collecting accurate dietary information. The methodological limitations associated with the study designs and inaccuracies in the measurement of fat may have obscured any relationship between fat and breast cancer that might exist.
When we consider the components of fat, saturated fat, the unhealthiest out of all the types of fat, when consumed in large amounts increases the risk of breast cancer. In the present review, several case-control studies have shown positive association with breast cancer risk [9,18,19,45]. However, there are only a few cohort studies [26,36,41] reported statistically significant increased breast cancer risk with increased saturated fat consumption.
Monounsaturated fats (MUFA) are considered as 'non-essential' because they can be synthesized within our bodies, from foods and oils such as oleic acid and olive oil. Even though a few studies showed a positive correlation between MUFA and breast cancer [16,18,34,37], most of the studies reviewed here have shown that oleic acid including olive oil has an inverse relationship with breast cancer [12,16,18,33,34,37]. Many previous epidemiological studies have pointed out the preventive properties of olive oil due to its high content of oleic acid [56,60,61].
In the present review on polyunsaturated fatty acids, we focused mainly on n-6 and n-3 families. Various studies have analyzed the possible relation of (n-3)/(n-6) PUFA ratio and the possible role that it may play in the risk of breast cancer [6,58]. Most of the studies reviewed have shown that a higher (n-3)/(n-6) PUFA ratio may reduce the risk of breast cancer [6,58,59,62]. Some studies, however, found results which suggests that high intake of monounsaturated fatty acids, n-3 fatty acids and n-6 fatty acids were related to lower risk [63] and the same study, which primarily investigated the role of recall bias, found no relation between PUFA and breast cancer risk.
Boyd and others [4] carried out a meta analysis of all published studies up to July 2003. They found that 14 cohort and 31 case-control studies gave similar results and suggested that total and saturated fat intake are positively associated with increased risk of breast cancer. Saadatian-Elahi and others [64] analyzed 3 cohort and 7 case-control studies including 2,031 cases and 2,334 controls and concluded that n-3 PUFA had a protective effect and total MUFA and oleic acid levels were significantly associated with an increase of breast cancer risk. In a pooled analysis, Hunter and others [65] gathered information about 4,980 cases from studies including 337,819 women and found that there was no evidence of association between total dietary fat intake and the risk of breast cancer. The meta-analytical studies arrived at a conclusion similar to the present review and suggest that certain dietary fats are associated with breast cancer risk. Only Hunter et al, [65] have differed in that they found no association. Prentice and Sheppard [83] have, in fact, concluded that 50% reduction of dietary fat from current USA consumption levels would result in a 2.5 fold reduced risk of breast cancer among postmenopausal women.
Previously published review papers reported increased risk of breast cancer with the increased consumption of total dietary fat [68,69,71-75]; saturated fat [69,76-79] and moderately decreased association with oleic acid and monounsaturated fatty acids [67,70,80,81]. These results are consistent with our present review except for an equivocal association observed in the case of monounsaturated fatty acids. A few reviews have shown no association of various types of fat with the etiology of breast cancer [76,82]. Previous epidemiologic evidence provides little support for any important relation between intake of either linoleic acid (n-6 fatty acid) or n-3 fatty acids and risk of breast cancer [67] as against a moderate inverse association between the increased consumption of n-3 fatty acids and the risk of breast cancer and a moderate positive association between the increased consumption of n-6 fatty acids and the risk of breast cancer.
Conclusion
Considering the totality of epidemiologic evidence, it is reasonable to conclude that a positive association exists between total dietary fat intake and breast cancer risk, even though all epidemiological studies do not provide a strong association between consumption of dietary fat and breast cancer risk. Also, increased consumption of saturated fat is positively associated with the development of breast cancer. There exists an inverse association of breast cancer risk with the consumption of oleic acid. A moderate inverse association between breast cancer risk and the consumption of n-3 fatty acids and a moderate positive association between n-6 fatty acids and breast cancer risk were also observed. These have lot of implications in view of the fact that breast cancer is a growing public health problem.
Competing interests
The present review paper will be quite useful for surgeons, clinicians, epidemiologists and other public health professionals particularly working in the filed of cancer.
Authors' contributions
BB – collected the literature, summarized the results in the form of tables and wrote the paper
AM: provided technical guidance for writing the paper and edited the paper
Funding sources
Nil
Supplementary Material
Additional File 1
Table showing case control studies and risk of breast cancer.
Click here for file
Additional File 2
Table showing cohort studies and risk of breast cancer.
Click here for file
==== Refs
Parkin DM Bray F Ferlay J Pisani P Estimating the world cancer burden: Globocon 2000 Int J Cancer 2001 94 153 156 11668491 10.1002/ijc.1440
McPherson K Steel CM Dixon JM Breast cancer- epidemiology, risk factors, and genetics BMJ 2000 321 624 628 10977847 10.1136/bmj.321.7261.624
WCRF (World Cancer Research Fund) American Institute for Cancer Research Expert Panel; Food, Nutrition, and Prevention of Cancer: A Global Perspective Am Inst Cancer Res 1997 Washington, DC
Boyd NF Stone J Vogt KN Connelly BS Martin LJ Minkin S Dietary fat and breast cancer risk revisited: a meta-analysis of the published literature Br J Cancer 2004 111 584 591
Saturated fat last accessed on July 10, 2005
Goodstine SL Zheng T Holford TR Ward BA Carter D Owens PH Mayne ST Dietary (n-3)/(n-6) fatty acid ratio: possible relationship to premenopausal but not to post-menopausal breast cancer risk in U.S. women J Nutr 2003 133 1409 1414 12730430
Horrocks LA Yeo YK Health benefits of docosahexanoic acid Pharmacol Res 1999 40 211 225 10479465 10.1006/phrs.1999.0495
Lee D Thomas G Fats, fish oil and omega 3 fatty acids accessed on July 10, 2005
Do MH Lee SS Jung PJ Lee MH Intake of dietary fat and vitamin in relation to breast cancer risk in Korean women: a case-control study J Korean Med Sci 2003 18 534 540 12923330
D'Avanzo B Negri E Gramenzi A Franceschi S Parazzini F Boyle P La Vecchia C Fats in seasoning and breast cancer risk: an Italian case-control study Eur J Cancer 1991 27 420 423 1828170
Ewertz M Gill C Dietary factors and breast cancer risk in Denmark Int J Cancer 1990 46 779 784 2228305
Franceschi S Favero A Decarli A Negri E La Vecchia C Ferraroni M Russo A Salvini S Amadori D Conti E Montella M Giacosa A Intake of macronutrients and the risk of breast cancer, Lancet 1996 347 1351 1356 8637339 10.1016/S0140-6736(96)91008-9
Graham S Hellmann R Marshall J Freudenheim J Vena J Swanson M Zielezny M Nemoto T Stubbe N Raimondo T Nutritional epidemiology of postmenopausal breast cancer in western New York Am J Epidemiol 1991 134 552 566 1951261
Hermann S Linseisen J Chang-Claude J Nutrition and breast cancer risk by age 50: a population based case-control study in Germany Nutr Cancer 2002 44 23 34 12672638 10.1207/S15327914NC441_4
Lee HP Gourley L Duffy SW Esteve J Lee J Day NE Dietary effects on breast-cancer risk in Singapore Lancet 1991 337 1197 1200 1673746 10.1016/0140-6736(91)92867-2
London SJ Sacks FM Stampfer MJ Henderson IC Maclure M Tomita A Wood WC Remine S Robert NJ Dmochowski JR Fatty acidcomposition of the subcutaneous adipose tissue and risk of proliferative benign breast disease and breast cancer J Natl Cancer Inst 1993 85 785 793 8487323
Potischman N Weiss HA Swanson CA Coates RJ Gammon MD Malone KE Brogan D Stanford JL Hoover RN Brinton LA Diet during adolescence and risk of breast cancer among young women J Natl Cancer Inst 1998 90 226 233 9462680 10.1093/jnci/90.3.226
Richardson S Gerber M Cenee S The role of fat, animal protein and some vitamin consumption in breast cancer: A case-control study in Southern France Int J Cancer 1991 48 1 9 2019449
Ronco A De Stefani E Mendilaharsu M Denno-Pellegrini H Meat, fat and risk of breast cancer: a case-control study from Uruguay Int J Cancer 1996 65 328 331 8575853 10.1002/(SICI)1097-0215(19960126)65:3<328::AID-IJC9>3.0.CO;2-1
Sala E Warren R Duffy S Welch A Luben R Day N High risk mammographic parenchymal patterns and diet: a case-control study Br J Cancer 2000 83 121 126 10883679 10.1054/bjoc.2000.1151
Santiago E Gonzalez MJ Matos MI Perez CM Association between dietary fat and breast cancer in Puerto Rican postmenopausal women attending a breast cancer clinic P R Health Sci J 1998 17 235 241 9883469
Van't Veer P Kok FJ Brants HA Ockhuizen T Sturmans F Hermus RJ Dietary fat and the risk of breast cancer Int J Epidemiol 1990 19 12 18 2351506
Yuan JM Wang QS Ross RK Henderson BE Yu MC Diet and breast cancer in Shanghai and Tianjin, China Br J Cancer 1995 71 1353 1358 7779738
Wakai K Dillon DS Ohno Y Prihartono J Budiningsih S Ramli M Darwis I Tjindarbumi D Tjahjadi G Soetrisno E Roostini ES Sakamoto G Herman S Cornain S Fat intake and breast cancer risk in an area where fat intake is low: a case-control study in Indonesia Int J Epidemiol 2000 29 20 28 10750599 10.1093/ije/29.1.20
Barrett-Connor E Friedlander NJ Dietary fat, calories, and the risk of breast cancer in postmenopausal women: a prospective population-based study J Am Coll Nutr 1993 12 390 399 8409101
Bingham SA Luben R Welch A Wareham N Khaw KT Day N Are imprecise methods obscuring a relation between fat and breast cancer Lancet 2003 362 212 214 12885485 10.1016/S0140-6736(03)13913-X
Byrne C Rockett H Holmes MD Dietary fat, fat subtypes, and breast cancer risk: lack of an association among postmenopausal women with no history of benign breast disease Cancer Epidemiol Biomarkers Prev 2002 11 261 265 11895875
Gago-Dominiguez M Yuan JM Sun CL Lee HP Yu MC Opposing effects of dietary n-3 and n-6 fatty acids on mammary carcinogenesis: The Singapore Chinese Health Study Br J Cancer 2003 89 1686 1692 14583770 10.1038/sj.bjc.6601340
Graham S Zielezny M Marshall J Priore R Freudenheim J Brasure J Haughey B Nasca P Zdeb M Diet in the epidemiology of postmenopausal breast cancer in the New York State Cohort Am J Epidemiol 1992 136 1327 1337 1336931
Kushi LH Sellers TA Potter JD Nelson CL Munger RG Kaye SA Folsom AR Dietary fat and post-menopausal breast cancer J Natl Cancer Inst 1992 84 1092 1099 1619683
Seiri S Krogh V Muti P Micheli A Pala V Crosignani P Berrino F Fat and protein intake and subsequent breast cancer risk in post-menopausal women Nutr Cancer 2002 42 10 17 12235640 10.1207/S15327914NC421_2
Velie E Kulldorff M Schairer C Block G Albanes D Schatzkin A Dietary fat, fat subtypes, and breast cancer in postmenopausal women: a prospective cohort study J Natl Cancer Inst 2000 92 833 839 10814679 10.1093/jnci/92.10.833
Voorrips LE Brants HA Kardinaal AFM van den Brant PA Goldbohm RA Intake of conjugated linoleic acid, fat and other fatty acids in relation to post-menopausal breast cancer: the Netherlands cohort study on diet and cancer Am J Clin Nutr 2002 76 873 882 12324303
Wirfalt E Mattisson I Gullberg B Johansson U Olsson H Berglund G Post-menopausal breast cancer is associated with high intakes of ω-6 fatty acids (Sweden) Cancer Causes Control 2003 13 883 893 12588084 10.1023/A:1021922917489
Howe GR Friedenreich CM Jain M Miller AB A cohort study of fat intake and risk of breast cancer J Natl Cancer Inst 1991 83 336 340 1995917
Cho E Spieglman D Hunter DJ Chen WY Stampfer MJ Colditz GA Willett WC Premenopausal fat intake and risk of breast cancer J Natl Cancer Inst 2003 95 1079 1085 12865454
Gaard M Tretli S Loken EB Dietary fat and the risk of breast cancer: A prospective study of 25,892 Norwegian women, Int J Cancer 1995 63 13 17 7558440
Knekt P Albanes D Seppanen R Aromaa A Jarvinen R Hyvonen L Teppo L Pukkala E Dietary fat and risk of breast cancer Am J Clin Nutr 1990 52 903 908 2239766
Rohan TE Jain M Miller AB A case-cohort study of diet and risk of benign proliferative epithelial disorders of the breast (Canada) Cancer Causes Control 1998 9 19 27 9486460 10.1023/A:1008841118358
Toniolo P Riboli E Shore RE Pasternack BS Consumption of meat, animal products, protein, and fat and risk of breast cancer: a prospective cohort study in New York Epidemiology 1994 5 391 397 7918807
van den Brandt PA van't Veer P Goldbohm RA Dorant E Volovics A Hermus RJ Sturmans F A prospective cohort study on dietary fat and the risk of postmenopausal breast cancer Cancer Res 1993 53 75 82 8416752
Holmes MD Hunter DJ Colditz GA Stampfer MJ Hankinson SE Speizer FE Rosner B Willett WC Association of dietary intake of fat and fatty acids with risk of breast cancer JAMA 1999 281 914 920 10078488 10.1001/jama.281.10.914
Horn-Ross PL Hoggatt KJ West DW Krone MR Stewart SL Anton-Culver H Bernstein L Deapen D Peel D Pinder R Reynolds P Ross RK Wright W Ziogas A Recent diet and breast cancer risk: the California Teachers Study (USA) Cancer Causes Control 2002 13 407 415 12146845 10.1023/A:1015786030864
Willett WC Hunter DJ Stampfer MJ Colditz G Manson JE Spiegelman D Rosner B Hennekens CH Speizer FE Dietary fat and fiber in relation to risk of breast cancer. An 8-year follow-up JAMA 1992 268 2037 2044 1328696 10.1001/jama.268.15.2037
Favero A Parpinel M Montella M Energy sources and risk of cancer of the breast and colon-rectum in Italy Adv Exp Med Biol 1999 472 51 55 10736615
Cade J Thomas E Vail A Case-control study of breast cancer in south east England: nutritional factors J Epidemiol Community Health 1998 52 105 110 9578857
Decarli A Favero A La Vecchia C Russo A Ferraroni M Negri E Franceschi S Macronutrients, energy intake, and breast cancer risk: implications from different models Epidemiology 1997 8 425 428 9209858 10.1097/00001648-199707000-00013
La Vecchia C Favero A Franceschi S Monounsaturated and other types of fat, and the risk of breast cancer Eur J Cancer Prev 1998 7 461 464 9926294
Katsouyanni K Trichopoulou A Stuver S Garas Y Kritselis A Kyriakou G Stoikidou M Boyle P Trichopoulos D The association of fat and other macronutrients with breast cancer: a case-control study from Greece Br J Cancer 1994 70 537 541 8080743
Klein V Chajes V Germain E Schulgen G Pinault M Malvy D Lefrancq T Fignon A Le Floch O Lhuillery C Bougnoux P Low alpha-linolenic acid content of adipose breast tissue is associated with an increased risk of breast cancer Eur J Cancer 2000 36 335 340 10708934 10.1016/S0959-8049(99)00254-3
Saadatian-Elahi M Toniolo P Ferrari P Goudable J Akhmedkhanov A Zeleniuch-Jacquotte A Riboli E Serum fatty acids and risk of breast cancer in a nested case-control study of the New York University Women's Health Study Cancer Epidemiol Biomarkers Prev 2002 11 1353 1360 12433711
Wolk A Bergstrom R Hunter D Willett W Ljung H Holmberg L Bergkvist L Bruce A Adami HO A prospective study of association of monounsaturated fat and other types of fat with risk of breast cancer Arch Intern Med 1998 158 41 45 9437377 10.1001/archinte.158.1.41
Nkondjock A Shatenstein B Ghadirian P A case-control study of breast cancer and dietary intake of individual fatty acids and antioxidants in Montreal, Canada Breast 2003 12 131 138 10.1016/S0960-9776(02)00284-9
Martin-Moreno JM Willett WC Gorgojo L Banegas JR Rodriguez-Artalejo F Fernandez-Rodriguez JC Maisonneuve P Boyle P Dietary fat, olive oil intake and breast cancer risk Int J Cancer 1994 58 774 780 7927867
Simonsen NR Fernandez-Crehuet Navajas J Martin-Moreno JM Strain JJ Huttunen JK Martin BC Thamm M Kardinaal AF van't Veer P Kok FJ Kohlmeier L Tissue stores of individual monounsaturated fatty acids and breast cancer: the EURAMIC study. European Community Multicenter Study on Antioxidants, Myocardial Infarction, and Breast Cancer Am J Clin Nutr 1998 68 134 141 9665107
Trichopoulou A Katsouyanni K Stuver S Tzala L Gnardellis C Rimm E Trichopooulos D Consumption of olive oil and specific food groups in relation to breat cancer risk in Greece J Natl Cancer Inst 1995 87 110 116 7503842
Bonilla-Fernandez P Lopez-Cervantes M Torres-Sanchez LE Tortolero-Luna G Lopez-Carrillo L Nutritional factors and breast cancer in Mexico Nutr Cancer 2003 45 148 155 12881007 10.1207/S15327914NC4502_02
Bagga D Anders KH Wang HJ Glaspy JA Long-chain n-3 to n-6 polyunsaturated fatty acid ratios in breast adipose tissue from women with and without breast cancer Nutr Cancer 2002 42 180 185 12416257 10.1207/S15327914NC422_5
Maillard V Bougnoux P Ferrari P Jourdan ML Pinault M Lavillonniere F Body G Le Floch O Chajes V N-3 and n-6 fatty acids in breast adipose tissue and relative risk of breast cancer in a case-control study in Tours, France Int J Cancer 2002 98 78 83 11857389 10.1002/ijc.10130
Owen RW Giacosa A Hull WE Haubner R Wurtele G Spiegelhalder B Bartsch H Olive-oil consumption and health: the possible role of antioxidants Lancet Oncol 2000 1 107 112 11905662 10.1016/S1470-2045(00)00015-2
Wahrburg U Assmann G Properties of olive oil Lancet 2001 357 1626 11408959 10.1016/S0140-6736(00)04774-7
Kohlmeier L Biomarkers of fatty acid exposure and breast cancer risk Am J Clin Nutr 1997 66 1548S 1556S 9394714
Mannisto S Pietinen P Virtanen M Kataja V Uusitupa M Diet and the risk of breast cancer in a case-control study: does the threat of disease have an influence on recall bias? J Clin Epidemiol 1999 52 429 439 10360338 10.1016/S0895-4356(99)00010-4
Saadatian-Elahi M Norat T Goudable J Riboli E Biomarkers of dietary fat intake and the risk of breast cancer: a meta-analysis Int J Cancer 2004 111 584 591 15239137 10.1002/ijc.20284
Hunter DJ Spiegelman D Adami HO Beeson L van den Brandt PA Folsom AR Fraser GE Goldbohm RA Graham S Howe GR Cohort studies of fat intake and the risk of breast cancer- a pooled analysis N Engl J Med 1996 334 356 361 8538706 10.1056/NEJM199602083340603
Franceschi S Favero A The role of energy and fat in cancers of the breast and colon-rectum in a southern European population Ann Oncol 1999 10 61 63 10676554 10.1023/A:1008384302955
Willett WC Fat, energy and breast cancer J Nutr 1997 127 921S 923S 9164264
Key TJ Allen NE Spencer EA Travis RC Nutrition and breast cancer Breast 2003 12 412 416 14659114 10.1016/S0960-9776(03)00145-0
Prentice RL Future possibilities in the prevention of breast cancer: fat and fiber and breast cancer research Breast Cancer Res 2000 2 268 276 11250720 10.1186/bcr68
Hunter DJ Willett WC Nutrition and breast cancer Cancer Causes Control 1996 7 56 68 8850435 10.1007/BF00115638
Zock PL Dietary fats and cancer Curr Opin Lipidol 2001 12 5 10 11176196 10.1097/00041433-200102000-00002
Law M Dietary fat and adult diseases and the implications for childhood nutrition: an epidemiologic approach Am J Clin Nutr 2000 72 1291S 1296S 11063471
Snyderwine EG Diet and mammary gland carcinogenesis Recent Results Cancer Res 1998 152 3 10 9928542
Santiago E Gonzalez MJ Matos MI Perez CM Dietary fat and breast cancer: a brief update on current knowledge P R Health Sci J 1998 17 273 279 9883473
Wynder EL Cohen LA Muscat JE Winters B Dwyer JT Blackburn G Breast cancer: weighing the evidence for a promoting role of dietary fat J Natl Cancer Inst 1997 89 766 775 9182974 10.1093/jnci/89.11.766
Richter WO Fatty acids and breast cancer – is there a relationship? Eur J Med Res 2003 8 373 380 12915333
Rock CL Demark-Wahnefried W Nutrition and survival after the diagnosis of breast cancer: a review of the evidence J Clin Oncol 2002 20 3302 3316 12149305 10.1200/JCO.2002.03.008
La Guardia M Giammanco M Breast cancer and obesity Panminerva Med 2001 43 123 133 11449184
Howe GR Dietary fat and breast cancer risks. An epidemiologic perspective Cancer 1994 74 1078 1084 8039142
Willett WC Dietary fat and breast cancer Toxicol Sci 1999 52 127 146 10630601
Kushi L Giovannucci E Dietary fat and cancer Am J Med 2002 113 63S 70S 12566141 10.1016/S0002-9343(01)00994-9
Feldman EB Breast cancer risk and intake of fat Nutr Rev 1999 57 353 356 10628187
Prentice RL Sheppard L Dietary fat and cancer: Consistency of epidemiologic data and disease prevention that may follow from a practical reduction in fat consumption Cancer Causes Control 1990 1 81 97 2102280 10.1007/BF00053187
|
16042767
|
PMC1199630
|
CC BY
|
2021-01-04 16:39:04
|
no
|
World J Surg Oncol. 2005 Jul 21; 3:49
|
latin-1
|
World J Surg Oncol
| 2,005 |
10.1186/1477-7819-3-49
|
oa_comm
|
==== Front
BMC OphthalmolBMC Ophthalmology1471-2415BioMed Central London 1471-2415-5-191610518110.1186/1471-2415-5-19Case ReportIs Microsporidial keratitis an emerging cause of stromal keratitis? – a case series study Vemuganti Geeta K [email protected] Prashant [email protected] Savitri [email protected] Joveeta [email protected] Usha [email protected] Shashi [email protected] Ophthalmic Pathology Service, L.V. Prasad Eye Institute, Hyderabad, India2 Cornea Service, L.V. Prasad Eye Institute, Hyderabad, India3 Jhaveri Microbiology Centre, L.V. Prasad Eye Institute, Hyderabad, India4 Centre for Cellular and Molecular Biology, Hyderabad, India2005 17 8 2005 5 19 19 15 4 2005 17 8 2005 Copyright © 2005 Vemuganti et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Microsporidial keratitis is a rare cause of stromal keratitis. We present a series of five cases of microsporidial keratitis from a single centre in southern India with microbiologic and histopathologic features.
Case presentation
Patient charts of five cases of microsporidial stromal keratitis diagnosed between January 2002 and June 2004 were reviewed retrospectively for clinical data, microbiologic and histopathologic data. The presence of microsporidia was confirmed by special stains on corneal scrapings and/or corneal tissues, and electron microscopy. All patients were immunocompetent with a preceding history of trauma in three. Four patients presented with unilateral, small, persisting deep stromal infiltrates, of uncertain etiology, in the cornea, which were not responding to conventional antimicrobial treatment and required penetrating keratoplasty in three. Fifth case was unsuspected and underwent keratoplasty for post-traumatic scar. Three of five cases were diagnosed on corneal scrapings, prior to keratoplasty, while two were diagnosed only on histology. The microsporidia appeared as oval well defined bodies with dense staining at one pole. None of the patients showed recurrence following keratoplasty.
Conclusion
Microsporidia, though rare, should be suspected in chronic culture-negative stromal keratitis. Organisms could lie dormant without associated inflammation.
==== Body
Background
Microsporidia are small, oval, obligate intracellular eukaryotic protozoan parasites, widely distributed in vertebrates and invertebrates, but mostly cause infections in immunocompromised individuals [1-3]. The ocular manifestations include superficial punctate keratoconjunctivitis [4-6], and corneal stromal keratitis [7-14] and these two manifestations are directed by the genus involved as well as the immune status of the patient. Keratoconjunctivitis is usually seen in immunocompromised individuals or in contact lens wearers; mostly by genus Encephalitozoon while the stromal keratitis is caused by Nosema and Microsporidium. However, recent reports suggest that keratoconjunctivitis can also occur in immunocompetent individuals [15,16]. A review of the literature in English using PubMed revealed eight cases of corneal stromal involvement, all of which were diagnosed based on identification of the organisms in the tissue sections and/or by ultrastructural studies [7-13]. We present a series of five cases of corneal stromal keratitis from a single tertiary eye care centre, caused by Microsporidium, two of which were diagnosed based on the clinical and microbiologic findings. These cases highlight the importance of identifying the clinical presentations, microbiology, and histopathology including electron microscopic findings for proper management of these cases.
Methods
In a retrospective study, all cases of microsporidial keratitis, diagnosed and managed at L V Prasad Eye Institute, Hyderabad, India between January 2002 and July 2002 were included. The medical records of the patients were specifically reviewed for past and personal history, clinical findings. The microbiologic work-up of the corneal lesion included 1) corneal scrapings stained by Gram, Giemsa stain and potassium hydroxide- calcofluor white preparation, 1% acid fast stain, 2) culture of the scrapings in culture media to facilitate growth of bacteria, fungi and parasites. In cases that underwent therapeutic penetrating keratoplasty, the corneal button was bisected and submitted to microbiology and histopathology. Routine histology stains and special stains like 1% acid fast stain, modified trichrome stain, Gram and Gomori's methenamine silver stain were performed on the permanent sections. In three cases, part of the tissue was submitted in 2% glutaraldehyde for electron microscopy. The patients underwent systemic evaluation to rule out any evidence of immunosuppression, HIV testing of the serum by ELISA (after obtaining informed consent) was done in all cases.
Case presentation
In this study period, five cases of microsporidial keratitis were diagnosed at L V Prasad Eye Institute, Hyderabad, India. The mean age of patients was 34 (range 2–70 years) years with three male patients. Four cases presented as keratitis, two of which were diagnosed by microbiological processing of the corneal scrapings and two by tissue diagnosis, while one presented with a corneal scar with a post traumatic descemetocele with unsuspected infection. Brief clinical history of all cases (Table 1) is given below.
Case 1
A 40-year-old lady presented to our institute with recurrent pain, redness, watering and blurred vision of seven months duration. During this time she had received topical steroids, antiviral and antifungal medications. However, there was no resolution in the signs or symptoms. The treating surgeon felt she was worsening and referred her as a case of crystalline keratopathy due to the nature of the stromal infiltrate. At presentation, she had a BCVA of perception of light and accurate projection of rays. There was 1+ lid edema and conjunctival congestion. There was a central mid to deep stromal infiltrate measuring 1.5 × 1.5 mm with surrounding stromal edema (figure 1). The overlying epithelium was intact with epithelial edema. Endothelium showed exudates arranged as a sheet over an area of 5.0 × 5.0 mm. There was 360° of superficial vascularisation. Since the infiltrate had not shown any signs of resolution and there was no definitive diagnosis in hand, all the medications were discontinued and she underwent a penetrating keratoplasty for that eye on 25/03/02. The post-operative period was uneventful and at last follow up she was doing well with a BCVA of 6/18. She was treated with topical Chloramphenicol, Prednisolone acetate1%, and Lacrigel. She was later prescribed Tablet Itraconazole 100 mg BD for one month. Her immune status was found to be normal.
Case 2
Parents of a 2-year-old male child brought him with complaints of white spot on the black of the left eye and swelling of the lids since one and half months. The child had red eye three months back which subsided with eyedrops (Ofloxacin and Tobramycin) but the opacity persisted. Vision could not be assessed as he was photophobic. Examination under anaesthesia revealed flat lids and a quiet conjunctiva. The cornea showed the presence of a mid to deep stromal infiltrate 1.0 × 1.0 mm with surrounding cellular reaction. There was surrounding stromal edema extending about 2 mm from the infiltrate edges. AC was deep and quiet. The IOP was within normal limits. A diagnosis of resolving infectious keratitis or a HSV stromal immune keratitis was kept in mind and he was started on Cefazolin eye drops 2 hourly with close follow up. He showed improvement and 15 days later four hourly Betamet e/d was added. The child was comfortable. He returned two months later with a recurrence of the stromal infiltration in the same location. Deep corneal scrapings were done under anaesthesia which revealed spores suggestive of Microsporidium. He was started on oral Itraconazole 50 mg BD. After 15 days of therapy there was no response and considering the risk of amblyopia a penetrating keratoplasty was done on 8/04/02. The post operative period was uneventful and the child is doing well and is relieved of his symptoms as noticed by his parents.
Case 3
A 23 years male patient presented with a history of pulling sensation in the left eye off and on since 2 years. This was associated with pain, redness and watering. He had been diagnosed and treated as HSV keratitis for the same. He was presently on treatment with Acyclovir ointment 3 times a day and Tears plus eye drops 3 times a day. His visual acuity was counting fingers close to face. The lids were edematous with a diffuse papillary reaction of the conjunctiva. Cornea showed the presence of a central, irregular epithelial defect 2.5 × 3.0 mm with surrounding stromal edema. There was scarring at the edges and the corneal sensations were decreased. Rest of the findings was within normal limits. Viral scrapings were taken and he was started on Acyclovir eye ointment 5 times a day with cycloplegics. The scrapings were inconclusive for viral etiology, there was a clinical resolution and as the epithelial defect reduced in size, he was started on Betamet eye drops drop to drop with the antiviral. The lesion completely resolved and he was continued on topical lubricants.
He returned with a recurrence 15 days later and was again started on topical corticosteroid under antiviral cover. Oral Acyclovir was also started in a dosage of 400 mg 5 times day. The lesion again resolved and the antivirals were discontinued. However, the lubricant eye drops were continued and the corticosteroid was tapered. He was on regular follow-up and seven months later a recurrence was seen with a stromal infiltrate and the presence of endothelial exudates (figure 2). There were deep vessels seen in the inferior half of the cornea. The central area showed marked thinning and a tissue adhesive with bandage contact lens was applied. The corneal scrapings in Gram stain revealed oval bodies suggestive of microsporidial spores. He was started on oral Itraconazole 100 mg BD with topical ciprofloxacin and cycloplegics. Despite treatment he showed a shallowing of the AC suggestive of leak under the glue and the procedure was repeated. He developed 360° deep and superficial vascularisation. He was continued on the same treatment and is on regular follow-up. On the last follow-up visit the glue and BCL had come off and revealed a scar with thinning of the corneas in that region. He was advised to continue oral therapy for 2 weeks, and come for review.
Case 4
A 70-year old female presented with complaints of pain, redness, watering and diminution of vision in the left eye of one month duration. She gave history of injury with grass six months ago. Visual acuity in right eye was 6/12 and hand movements with counting finger in the left eye. Right eye on examination was within normal limits. Left eye lids were edematous, conjunctiva was congested, cornea showed the presence of an anterior mid stromal infiltrate in nasal paracentral area with overlying epithelial defect of 1.5 × 1.5 mm (figure 3). In the temporal paracentral area, there was an area of deep stromal infiltrate with underlying endothelial exudate. Smears revealed gram-negative oval bodies. Cultures were positive for gram-positive cocci and gram-positive bacilli. Patient was started on Cefazolin eye drops half hourly. The clinical picture was the same at one and half months, therefore a decision was taken to penetrating keratoplasty in the left eye. Post-operative period was uneventful.
Case 5
A 37-year-old male presented with history of injury in left eye during a fight 4 years ago. His vision decreased but he did not use any medications. On examination, vision in right eye was 6/6 and in the left eye it was PL positive, PR accurate. Right eye examination was within normal limits. Left eye showed trace conjunctival congestion, there was diffuse corneal scarring with central descemetocoele. Anterior chamber view was hazy and there was total cataract. He underwent optical penetrating keratoplasty with extracapsular cataract extraction with intraocular lens implantation under local anesthesia in left eye. The post-operative period was uneventful and at last follow-up his vision in the left eye was 20/400.
Microbiology
Four patients in this series had presented as ulcerative keratitis and one as post-traumatic corneal ectasia. Two of the four patients, diagnosed as stromal keratitis, underwent deep stromal scrapings, which demonstrated microsporidial spores. The diagnosis was made on observing oval refractile bodies in the corneal scrapings, both within cells as well as in extracellular location. The Gram (figure 4) and Giemsa stained smears revealed variable staining of the parasites. The oval body had a dark polar nucleus. The parasite was acid fast and appeared as red oval structure on smears stained with 1% acid fast stain (figure 5). Corneal scrapings stained with calcofluor white showed oval fluorescent bodies in clumps. Bacterial and fungal cultures from these cases did not reveal any organisms except in one case (case 4) that showed a significant growth of Staphylococcus epidermidis.
Treatment
Three cases were initially treated medically based on the findings of smear examination. In case 1 the treatment was started with fortified cefazolin and fortified gentamicin even though oval bodies were seen in the smear examination. In other two cases where scraping was performed the treatment was started with oral itraconazole 100 mg twice daily and topical ciprofloxacin 0.3%. The infiltrate resolved with medical therapy in one case. Even this case developed progressive thinning for which tissue adhesive and bandage contact lens was applied. Penetrating keratoplasty was performed in four cases. The indications were non-responding keratitis in three cases and thinning with ectasia in one case. Postoperatively all cases were managed with corticosteroids (Prednisolone acetate 1%). The graft was clear in all cases with no evidence of recurrence of infection.
Histopathology
Out of four corneal buttons included in the study three were from patients diagnosed as stromal keratitis and one was from a corneal scar with descemetocele, undergoing optical PK. The epithelium was intact in two, showed edema in one and was ulcerated in one case. Bowmans layer was destroyed in 3 out of 4 cases. There was moderate to severe stromal inflammation, at places forming micro abscess (fig 6). The stromal inflammation consisted of polymorphonuclear cells, mostly seen in the pre-Descemet region and extending to the superficial layers. A few macrophages were also seen. Inflammation was absent in the corneal scar tissue. Routine staining showed Microsporidial spores as oval bodies, 2 – 3 μ wide and 3 – 5 μ in length involving mostly the deep stroma, with extension into anterior layers in 2 cases. The spores showed a thick band like nucleus at one pole and stained positively with modified Gram Trichrome stain (figure 7), 1% acid fast stain (figure 8) and variably with GMS. The unstained and faintly stained spores showed thick walled capsule that was birefringent on polarized light. The case with corneal scar and a descematocele showed a corneal scar with marked thinning in the central stroma (figure 9). There was no inflammatory infiltrate, but the stroma showed oval, spores of microsporidia in all section (figure 10). These were seen mostly in the deeper stroma, at places extending to mid and superifical stroma. The semithin sections of these tissues showed many viable, mature, immature, and degenerated spores, many of them being intracellular. The electron microscopic examination revealed an electron lucent thick capsule with a single nucleus and 11 – 13 tubules in the cytoplasm (figure 11 with inset as figure 12). Some of the sporoblasts were uniformly dark and osmophilic.
Discussion
Microsporidia are eukaryotic, spore forming, obligate intracellular, protozoan parasites with two developmental phases – schizogenic and sporogenic. The size of the spores varies from 1 to 20 μm and they are spherical, oval or elongate. They can infect a broad range of vertebrates and invertebrates and are considered ubiquitous [17]. They are becoming increasingly recognized as opportunistic infectious pathogens in immuncompromised patients, causing intestinal, ocular, sinus, pulmonary, muscular and renal diseases [1-3]. Like in all other organs, ocular microsporidial infection was rare before the era of AIDS, the first case being reported by Ashton et al in 1973 [7], in a vascularized corneal scar, unsuspected of infectious etiology. Since 1991, cases of microsporidial keratoconjunctivitis have been reported from immunocompromised patients and of late even in immunocompetent individuals [4-15], one of which was from this centre [16].
Diagnosis of microsporidiosis currently depends on morphological demonstration of the organisms in one or more readily obtainable specimens such as stool, duodenal aspirates, urine, sputum, nasal discharge, bronchoalveolar lavage fluid, conjunctival smears [18]. Definitive species identification is made by using the specific fluorescein-tagged antibody (immunofluorescence) technique, electron microscopy and also by PCR [18-21].
Ocular microsporidia could be isolated or part of systemic microsporidiosis in immunocompromised patients and can manifest as either stromal keratitis or keratoconjuncitivitis. A careful review of previously published cases and the cases reported in this series suggests that microsporidial stromal keratitis is a slowly progressive keratitis-affecting individual of any age. No definitive predisposing factors could be identified in this series. The history of trauma was elicited in only 2 of 5 cases from our series and 1 of 7 previously published reports [7].
Clinically, microsporidial stromal keratitis could mimic suppurative or non suppurative inflammation and vascularization of the cornea and includes a differential diagnosis of herpes simplex virus keratitis, fungal or bacterial keratitis. The diagnosis can be made either by corneal biopsy or deep corneal scrapings as was done in two of our cases. One of the non-invasive emerging technique of diagnosing microsporidial keratitis is with the help of a confocal microscopy [21] using a 24× contact objective lens and a Nipkow disc in imaging. The spores appear as high contrast intraepithelial or intrastromal opacities.
The duration of symptoms in this series ranged from one month to 2 years suggesting a slow indolent nature in the initial phase. Except one case, none of them gave any history of preceding trauma, illness, immunosupression or any local predisposing factors. The youngest patient in this series was 2 years, which is very unusual. Four of the cases in this series were diagnosed as infectious keratitis; two were diagnosed and treated as viral keratitis to which they showed response but after few months showed recurrence of infiltrates. The subsequent corneal scrapings in three cases were diagnosed as microsporidial keraitits. This is in contrast to the other reports where the diagnosis was made only on corneal sections.
On microbiologic examination of the corneal scraping, the differential diagnosis of microsporidia includes bacteria or yeast, but can be confirmed by special stains like Grams trichrome stain, 1% acid fast stain [19,22-24]. These oval or round structures are non-budding which differentiates them from yeast. Histologically, the organisms are seen within the stromal keratocytes, histocytes as well as between the stromal lamellae. They are seen as oval gram positive structures with a thick wall. They stain red with modified trichrome stain and acid fast stain best demonstrates the organisms. However, it should be noted that all species may not be acid fast. The inflammatory cells in the corneal stroma consisted of polymophonuclear cells as well as few histiocytes, with varying degrees of stromal necrosis, thus explains the corneal thinning in these cases.
Under electron microscope the spores of microsporidia show sporoplasm and a tubular polar filament with varying number of coils depending on the species. Therefore, ultrastructurally, it is possible to identify the species. The organisms of Nosema species, as seen in our series, measure 3 – 5 microns in length and 2 – 3 micron in width. The organisms are not surrounded by parasitoporous vacuole within the host cells and the number of coils in the cytoplasm varies between 10 – 14.
Therapeutically, there is no definitive medical therapy for this entity. Font et al treated their case with fumagillin 0.3% and oral Albendazole but there was no response. After making a diagnosis of microsporidial keratitis in our series, two received itraconazole for few weeks, while one case received cefazolin. Only one case responded to itraconazole treatment while the other two cases showed persistence of infiltrates with corneal thinning, and necessitated surgical intervention. Even in the case that responded to medical treatment there was progressive thinning that required application of tissue adhesive. Though this is a small case series, but reviewing the literature and in the light of these findings, it is possible for us to suggest that in the absence of effective medical therapy it is reasonable to manage these cases surgically. Penetrating keratoplasty helped eradicating infection in 4 cases of our series and 6 cases from previously published reports. It is important to note that the infection is mostly restricted to stroma and has not been demonstrated in deeper ocular tissues. The other favorable feature in these cases is that none of them had recurrence of infection postoperatively.
It is not clear how the microsproidia enter into the eye- it could be either due to trauma or contact with contaminated water, food. The normal lifecycle of microsporidia includes: once invasion of the spore into the human host cell occurs, the contents are discharged into the cytoplasm. Within the cell the sporoplast divides by binary fission to form schizont with 2–6 nuclei, which split into unicellular meronts. The meronts then secrete a rigid capsule and the fully formed spore measures about 2.5 × 1.5 microns. The cell eventually ruptures to continue the cycle and further destruction of the host tissue.
Conclusion
This is the largest series of ocular infection by Microsporidia in immunocompetent individuals manifesting as stromal keratitis. This series suggests the need for a high index of suspicion in diagnosing microsporidial infection in culture negative stromal keratitis. Use of special stains, tissue biopsy and electron microscopy aids in confirming the diagnosis of this infection. Due to lack of effective medical therapy, surgical treatment appears to be a reasonable option in the management of these cases.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
GKV was involved in collecting data, carrying out the analysis and drafting the manuscript. PG provided the clinical information and edited the manuscript. SS participated in its design, carried out the microbiological work up and edited the manuscript. JJ collected data, participated in its design and helped to draft the manuscript. UG participated in its design and carried out the microbiological work up. Ssingh carried out the electron microscopic analysis. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We are grateful to Dr. Govind Visvesvara, CDC, Atlanta for reviewing our slides. Financial support: Hyderabad Eye Research Foundation, Hyderabad
Figures and Tables
Figure 1 Slit lamp biomicroscopy of cornea of case 1 in diffuse illumination with central mid to deep stromal infiltrate and surrounding stromal edema.
Figure 2 Slit lamp biomicroscopy of cornea of case 3 in diffuse illumination shows a large stromal infiltrate.
Figure 3 Slit lamp biomicroscopy of cornea of case 4 in diffuse illumination shows two discreet stromal infiltrates.
Figure 4 Gram stain of corneal scraping showing gram positive oval microsporidial spores (×500, 4a).
Figure 5 1% acid fast stain of corneal scraping showing acid fast positive oval microsporidial spores (×500, 4b).
Figure 6 Section showing intact epithelium, with deep stromal infiltrates (arrow) (Hematoxylin and Eosin stain, ×20).
Figure 7 Section of corneal button which shows oval bluish purple spores of microsporidia with a dark band (modified grams chromotrope stain, ×1000).
Figure 8 Section of corneal button in which the spores appear red with darkly stained band at the tip (1% acid-fast stain, ×1000).
Figure 9 Section of corneal button of case 4 showing marked thinning in the central stroma (descematocele, arrow).
Figure 10 Section of corneal button of case 4 showing the presence of oval, spores of microsporidia in the stroma, unaccompanied by inflammatory infiltrates (PAS stain, ×500).
Figure 11 The electron microscopic examination of the corneal button revealing macrophage with three spores.
Figure 12 Inset shows an electron lucent thick capsule with a single nucleus and 11 – 13 tubules in the cytoplasm.
Table 1 Summary of clinical features in Microsporidial keratitis
Age/ Dur of Sym Med. Tx trauma vision Infiltrate Vasc Clin diag Micro Treat FU
40/F 7 mon AV, AF - 6/8 + Crys K Neg - PK
2/F 1.5 mon AB - HSV Micros Itracon PK
23/M 2 yrs AV - cfcf + HSV Micros Itracon Scar
70/F 1 mon ? - cfcf IK Micros Cefaz PK
37/F 4 mon - + Desce Neg Nil PK
AB = antibacterial
AV = antiviral
AF = antifungal
Cfc = counting fingers close to face
HSV = herpes simplex keratitis
Crys K = crystalline keratopathy
PK = Penetrating Keratoplasty
Micros = Microsporidia
==== Refs
Shadduck JA Greely E Microsporidia and human infections Clin Microbiol Rev 1989 2 158 65 2650860
Weber R Bryan RT Schwartz DA Owen RL Human microsporidian infections Clin Microbiol Rev 1994 7 426 61 7834600
Friedberg DN Ritterband DC Wittner M, Weiss LM Ocular microsporidiosis The microsporidia and Microsporiosis 1999 ASM Press, Washington DC 293 313
Friedberg DN Stenson SM Orenstein JM Tierno PM Charles NC Microsporidial keratoconjunctivitis in acquired immunodeficiency syndrome Arch Ophthalmol 1990 108 504 8 2108656
Metcalfe TW Doran RM Rowlands PL Curry A Lacey CJ Microsporidial keratoconjunctivitis in a patient with AIDS Br J Ophthalmol 1992 76 177 8 1540567
Schwartz DA Visvesvara GS Diesenhouse MC Weber R Font RL Wilson LA Corrent G Serdarevic ON Rosberger DF Keenen PC Goossniklaus HE Hewan-Lowe K Bryan RTL Pathologic features and immunofluorescent antibody demonstration of ocular microsporidiosis in seven patients with acquired immunodeficiency syndrome Am J Ophthalmol 1993 115 285 92 8095123
Ashton N Wirasinha PA Encephalitozoonosis of the cornea Br J Ophthalmol 1973 57 669 74 4202149
Pinnolis M Egbert PR Font RL Winter FC Nosematosis of the cornea. Case report including electron microscopic studies Arch Ophthalmol 1981 99 1044 47 6786272
Davis RM Font RL Keisler MS Shadduck JA Corneal microsporidiosis. A case report including ultrastructural observations Ophthalmology 1990 97 953 57 2381710
Cali A Meisler DM Lowder CY Lembach R Ayers L Takvorian PM Rutherford I Longworth DL McMahon J Bryan RT Corneal microsporidioses: characterization and identification J Protozool 1991 38 215S 217S 1818175
Font RL Samaha AN Keener MJ Chevez-Barrios P Goosey JD Corneal microsporidiosis. Report of case, including electron microscopic observations Ophthalmology 2000 107 1769 75 10964843 10.1016/S0161-6420(00)00285-2
Font RL Su GW Matoba AY Microsporidial stromal keratitis Arch Ophthalmol 2003 121 1045 7 12860813 10.1001/archopht.121.7.1045
Rauz S Tuft S Dart JK Bonshek R Luthert P Curry A Ultrastructural examination of two cases of stromal microsporidial keratitis J Med Microbiol 2004 53 775 81 15272065 10.1099/jmm.0.45524-0
Font RL Samaha AN Keener MJ Chevez-Barrios P Goosey JD Corneal microsporidiosis. Report of case including electron microscopic observations Ophthalmology 2000 107 1769 75 10964843 10.1016/S0161-6420(00)00285-2
Chan CM Theng JT Li L Tan DT Microsporidial keratoconjunctivitis in healthy individuals: a case series Ophthalmology 2003 110 1420 5 12867402 10.1016/S0161-6420(03)00448-2
Sridhar MS Sharma S Microsporidial keratoconjunctivitis in a HIV-seronegative patient treated with debridement and oral itraconazole Am J Ophthalmol 2003 136 745 6 14516821 10.1016/S0002-9394(03)00391-X
Lowder CY Wilson LA Pepose JS, Holland GN, Wilhelmus KR Microsporidiosis Ocular Infection and Immunity 1996 Mosby, St. Louise 1072 1079 Chapter 79
Lamps LW Bronner MP Vnencak-Jones CL Tham KT Mertz HR Scott MA Optimal screening and diagnosis of microsporida in tissue sections: a comparison of polarization, special stains, and molecular techniques Am J Clin Pathol 1998 109 404 10 9535393
Webber R Schwartz DA Deplazes P Wittner M, Weiss LM Laboratory diagnosis of microsporidiosis The Microsporidia and Microsporiosis 1999 ASM Press, Washington DC 315 361
Conners MS Gibler TS Van Gelder RN Diagnosis of microsporidia keratitis by polymerase chain reaction Arch Ophthalmol 2004 122 283 4 14769611 10.1001/archopht.122.2.283
Shah GK Pfister D Probst LE Ferrieri P Holland E Diagnosis of microsporidial keratitis by confocal microscopy and the chromatrope stain Am J Ophthalmol 1996 121 89 91 8554085
Moura H Da Silva JL Sodre FC Brasil P Wallmo K Wahlquist S Wallace S Croppo GP Visvesvara GS Gram-chromotrope: a new technique that enhances detection of microsporidial spores in clinical samples J Eukaryot Microbiol 1996 43 94S 95S 8822884
Moura H Schwartz DA Bornay-Llinares F Sodre FC Wallace S Visvesvara GS A new and improved "quick-hot Gram-chromotrope" technique that differentially stains microsporidian spores in clinical samples, including paraffin-embedded tissue sections Arch Pathol Lab Med 1997 121 888 93 9278620
Awadalla HN el Naga IF el-Temsahi MM Negm AY Detection of Microsporidia by different staining techniques J Egypt Soc Parasitol 1998 28 729 38 9914696
|
16105181
|
PMC1200253
|
CC BY
|
2021-01-04 16:03:50
|
no
|
BMC Ophthalmol. 2005 Aug 17; 5:19
|
utf-8
|
BMC Ophthalmol
| 2,005 |
10.1186/1471-2415-5-19
|
oa_comm
|
==== Front
PLoS GenetPLoS GenetpgenplgeplosgenPLoS Genetics1553-73901553-7404Public Library of Science San Francisco, USA 1615151510.1371/journal.pgen.001002805-PLGE-RA-0123R2plge-01-03-02Research ArticleCancer BiologyCell BiologyDevelopmentHematologyImmunologyMus (Mouse)Differential Expression of Novel Potential Regulators in Hematopoietic Stem Cells Novel Hematopoietic Stem Cell RegulatorsForsberg E. Camilla 1*Prohaska Susan S 1Katzman Sol 2Heffner Garrett C 1Stuart Josh M 2Weissman Irving L 11 Departments of Pathology and Developmental Biology, Institute of Cancer and Stem Cell Biology and Medicine, Stanford University Medical School, Stanford, California, United States of America
2 Biomolecular Engineering, University of California at Santa Cruz, Santa Cruz, California, United States of America
Roopenian Derry EditorThe Jackson Laboratory, United States of America*To whom correspondence should be addressed. E-mail: [email protected] 2005 2 9 2005 1 3 e283 6 2005 14 7 2005 Copyright: © 2005 Forsberg et al.2005This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.The hematopoietic system is an invaluable model both for understanding basic developmental biology and for developing clinically relevant cell therapies. Using highly purified cells and rigorous microarray analysis we have compared the expression pattern of three of the most primitive hematopoietic subpopulations in adult mouse bone marrow: long-term hematopoietic stem cells (HSC), short-term HSC, and multipotent progenitors. All three populations are capable of differentiating into a spectrum of mature blood cells, but differ in their self-renewal and proliferative capacity. We identified numerous novel potential regulators of HSC self-renewal and proliferation that were differentially expressed between these closely related cell populations. Many of the differentially expressed transcripts fit into pathways and protein complexes not previously identified in HSC, providing evidence for new HSC regulatory units. Extending these observations to the protein level, we demonstrate expression of several of the corresponding proteins, which provide novel surface markers for HSC. We discuss the implications of our findings for HSC biology. In particular, our data suggest that cell–cell and cell–matrix interactions are major regulators of long-term HSC, and that HSC themselves play important roles in regulating their immediate microenvironment.
Synopsis
Hematopoietic, or blood-forming, stem cells (HSC) are responsible for the continual replenishment of all blood cells throughout life. This ability to both renew themselves and give rise to expanded populations of differentiating and mature cells is a hallmark of stem cells and is therefore an area of intense research. The rarity of HSC as well as their location in the bone marrow environment has made it difficult to identify the genes that regulate these properties. The earliest stages of blood development begins with the long-term (LT) repopulating HSC that then differentiate into short-term (ST) repopulating HSC and non-self renewing multipotent progenitors (MPP). The authors investigated the gene expression differences in these highly purified populations that differ mainly in their capacity to self renew, and identified a number of genes specific to each of these populations. Intriguingly, many of these genes code for proteins that are involved in cell–cell and cell–matrix interactions that were not previously identified on these populations. These novel discoveries will, together with future experiments, enhance our understanding of the basic biology of stem cells and their clinical uses.
Citation:Forsberg EC, Prohaska SS, Katzman S, Heffner GC, Stuart JM, et al. (2005) Differential expression of novel potential regulators in hematopoietic stem cells. PLoS Genet 1(3): e28.
==== Body
Introduction
Mature blood cells have a high turnover rate and need to be constantly replaced as well as respond to more acute conditions such as blood loss or infections, requiring the rapid generation of millions of new blood cells. This demand is fulfilled for life by a pool of hematopoietic stem cells (HSC). The long-term repopulating HSC (LT-HSC) thus has to be capable of differentiating without depleting the stem cell pool, thereby satisfying the definition of a stem cell: the ability at the single cell level to both self-renew and differentiate into more mature cell types. LT-HSC normally reside in the bone marrow and have essentially six developmental choices: remain quiescent, differentiate, self-renew, migrate, enter senescence, or undergo apoptosis. Such fate decisions are likely controlled both by HSC-intrinsic mechanisms and by the bone marrow microenvironment or “niche.” It has proved difficult to define the complex intrinsic and extrinsic mechanisms that govern the balance of these decisions. In hematopoiesis, only LT-HSC are capable of lifelong self-renewal and, therefore, is the operative population in hematopoietic transplantation. Understanding how HSC fate decisions are controlled is therefore of critical importance.
The expression profiles of primitive hematopoietic cells defined by various criteria have previously been compared to other hematopoietic and non-hematopoietic cell types [1–5]. In addition, several molecules and pathways have been implicated in HSC self-renewal, including HoxB4, Bmi1, the Wnt/β-catenin signaling pathway, and Notch [6–9]. As the bone marrow microenvironment likely provides unique cues necessary for proper HSC function, cell surface proteins mediating these signals should play important roles in HSC fate decisions. While there have been recent advances in defining a potential HSC niche within the bone marrow [10,11], little is known about the specific signals regulating HSC in vivo. A central question is how the interplay of soluble ligands, matrix interactions, and cell–cell contacts influence HSC fate. Recent evidence points to a role for the angiopoietin receptor Tek, also known as Tie2, in maintaining transplantable HSC [12]. Likewise, HSC in mice null for Mmp9, a matrix metalloproteinase (Mmp) that facilitates cell migration by proteolytic cleavage, have impaired proliferation and differentiation capabilities [13]. In addition, mice lacking integrin β1, part of the HSC homing receptor α4β1 [14], cannot establish fetal liver hematopoiesis [15], presumably due to a failure of LT-HSC to engraft at this fetal site. Thus, there is increasing evidence that interactions with the environment are important for the maintenance of HSC self-renewal capability. A thorough understanding of cell surface molecules expressed on HSC is an important step in identifying functional interactions with the environment.
Here, we have carefully analyzed the transcription profiles of three highly purified subpopulations within the mouse adult bone marrow lineage−/c-kit+/Sca1+ (KLS) fraction: LT-HSC (defined as Thy1.1lo/Flk2− KLS), short-term (ST)-HSC (Thy1.1lo/Flk2+ KLS), and multipotent progenitors (MPP) (Thy1.1−/Flk2+ KLS) [16]. These three populations have the ability to give rise to both lymphoid and myeloid lineages [16] and platelets (E. C. F., E. Passegué, and I. L. W., unpublished data) when transplanted into irradiated mice. Thus, LT-HSC, ST-HSC and MPP have similar multilineage potential, but differ in their self-renewal and proliferative capacity. All long-term repopulating activity is contained in the LT-HSC fraction; thus, cells within this fraction are the only cells capable of maintaining hematopoiesis for the life of the host. As LT-HSC differentiate to ST-HSC and then to MPP, self-renewal capability progressively declines [16]. Therefore, cells derived from transplanted ST-HSC and MPP decrease to undetectable levels in peripheral blood by about ten weeks posttransplant. LT-HSC, ST-HSC and MPP also differ in their cell cycle status. While most LT-HSC are quiescent, a larger fraction of ST-HSC and MPP are in cycle under steady-state conditions (E. Passegué, A. J. Wagers, and I. L. W., unpublished data). Based on these small, but functionally critical differences, we reasoned that comparing the transcriptional profile of these three populations would reveal genes specifically involved in self-renewal, quiescence, and proliferation mechanisms. Here, we present new data on selective gene expression in highly purified HSC subsets and discuss the possible implications for HSC biology.
Results
Experimental Design and Data Analysis
To obtain accurate and reliable transcription profiles of LT-HSC, ST-HSC and MPP, we isolated highly pure, functionally defined cell populations and performed pairwise competitive hybridizations in three independent experiments, each with a “dye swap,” so that each comparison was performed six times. This allowed us to identify small but reproducible changes in gene expression. We used Stanford microarrays spotted with about 42,000 cDNAs and expressed sequence tags (ESTs) from a wide variety of tissue- and developmental stage-specific libraries, including a library made from bone marrow-subtracted HSC cDNA [4] and a selection of genes involved in development and hematopoiesis. We employed stringent criteria when evaluating and scoring genes as differentially expressed. This resulted in six gene lists (two for each comparison; Figure 1 and Tables S1−S6). As expected, most of the differentially regulated transcripts came from the comparison of LT-HSC to MPP, with 527 transcripts upregulated in LT-HSC and 323 transcripts upregulated in MPP, totaling 850 differentially regulated transcripts. Between LT-HSC and ST-HSC 565 transcripts were differentially regulated, as were 111 transcripts between ST-HSC and MPP (Table 1). The top 25 statistically significant up- and downregulated genes from each comparison are listed in Figure 1. Complete lists of all differentially regulated genes, including ESTs and transcripts represented by multiple spots, are presented in Tables S1−S6 and summarized as a heat map representation in Figure 2.
Figure 1 Top 25 Differentially Regulated Transcripts with Corresponding SAM Plots for Each Comparison
Only known, unique genes are listed; thus, ESTs were removed, and genes appearing more than once in the same list are denoted with number of appearances in parentheses. An “A” in parenthesis indicates that results from analogous experiments using Agilent arrays are consistent with the Stanford Microarray Database array data. The LT-HSC to ST-HSC comparison was not performed with Agilent arrays.
Table 1 Summary of the Numbers of Differentially Regulated Transcripts
Figure 2 Heat Map Representation of Differentially Regulated Transcripts
Rows represent genes and columns represent array comparisons between cell populations as indicated.
(A) Significance score by SAM is indicated from top to bottom of each comparison (n = 6) by gradients of yellow or blue, with brighter yellow or blue indicating higher significance. Yellow or blue in additional columns indicate that the gene was also differentially expressed between these cell types (e.g., many of the genes upregulated in MPP versus LT-HSC were also upregulated in ST-HSC compared to LT-HSC).
(B) Conventional red-green expression data for all the differentially regulated genes for each array. Red indicates genes upregulated in the more differentiated cell population (e.g., upregulated in MPP when compared to LT-HSC); green indicates upregulated in the less differentiated cell population.
The reliability of the resulting lists of differentially expressed genes was then assessed in several ways. First, as illustrated in Figure 2, the reproducibility between and across arrays is extremely good and consistent with the previously described linear relationship between these populations, where LT-HSC give rise to, first, ST-HSC, which in turn differentiate into MPP. Second, these data accurately reflect the expression of several transcripts known to be differentially expressed in hematopoietic stem and progenitor cells. These include the sort markers Thy1.1 (upregulated in LT-HSC) and Flk2 (upregulated in MPP). The transporter Abcg2, responsible for the “side population” phenotype [17], is upregulated in LT-HSC, while CD34 transcripts increase with differentiation, consistent with CD34 protein levels in previous studies [18]. A fraction of LT-HSC is negative for CD34 cell surface protein, while the vast majority of ST-HSC and MPP are CD34-positive (Figure 3). Aldehyde dehydrogenase, Aldh1a1, and Tek transcript levels are also higher in LT-HSC in agreement with their selective expression by HSC reported previously [12,19,20]. In addition, numerous transcripts scored consistent with each other on the Stanford cDNA arrays if represented by two or more spots, and when assayed using Agilent oligonucleotide arrays (see Figure 1; unpublished data). We also confirmed the expression patterns of several of the transcripts by quantitative RT-PCR of unamplified RNA from double-sorted cells (Figure S1) and by flow cytometry of protein levels (Figure 3). In no case did RT-PCR or cell surface protein levels contradict the array data. Thus, we believe that these analyses accurately reflect the true transcription profile of these three populations.
Figure 3 Flow Cytometric Analysis of Cell Surface Protein Levels on LT-HSC, ST-HSC, and MPP Relative to Control
Green represents LT-HSC, blue represents ST-HSC, red represents MPP, and grey represents the control.
To detect potential self-renewal or proliferation loci we determined the chromosomal location of genes differentially regulated between LT-HSC and MPP (Figure S2). The most striking difference was observed on the X chromosome, where 23 genes upregulated in LT-HSC but only one MPP-specific gene reside. The highest enrichment of MPP-specific genes was a 2.4-fold increase on chromosome 19. The 23 LT-specific genes are listed in Table S7. We detected no enrichment of LT-HSC-specific genes on chromosome 17, which we and others previously showed to contain a stem cell frequency locus [2,21].
To assess objectively what types of genes were differentially regulated, we used the gene classification tool Gene Ontology (GO) [22]. GO has classified a large number of genes based on biological processes, molecular function, and cellular component. Genes enriched in the more primitive cells are overrepresented in the GO categories involving cell junction, anion transport, extracellular space, and oxidoreductase activity, whereas cell differentiation enriched for transcripts in the cell cycle, immune response, taxis/chemotaxis, and regulation of transcription categories. However, as only a fraction of the genes in our data set has been classified by GO, the majority were broadly categorized by reviewing published literature and discussed based on cellular location moving from the extracellular space inward. Genes discussed in the text are summarized in Table 2 and selected transcripts are depicted in Figure 4. Additional references are presented in Table S8.
Table 2 Differentially Regulated Transcripts Discussed in the Text
Figure 4 Hypothetical Model of Selected Potential Interactions of Proteins Corresponding to Differentially Expressed Transcripts
Molecules upregulated in LT-HSC are in green; those upregulated in MPP are red. Depicted protein interactions and colocalization are based on published reports in various mammalian and nonmammalian systems. Single cells expressing all the proteins as depicted in this cartoon may not exist. Not drawn to scale.
Extracellular Matrix and Other Secreted Proteins
The extracellular matrix (ECM) proteins upregulated in LT-HSC include transglutaminase2 (Tgm2), biglycan, Dag1/dystroglycan, ApoE, Emilin1, Calsyntenin1, collagens type 4a1 and 4a2, and the heparin-binding protein Fstl1. Tgm2 is a widely expressed protein cross-linking enzyme that can function in the ECM, cytosol, and nucleus. The proteoglycan biglycan is highly expressed in connective tissues including bone, and mice lacking biglycan are prone to osteoporosis. ApoE, an ECM protein and a component of low-density lipoproteins, interacts directly with biglycan, and both ApoE and biglycan bind to membrane proteins preferentially expressed by LT-HSC (see below). By expressing genes encoding structural components of the ECM such as collagens and proteoglycans, LT-HSC may contribute to the architecture of its immediate microenvironment. In contrast, matrix metalloproteinases (Mmp) facilitate cell motility and are important for HSC differentiation [13]. Mmp15 and Mmp16 are upregulated in LT-HSC, but their activity may be inhibited by high levels of the Mmp inhibitors Timp2 and Timp3. Interestingly, concurrent with Timp2 downregulation in MPP, the main target of Timp2, Mmp2, is upregulated. Another facilitator of cell motility, heparanase, is also upregulated in MPP. Heparan sulfate-bearing membrane receptors, potential substrates for heparanase, are upregulated in LT-HSC (below). Together, the increases in heparanase and Mmp2 expression and decreases in Timp2 and Timp3 levels may act to destabilize cell interactions with the matrix and allow a subset of HSC to exit the niche and undergo differentiation.
Membrane, Cell Junction, and Adhesion Molecules
The “cell junction” GO category includes a number of genes not previously implicated in HSC biology. Transcripts upregulated in LT-HSC in this category include the junction adhesion molecules Jam1/F11R, Jam2 and Jam3, Claudin12, Claudin22, the tight junction protein Tjp1/ZO-1, and the connexins Gja1 and Gjb5. Jam2 protein is detectable on the surface of LT-HSC, and its levels decrease as LT-HSC differentiate into ST-HSC and MPP (see Figure 3). Jams, Tjps, and claudins commonly colocalize in epithelial cells to form tight junctions. Tjp1 is also found at gap junctions with the channel-forming connexins. The differential expression of two connexins in LT-HSC, Gja1 and Gjb5, is consistent with evidence for gap junctions functioning in stem cell regulation in human ES cells [23], and in Drosophila, where the gap junction protein Zpg is necessary for proper germ stem cell differentiation [24]. It will be interesting to determine whether these proteins form functional channels in HSC, and if so, with what partner cells and what types of molecules are exchanged between cells.
VCAM1 and ESAM1 are related adhesion molecules upregulated in LT-HSC both at the transcript and protein level (see Figure 3). VCAM1 interaction with integrin α4β1 mediates cell–cell interactions in multiple cell types, and both VCAM1 and integrin α4β1 have been implicated in HSC homing to the bone marrow. As at least a subset of LT-HSC express α4β1 [14], detection of VCAM1 protein on all LT-HSC (Figure 3) indicates that a single LT-HSC can express both α4β1 and VCAM1. ESAM1 was first described on endothelial cells, but is also expressed on megakaryocytes and platelets [25,26]. ESAM1 mediates homophilic interactions between endothelial cells and colocalizes with tight junction proteins and claudins. The self-ligand Slamf1, known for its roles in bidirectional T- and B-cell stimulation, is upregulated in LT-HSC, as shown here and in other studies (S. Morrison, personal communication). Interestingly, the cell surface protein profile divides LT-HSC into Slamf1 positive and negative fractions, whereas ST-HSC and MPP are uniformly Slamf1 negative (Figure 3). As HSC negative for CD34 are Slamf1 positive, and there is an inverse relationship between Slamf1 and Flk2 surface expression on both HSC and myeloid progenitor populations, Slamf1 functions as a useful marker to separate these populations (D. Bryder, E. C. F., and I. L. W., unpublished data). Ongoing studies indicate that there are functional differences between Slamf1 positive and negative stem and progenitor populations.
LT-HSC also express higher levels of the receptor Robo4, its putative ligand Slit-like2, and the effector protein Srgap2 than do ST-HSC or MPP. In Drosophila, Slit/Roundabout signaling promotes asymmetric division, and it inhibits cell migration in both Drosophila and mammalian systems [27,28], perhaps by inhibition of Mmp2 and Mmp9 by Slit [29]. Robo4 can interact directly with DCC, a homolog of Neogenin which is also upregulated in LT-HSC over MPP. In other systems, Neogenin and DCC are implicated in cell adhesion, polarity, and migration, and are receptors for the Netrin family of chemoattractants. Interestingly, Slit and Netrin signaling influence each other [30] and can have opposing effects [31]. Slit signaling can also modulate Cxcr4-mediated chemotaxis [29], and may therefore play a role in HSC migration toward Sdf1 gradients [32,33].
Drosophila Roundabout binds directly to Syndecan [27], and Roundabout mutants phenocopy the Syndecan mutant [34]. The syndecans are transmembrane heparan sulfate-decorated molecules that link the cytoskeleton to the extracellular space and participate in cell–cell and cell–matrix interactions, cell proliferation, and growth factor signaling. We found Sdc1 transcripts upregulated in MPP, and a fraction of MPP also expresses the protein on the cell surface (Figure 3). In contrast, Sdc2 transcripts are enriched in LT-HSC. Both Sdc1 and Sdc2 are coreceptors for FGF and GM-CSF [35,36], molecules with known functions in hematopoiesis. Syndecan-mediated adhesion is regulated by metalloproteinases and their inhibitors as well as by heparanase. Cleavage of the heparin chains by heparanase or proteolysis by metalloproteinases increases cell migration, the latter of which can be inhibited by Mmp inhibitors such as Timp3 [27]. Thus, differential expression of syndecans and regulation of their cell–cell and cell–matrix interactions may control retention versus release of cells from the niche.
Amyloid beta precursor protein (App) is a heparin-binding cell adhesion molecule that interacts with two of the extracellular matrix molecules upregulated in LT-HSC, ApoE [37] and biglycan [38]. Mutations of App have been implicated in Alzheimer's disease. Proteolytic cleavage of App by γ-secretase and presenilin leads to nuclear translocation and transcriptional activation. App may thus directly transmit extracellular signals to regulate transcription of specific target genes in LT-HSC.
Tek is expressed by HSC [12,19], and a role for Tek in maintaining HSC quiescence through interaction with the niche was suggested [12]. Tek transcript levels are indeed higher in LT-HSC than in ST-HSC or MPP, and all detectable cell surface expression of Tek protein is contained within the LT-HSC fraction (Figure 3). We and others [1] also detected robust and differential transcript levels of the Tek ligand angiopoietin1 in LT-HSC, suggesting potential autocrine or paracrine Tek activation.
Several other proteins corresponding to transcripts enriched in LT-HSC might mediate cell–cell and cell–matrix interactions. Map17 is expressed mainly in the kidney but is also enriched in LT-HSC over MPP. Although upregulated in carcinomas, overexpression of Map17 in colon carcinoma cells can inhibit cell proliferation and tumor growth [39]. The tetraspanins CD9 and CD63 can function as metastasis suppressors, potentially due to association with integrins [40]. The surface protein levels of CD9 correlate with the transcript profile (Figure 3). Elsewhere we have shown that CD9 is a marker for the megakaryocyte progenitor, MKP [41]. CD9 can interact with integrin α6 (Itga6) [42–44], which, together with integrin β1, form a cell surface laminin-binding receptor. Cell surface expression of Itga6 and Itgb1 on HSC was demonstrated previously [14]. Itga6 may be a general stem cell marker since it is upregulated on LT-HSC in addition to stem cells of several other tissues [45,46]. Another surface protein upregulated in LT-HSC, Mllt4, is an actin filament-binding protein that belongs to a cadherin/catenin adhesion system with roles in the organization of homotypic and heterotypic cell–cell adherens junctions [47].
In contrast to the abundance of cell adhesion molecules upregulated in LT-HSC, MPP express a number of cell surface receptors implicated in immune response and chemokine activation. These include IL10rα and β, IL1r1, IL17r, Mpeg1, Blnk, Notch1, Ccr5, and Igh6. These and other receptors selectively expressed by MPP can render cells responsive to chemotactic and inflammatory response molecules such as Ccl3, Ccl4, Ccl9, and IL1β, whose transcripts are also upregulated in MPP. It is possible that distinct subsets of MPP express high levels of particular transcripts as an early indicator of commitment to different downstream lymphoid and myeloid lineages, or that the presence of both types of transcripts are indicative of promiscuous expression associated with lineage commitment [48,49].
Intracellular Adaptor and Signaling Molecules
Links between the extracellular space and the actin cytoskeleton are provided by several transcripts preferentially expressed by LT-HSC. This includes direct interaction with actins by transmembrane proteins such as Mllt4 and sarcoglycan E, as well as by use of adaptor molecules such as the erythrocyte cell shaping protein Epb4.1, the Slit/Robo effector protein Srgap2, cortactin, and Fgd1. Cortactin functions in cell shape, actin organization, and cell adhesion by interacting directly with actin and with components of adherens-type junctions, including cadherins, catenins, and Tjp1 [50]. Cortactin also interacts with Fgd1 [51], a protein with roles in actin organization and cell shape.
The growth factor receptor-bound adaptor proteins Grb2 and Grb10 are also upregulated in LT-HSC. In other systems, Grb10 regulates signaling from several receptors, including IGFR, PDGFR, EGFR, and VEGFR2, at least in part by antagonizing receptor degradation by Nedd4 [52], a ubiquitin ligase robustly expressed in HSC. Depending on the context, Grb10 may promote or inhibit cell growth [53]. Grb2 can interact with App to regulate apoptosis in a Map kinase-dependent manner in neuroblastoma cells [54]. Interestingly, Grb2 also binds catalase [55], an enzyme recently implicated in HSC self-renewal [56]. Catalase is upregulated in LT-HSC together with other transcripts implicated in oxidoreductase activity such as Sestrin1, the known stem cell marker aldehyde dehydrogenase Aldh1a1 and its homolog Aldh7a1, the cytochrome P450 Cyp2j6, and glutathione-S-transferases Gsta1, Gstk1, and Gstm1. These enzymes may render LT-HSC uniquely resistant to stress, a property that would be particularly important for long-lived cells such as stem cells.
MPP have higher transcript levels for proteins with potentially opposite effects of those upregulated in LT-HSC. Two such examples that regulate cell growth, proliferation, and differentiation are Gelsolin, a calcium-dependent actin-depolymerizing protein, and Lims1, an adaptor protein mediating integrin and growth factor signaling at focal adhesions [57,58].
Proliferation and Cell Cycle Genes
The proliferative activity of LT-HSC increases as they differentiate into ST-HSC and MPP (E. Passegué, A. J. Wagers, and I. L. W., unpublished data). Thus, we expected transcripts related to cell cycle control to be differentially regulated. Indeed, the most significantly enriched GO categories for MPP are related to cell proliferation. The array data presented here are corroborated by the qRT-PCR data performed on the same cell populations (E. Passegué, A. J. Wagers, and I. L. W., unpublished data) and indicate that LT-HSC express higher levels of cyclin D3 and the cyclin-dependent kinase inhibitor p57/Cdkn1c, whereas ST-HSC/MPP express higher levels of cyclins A2, B1, B2, and D1 and the proliferation marker Ki67. Our array analysis identified several additional transcripts that may play a role in regulating HSC cell cycle status. Transcripts enriched in LT-HSC include p53, the growth suppressor necdin; Sesn1, a regulator of cell cycle arrest and a negative regulator of cell proliferation; and Ches1, a probable transcriptional activator that may be involved in DNA damage-induced cell cycle arrests.
Ier5, a mediator of mitogenic signals, is upregulated in ST-HSC and MPP, as are several transcripts associated with mitotic chromosomes. These include Spag5, the centromere proteins Cenph and Cenpe, and the kinesins Kif4, Kifc5a, Kif22, and Kif23. Marcks-like protein (Mlp), a component of the anaphase-promoting complex, and the topoisomerase Top2a are also increased. Two cohesin-complex transcripts, Smc2l1 and Smc4l1, are upregulated as LT-HSC differentiate, potentially to promote chromosome segregation during the increased cycling of ST-HSC and MPP. In contrast, Smc1l1 is upregulated in LT-HSC. Smc1l1 is phosphorylated by ataxia telangiectasia mutated, indicating a potential role for this protein in DNA repair. This is also of note, since ataxia telangiectasia mutated-mediated responses to oxidative stress recently were implicated in HSC self-renewal [56]. Although several of the transcripts in this category may be upregulated as a consequence of increased cycling, some likely have more causal roles. Proliferation is likely regulated to a significant extent by high expression of cell cycle inhibitors such as p57/Cdkn1c in LT-HSC. However, of note is that cyclin D3 is upregulated in LT-HSC, and it will be interesting to determine whether cyclin D3 activity favors self-renewing divisions over divisions associated with differentiation.
Transcription Factors and Other Nuclear Proteins
Transcriptional regulators with higher expression in LT-HSC include the erythroid- and megakaryocyte-specific transcription factors Gata1, Gata2, and Nfe2; members of the Hox family of transcription factors, Hoxa5 and Hoxb5; the Hox-interacting Pbx1 and Pbx3; the high-mobility group proteins Sox6, Sox18, and Hmgb3; and Ski, Dachshund1, and Eya2. Ski and Dachshund belong to the Ski/Sno family of proto-oncogenes implicated in TGF-β/Smad signaling. They generally act as corepressors of transcription in concert with DNA-binding proteins and histone deacetylases. Ski has previously been shown to be expressed in hematopoietic cells [59] and to play a role in regulating cell proliferation by interfering with TGF-β signaling [60] or Myb-mediated transcriptional activation [61]. Indeed, as transcript levels of Ski decline in MPP, there is increased expression of the Myb target gene c-myc. This transcriptional repressor or activator depending on the context, might affect the balance of HSC self-renewal, as mice lacking hematopoietic expression of c-myc have increased numbers of phenotypic HSC and fewer differentiated progeny [62].
Dachshund proteins operate together with the Eya family of proteins to regulate gene expression. While Dach can repress transcription either by binding directly to DNA or as a corepressor for the Six family of proteins, Eya factors are transcriptional coactivators [63,64]. Both Eya1 and Eya2 transcripts are enriched in HSC when compared to hematopoietic progenitor populations [4], but only Eya2 is differentially expressed among LT-HSC, ST-HSC and MPP. The Eya proteins interact with G proteins in the cytosol, translocate to the nucleus, and use their phosphatase activity to regulate transcription [65]. The Eya/Dach complex has been implicated in modulating precursor cell proliferation [65] by regulating transcription of cell cycle inhibitors such as p27/Cdkn1b [63].
The high-mobility group proteins Hmgb3, Sox6, and Sox18 are upregulated in LT-HSC compared to MPP. Sox18 has been implicated in vascular development and transactivates the Vcam1 promoter [66], consistent with the higher levels of Vcam1 in LT-HSC (Figure 3). Sox6 has mainly been implicated in cartilage formation as a downstream mediator of Bmp signaling [67], but it can also regulate transcription of the cell adhesion molecules N- and E-cadherin and of Wnt1 [68]. By affecting these regulators of HSC function, Sox6 could play important roles in HSC biology. However, Sox6 null mice have no obvious hematopoietic phenotype, suggesting the presence of redundant factors or mechanisms. Hmgb3 has previously been shown to be expressed by HSC, and its downregulation was important for differentiation into myeloid and lymphoid but not erythroid lineages [69]. The lower levels of Hmgb3, Gata1, Gata2, and Nfe2 transcripts may contribute to attenuated erythroid and megakaryocyte potential of MPP compared to LT-HSC.
Nuclear factors upregulated in MPP include Ddx4, Hoxa9, Satb1, Atrx, Notch1, and Tcf12. Tcf12 may play a role in the proliferation of neural stem and progenitor cells [70], and its upregulation in MPP could indicate a similar role in hematopoiesis. Hoxa9 and several other Hox genes have important roles in hematopoiesis, and several Hox members are involved in chromosomal translocations in leukemogenesis. Hoxa9 null mice have significant reductions of both myeloid and lymphoid progenitor populations [71], and overexpression of Hoxa9 can expand the HSC pool [72], indicating that Hoxa9 is a regulator of hematopoietic stem and progenitor cell pool size. The expression of Ddx4, a homolog of Drosophila Vasa, in MPP is surprising as it has been shown previously to be selectively expressed in the germline [73]. Ddx4 has RNA helicase activity and is involved in RNA-related processes such as translation initiation, splicing, and nucleosome assembly. Upregulation of Ddx4 in MPP may reflect increased or selective translational activity compared to LT-HSC.
Although Notch1 has been implicated in HSC self-renewal [9], transcript levels of Notch1 increase as LT-HSC differentiate to MPP. Notch1, as well as the chromatin-remodeling protein Satb1, have roles in lymphoid development and cell fate decisions. Mutations of another chromatin remodeling factor upregulated in MPP, Atrx, are associated with mental retardation and α-thalassemia. Atrx may play roles in spindle organization and chromosome alignment during cell division [74], in addition to modifying chromatin structure to regulate transcription. As several transcriptional activators and repressors have been implicated both in HSC self-renewal and as master regulators of hematopoiesis, the role of transcriptional regulators not previously known to be expressed by HSC, such as Dachshund, will be interesting to examine, and genes regulated by a potential Dach/Ski/Eya2 complex may play important roles in LT-HSC fate decisions.
Discussion
Here, we have analyzed in detail the difference in transcriptional profiles between LT-HSC and two closely related isolatable progeny during steady-state hematopoiesis. This analysis revealed numerous transcripts not previously known to be expressed by HSC. Directly comparing highly purified LT-HSC to closely related progeny as opposed to heterogeneous cell populations or cells of nonhematopoietic lineages likely enhanced the sensitivity of the analysis. It is important to note that our analyses focused on differential expression and therefore did not seek to identify genes that are important for HSC biology but are not differentially expressed at the transcript level among LT-HSC, ST-HSC, and MPP. Posttranscriptional regulation and interaction with differentially expressed molecules may regulate the activity of such proteins. The differentially regulated transcripts we identified likely play important roles in early hematopoietic decisions, including self-renewal and quiescence of LT-HSC, and differentiation into the more proliferative ST-HSC and MPP. Collectively, these data support previous evidence suggesting that processes such as quiescence, adhesion, and cytoprotection are particularly important for LT-HSC integrity, while differentiation, proliferation, and chemotaxis are increasingly important for the immediate progeny (Figure 5). Novel candidates regulating these processes provide a model that will guide functional studies.
Figure 5 Schematic of Biological Processes that Gradually Decline or Increase with LT-HSC Differentiation Based on the Relative Transcript Levels Presented in this Report
Green circles represent LT-HSC, and color gradients from green to red represent increasingly mature progeny.
As shown in Figure 3, our analysis also led to the identification of new differentially expressed cell surface markers. It is important to note that there are multiple ways to isolate HSC and multipotent progenitors [1,16,18,75–79]. Although populations isolated by alternative methods show significant overlap with the subsets described here, there is likely even more complexity beyond the ones reported here. For example, functionally distinct multipotent progenitors can be separated based on CD4 and Mac1 expression [76], and as these subsets express low levels of Thy1.1, they are likely more primitive than the Thy1.1-negative MPP described here. It is possible that the MPP we identified by Flk2 expression are determined to respond to inflammatory stimuli. In addition, since LT-HSC as defined here can be divided into two fractions by at least two different surface markers (Slamf1 and CD34), we anticipate that these and other novel antigens will be useful in the ongoing refinement of the relationship between hematopoietic cell types and will increase our understanding of cell fate decisions.
The relative quiescence of LT-HSC and stem cells of nonhematopoietic lineages likely protects these cells from exposure to reactive oxygen species and toxic metabolites that could lead to DNA damage. Avoiding inheritable damage would be particularly important for stem cells, since they give rise to billions of mature cells throughout the life of the host. Insults to the hematopoietic system could be minimized if LT-HSC were resistant to cellular damage and then capable of replacing the shorter-lived mature cells. In addition to quiescence, the upregulation of transcripts associated with cytoprotection such as p53, Ches1, Sesn1, catalase, and Abcg2 in LT-HSC may play important roles in maintaining the integrity of LT-HSC.
Importantly, Abcg2 and other ATP-binding cassette transporters are associated with multidrug resistance of cancer cells, and an important goal in cancer biology is to understand the differences and similarities between normal stem cells and cancer stem cells [80]. We recently identified the leukemia-initiating cell fraction in a mouse model resembling human chronic myelogenous leukemia (CML), where the transplantable leukemia was contained in the phenotypic LT-HSC compartment [81]. To understand the molecular events leading to CML, we compared the expression profile of these leukemic HSC to normal HSC. Many of the transcripts preferentially expressed by normal LT-HSC are downregulated in the leukemic HSC (E. Passegué, E. C. F., and I. L. W., unpublished data), emphasizing the importance of these genes in the regulation of normal stem cells. We have also noted that in human CML the leukemic stem cell is at the HSC stage during the early, more indolent form of the disease, but when myeloid blast crisis emerges the candidate leukemic stem cell is found in a population of normally non-self-renewing cells [82]. In addition to clarifying the events leading to neoplastic transformation, understanding the similarities and differences between normal and oncogenic self-renewing cells will help specific therapeutic targeting of cancer-initiating cells while sparing normal HSC.
It is intriguing that many of the differentially regulated transcripts represent components of functional units such as receptor-ligand pairs and protein complexes as exemplified by the cell junction category (see Figure 4). Although some of these molecules can be found at adherens junctions, several are specifically associated with tight and gap junctions. Roles for these types of junctions have not been described in HSC; thus, determining their function will provide novel insights to HSC biology. We are currently assessing the functional role of Esam1 in hematopoiesis and the use of this antigen as a highly selective marker for LT-HSC. The details of those studies will be presented elsewhere.
The bone marrow microenvironment is crucial for adult HSC regulation. However, in view of the number of differentially expressed matrix molecules (i.e., Biglycan, ApoE, collagens), homotypic-interacting proteins (i.e., Esam1, Mllt4, Slamf1, claudins, and connexins), and ligand-receptor pairs (Angpt1:Tek, Slitl2:Robo4, Ccl3:Ccr5, and IL1r1:IL1b) these cell populations seem surprisingly self-sufficient in controlling their fates. The importance of these types of interactions for HSC function has been demonstrated previously, for example between integrin α4β1 and Vcam1, and between c-kit and SLF [14,83]. It should be noted, however, that the simultaneous expression of interacting molecules by the same cell presents cell biological puzzles concerning the prevention or use of intermolecular interactions and potential signaling during processing through the endoplasmic reticulum, the Golgi, and post-Golgi vesicles. Although other factors and cell types likely play important roles, LT-HSC might be capable of modifying their immediate environment by secreting matrix molecules as well as ligands for surface receptors. The abundance of homotypic-interacting surface molecules on LT-HSC suggests that LT-HSC interact with cells expressing at least some of the same molecules. Thus, it seems possible that HSC interact with each other and that they may reside in localized clusters in the bone marrow. The difficulty of identifying functional HSC in their native environment makes it even more challenging to determine what cell types interact with HSC in situ. Our identification of these novel HSC surface proteins could support current evidence for interactions between HSC and osteoblasts [10,11,62,84] and facilitate the search for other cell types capable of interacting with LT-HSC. In this regard, the expression of some of the same molecules upregulated by LT-HSC (biglycan, Gjb5, Slitl2) by AFT024 [85] may account for this stromal line's capability of supporting LT-HSC in culture for several weeks [86].
Although interaction with the niche may be a prerequisite for both HSC quiescence and self-renewal, these two processes might be developmental choices since a truly quiescent cell is not dividing. Thus, self-renewal requires signals to proliferate, in combination with prevention of apoptosis and differentiation. As many cell types are capable of self-renewal, although for a limited time or number of divisions, extended self-renewal may be regulated to a greater extent by inhibition of differentiation than by active promotion of self-renewal. The two main mechanisms regulating LT-HSC properties could be location and unresponsiveness to mitogenic and differentiation signals. Proper localization of LT-HSC within the bone marrow would be mediated by adhesion molecules preferentially expressed by LT-HSC. This niche might be characterized by high levels of molecules promoting quiescence and preventing differentiation, and by relatively low levels of signals favoring proliferation and differentiation. LT-HSC may contribute to the establishment of such gradients by expressing ligands, protease inhibitors, and extracellular matrix molecules, all of which would help retain the LT-HSC in the niche, hiding from toxic metabolites and mitogenic and differentiation signals (see Figure 5). We propose that at the appropriate time, HSC are induced to express relatively high levels of cyclin D3, which may enable them to overcome the antiproliferative signals and allow LT-HSC to undergo self-renewing divisions in the niche. More extensive proliferation and differentiation may require cells to move out of the niche. This might be facilitated by enzymes that inhibit cell adhesion, such as heparanase and metalloproteinases. Because MPP express higher levels of these enzymes, as well as cytokines associated with more differentiated cell types, some of this regulation could be cell autonomous.
The second proposed major regulator of LT-HSC, relative unresponsiveness to proliferation and differentiation signals, could be due to lack of receptors and signaling molecules mediating such signals and by high levels of cell cycle inhibitors. In contrast, ST-HSC and MPP may be poised to move down a differentiation pathway, in part by their inability to adhere to the niche, and in part by being more responsive to differentiation and mitogenic signals (Figure 5). Differentially expressed genes will cause two different cells exposed to the same signals to respond differently. For example, HSC with long-term repopulating ability circulate in the blood under normal homeostatic conditions [33], upon transplantation, and when induced to mobilize. Thus, leaving the niche does not induce irreversible differentiation in all cases. It is possible that passage through the blood and reentry into the bone marrow allows for stochastic relocation of cells to different kinds of available niches, some that direct the cells to regain quiescence, others to self-renew, or still others to differentiate. Such a model also provides a means for LT-HSC to sense and respond to homeostatic changes in the hematopoietic system more easily than if they remained in the niche. Quiescence itself may contribute to the unresponsiveness of LT-HSC, since differentiation requires proliferation. In addition, adhesion reinforces quiescence and vice versa, creating a positive feedback loop.
As the definition of HSC currently depends mainly on transplantation into irradiated hosts followed by assessment of long-term multilineage readout, it is difficult to determine to what extent an endogenous cell with a ST-HSC or MPP cell surface phenotype self-renew normally as opposed to upon transplantation. In addition, long-term engraftment upon transplantation is dependent on cell cycle status, with quiescent cells having higher engraftment potential (E. Passegué, A. J. Wagers, and I. L. W., unpublished data) [87,88]. The current transplantation-based definition of LT-HSC is therefore directly tied to quiescence. When assayed experimentally, long-term engraftment potential may thus be more a measure of a cell's ability to find and then remain in an environment supporting self-renewal divisions than a measure of greater intrinsic ability to self-renew. Given this perspective, it is not surprising to find an abundance of cell–cell and cell–matrix molecules on the surface of LT-HSC. At this point, we can only speculate on the potential implications of the novel LT-HSC-enriched transcripts. However, the number and variety of adhesion molecules we have identified as upregulated on LT-HSC support an important role for such molecules in LT-HSC function. In addition, the types of cell surface molecules identified in this report allow for dynamic regulation, since several receptors are regulated by soluble ligands and proteases, and recent literature has provided convincing evidence for the importance of such mechanisms in HSC function. Arai et al. proposed that Tek actively promotes adhesion and quiescence when bound by its ligand angiopoietin1 [12], and Mmps facilitate HSC differentiation by facilitating exit from the niche [13]. Our data support these reports and reveal numerous additional candidates that are likely to participate in dynamically regulated niche-cell interactions. Clearly, LT-HSC express genes that enable them to participate in complex cell–cell and cell–matrix interactions that direct their function and developmental decisions. These and other differentially expressed transcripts allowed us to generate a model that will guide our current functional approaches to understand HSC biology.
Materials and Methods
Isolation of cells.
LT-HSC (Lin−/c-kit+/Sca1+/Thy1.1lo/Flk2−), ST-HSC (Lin−/c-kit+/Sca1+/Thy1.1lo/Flk2+), and MPP (Lin−/c-kit+/Sca1+/Thy1.1−/Flk2+) were isolated from bone marrow of 8- to 12-wk-old BA mice by double-sorting on a modified FACS Vantage (Becton-Dickinson, Mountain View, California, United States) as described [16]. Appropriate functional readout was confirmed by transplantation into lethally irradiated hosts (unpublished data).
RNA isolation, amplification, and labeling for array analysis.
Cells from individual isolations were pooled into 30,000- to 50,000-cell aliquots and RNA was extracted by Trizol. DNaseI treated RNA was amplified in two rounds by Arcturus RNA amplification kit following instructions from the manufacturer (Arcturus, Mountain View, California, United States). Amplified RNA (2.5 μg) was reverse transcribed with amino-allyl labeled deoxynucleotides (Ambion, Austin, Texas, United States) and dye-labeled by incubation with Cy3 or Cy5 for 1 h (http://cmgm.stanford.edu/pbrown/protocols/index.html). Probes were combined for the appropriate comparisons, purified with QiaQuick PCR purification kit (Qiagen, Valencia, California, United States) and eluted twice with 30 μl of EB. In each experiment, all probes were labeled by both Cy3 and Cy5 in separate reactions and combined with its comparison partner labeled in the other color in “dye-swap” experiments to eliminate bias due to differences in dye labeling efficiency. Volumes were adjusted to 28 μl by vacuum centrifugation and brought to 35 μl with a final concentration of 0.3% SDS and 3.4× SSC.
cDNA array hybridization.
Competitive hybridizations on Stanford Microarray Facility 42k mouse cDNA arrays (http://www.microarray.org) were performed in three independent experiments, each time in duplicate to accommodate “dye swaps,” for a total of six hybridizations for each pairwise comparison. Combined probes were denatured and layered on Stanford cDNA microarray slides, coverslipped, and placed in sealed, humidified chambers in a 65 °C waterbath overnight (12–16 h). Slides were washed in 0.5× SSC with 0.01%SDS for 5 min and 0.06× SSC for 2 min. Slides were scanned with a Genepix 4000A scanner using GenepixPro software. Care was taken to match Cy3 and Cy5 emission spectra during scanning.
Data analysis.
Scanned images were gridded with GenepixPro. “Bad” or missing spots were flagged by the software as well as visually. Resulting files were submitted to the Stanford Microarray Database (http://genome-www.stanford.edu/microarray) to assign identity to each spot and to normalize Cy3/Cy5 relative intensities for each slide. The log2 net intensity ratios were print-tip normalized (see below). Type I hybridizations directly comparing two different populations on the same array and therefore under exactly the same conditions enables a highly sensitive detection of differences between two populations without normalizing to a common reference that may vary from sample to sample. To further enable detection of the small, but potentially biologically highly significant, differences between the small developmental steps our three populations represent, no filters for intensity levels or spot size and shape were applied. Instead, our most stringent criterion was reproducibility. Six hybridizations in three independent experiments were performed for each pairwise comparison, and spots had to be present (not flagged as undetectable or bad) on at least five out of the six pairwise comparisons. Spots passing this criterion were analyzed for statistically significant differences using Significance Analysis of Microarrays (SAM; http://www-stat.stanford.edu/~tibs/SAM/) [89]. Because some genes were represented by multiple spots, each list was filtered by Unigene ID, clone ID, and GenBank accession number to determine the number of unique genes in each list. The number of spots and transcripts resulting from these analyses are summarized in Table 1. Significantly differentially regulated genes as defined by SAM at a false discovery rate of 10% are listed in Tables S1–S6.
Print tip normalization.
Chip- or dye-specific biases were corrected for by applying print-tip loess using the BioConductor R package to all log-ratios. First, spots with intensity signals indistinguishable from background were filtered out. To achieve high signal-to-noise, any spot with average foreground intensity within 10% of the average background level was flagged as absent. Within each print-tip group, the mean and standard deviation of the log-ratios were computed within a window of values for the log-product. The log-ratios within this window were then centered and scaled using the estimated mean and standard deviation. This was applied to each window and repeated across all print tips.
Gene Ontology (GO) analysis.
GO classification and statistics were performed using GOstat (http://gostat.wehi.edu.au/) and eGOn (http://nova2.idi.ntnu.no/egon/) with similar results. Genes upregulated in the less differentiated cell type of each comparison were combined to make an “undifferentiated” list (genes up in LT-HSC compared to either ST-HSC or MPP and genes up in ST-HSC compared to MPP). The “differentiated” list was made by combining genes upregulated in the more differentiated cell type (genes up in MPP versus either ST-HSC or LT-HSC, and genes up in ST-HSC versus LT-HSC). Similar results were obtained when genes upregulated and downregulated in the “LTvsMPP” comparison were used as the only input.
Real-time RT-PCR.
Total RNA was isolated using Trizol reagent (Invitrogen, Carlsbad, California, United States) from equal numbers of purified LT-HSC, ST-HSC, and MPP (typically 8,000−10,000 cells), digested with DNaseI, and used for reverse transcription according to the manufacturer's instructions (SuperScript II kit, Invitrogen). For qRT-PCR analysis, primer sequences were designed using Primer Express software (Applied Biosystems, Foster City, California, United States) and are available upon request. PCR amplification was performed in triplicate in a 10 μl final volume containing cDNA corresponding to approximately 200 cells, and 1× SYBR Green PCR buffer, 2 mM magnesium chloride, 0.5 mM dNTP mix with dUTP, 1 μM of each primer, 0.1 U AmpErase UNG, and 0.25 U AmpliTaq Gold. All reactions were performed in an ABI-7000 sequence detection system (Applied Biosystems) at 50 °C for 2 min and 95 °C for 10 min followed by 40 cycles at 95 °C for 15 s and 60 °C for 1 min. For each sample, expression of the HPRT gene was used to normalize the amount of the investigated transcript.
Cell surface protein levels by flow cytometry.
Cell surface protein expression was assessed by staining for LT-HSC, ST-HSC, and MPP with antibodies as described above, except an additional color was used (Cy7PE) to accommodate the additional antigen. Antibodies used were αVCAM1-biotin (eBioscience, San Diego, California, United States; clone #429), αCD9-biotin (BD Pharmingen, San Diego, California, United States; #KMC8), αSdc1-biotin (BD Pharmingen; #281–2), αESAM1 (clones 4G8 and 3C10) [25] conjugated to Alexa488, αSlamf1-PE (BioLegend, San Diego, California, United States; #TC15-12F12.2), αCD34-FITC (BD Pharmingen; #RAM34), αTek-biotin (eBioscience; #Tek4), and αJam2-FITC (SeroTec, Kidlington, Oxford, United Kingdom; #CRAM-18 F26).
Large-scale cDNA microarrays similar to the ones used for this study are available to researchers at not-for-profit research institutions by contacting the Stanford Functional Genomics Facility, CCSR 4256, 269 Campus Drive, Stanford, CA 94305-5177, United States, or by visiting http://www.microarray.org.
Supporting Information
Figure S1 Quantitative RT-PCR of a Subset of Transcripts Identified as Differentially Regulated by Array Analysis
Transcript levels are plotted as fold difference relative to LT-HSC after normalization to HPRT transcript levels. Blue bars represent ST-HSC and red bars represent MPP. Fold difference for genes off scale: Grb10, 10×; Nupr1, 29×; Sdc2, 62×; and Tgm2, 37×.
(554 KB PDF)
Click here for additional data file.
Figure S2 Chromosomal Distribution of Genes Differentially Regulated Between LT-HSC and MPP
The distribution of genes that are differentially regulated between LT-HSC (green) and MPP (red) were normalized to total number of differentially regulated genes with known chromosomal location for each population.
(71 KB PDF)
Click here for additional data file.
Table S1 Transcripts Upregulated in ST-HSC Compared to LT-HSC
(181 KB XLS)
Click here for additional data file.
Table S2 Transcripts Upregulated in LT-HSC Compared to ST-HSC
(99 KB XLS)
Click here for additional data file.
Table S3 Transcripts Upregulated in MPP Compared to LT-HSC
(160 KB XLS)
Click here for additional data file.
Table S4 Transcripts Upregulated in LT-HSC Compared to MPP
(260 KB XLS)
Click here for additional data file.
Table S5 Transcripts Upregulated in MPP Compared to ST-HSC
(18 KB XLS)
Click here for additional data file.
Table S6 Transcripts Upregulated in ST-HSC Compared to MPP
(53 KB XLS)
Click here for additional data file.
Table S7 Transcripts Upregulated in LT-HSC Mapped to the X Chromosome
(62 KB PDF)
Click here for additional data file.
Table S8 Additional References
(52 KB XLS)
Click here for additional data file.
Accession Numbers
The LocusLink accession numbers (LLIDs) (http://www.ncbi.nlm.nih.gov/projects/LocusLink/) of the proteins discussed in this paper are 4a2 (12827), Abcg2 (26357), Aldh1a1 (11668), Aldh7a1 (110695), angiopoietin 1 (11600), ApoE (11816), App (11820), Atrx (22589), biglycan (12111), Blnk (17060), Calsyntenin1 (65945), catalase (12359), Ccl3 (20302), Ccl4 (20304), Ccl9 (20308), Ccr5 (12774), CD34 (12490), CD63 (12512), CD9 (12527), Cenpe (229841), Cenph (26886), Ches1 (71375), Claudin12 (64945), Claudin22 (75677), c-myc (17869), collagens type 4a1 (12826), cortactin (13043), cyclin A2 (12428), cyclin B1 (268697), cyclin B2 (12442), cyclin D1 (12443), cyclin D3 (12445), Cyp2j6 (13110), Dachshund1 (13134), Dag1/dystroglycan (13138), Ddx4 (13206), Emilin1 (100952), Epb4.1 (269587), ESAM1 (69524), Eya2 (14049), Fgd1 (14163), Fstl1 (14314), Gata1 (14460), Gata2 (14461), Gelsolin (227753), Gja1 (14609), Gjb5 (14622), Grb10 (14783), Grb2 (14784), Gsta1 (14857), Gstk1 (76263), Gstm1 (14862), heparanase (15442), Hmgb3 (15354), Hoxa5 (15402), Hoxa9 (15405), Hoxb5 (15413), Ier5 (15939), Igh6 (16019), IL1β (16176), IL10rα (16154), IL10rβ (16155), IL17r (16172), IL1r1 (16177), Itga6 (16403), Jam1/F11R (16456), Jam2 (67374), Jam3 (83964), Ki67 (17345), Kif22 (110033), Kif23 (71819), Kif4 (16571), Kifc5a (94116), Lims1 (110829), Map17 (67182), Mllt4 (17356), Mlp (17357), Mmp15 (17388), Mmp16 (17389), Mmp2 (17390), Mpeg1 (17476), necdin (17984), Neogenin (18007), Nfe2 (18022), Notch1 (18128), Notch1 (18128), p53 (22059), p57/Cdkn1c (12577), Pbx1 (18514), Pbx3 (18516), Robo4 (74144), sarcoglycanE (20392), Satb1 (20230), Sdc1 (20969), Sdc2 (15529), Sesn1 (140742), Ski (20481), Slamf1 (27218), Slit-like2 (246154), Smc1l1 (24061), Smc2l1 (14211), Smc4l1 (70099), Sox18 (20672), Sox6 (20679), Spag5 (54141), Srgap2 (14270), Tcf12 (21406), Tek (21687), Tgm2 (21817), Timp2 (21858), Timp3 (21859), Tjp1/ZO-1 (21872), Top2a (21973), and VCAM1 (22329).
We thank D. Vestweber and S. Butz for anti-Esam1 antibodies; J. Christensen for advice on cell sorting; and E. Passegué, D. Rossi, D. Bryder, D. Battacharya, L. Ailles, and other members of the Weissman lab for critical reading of the manuscript and helpful suggestions. We also thank L. Jerabek for superb laboratory management and C. Richter for antibody preparations. Supported in part by grants 5P01DK053074 and 2R01CA086065 (to ILW), 1R01GM068570–01 (to SK) and the Cancer Research Institute (to ECF).
Competing interests. As a former advisory board member of Amgen, ILW owns significant Amgen stock. He also cofounded and consulted for Systemix; cofounded Cellerant Therapeutics, a spin-off from Systemix Novartis to transplant human HSC; and is a cofounder and a director of the company Stem Cells, which is involved in the isolation and study of human central nervous system stem cells, liver-repopulating cells, and pancreatic islet/progenitor cells.
Author contributions. ECF, SSP, and ILW conceived and designed the experiments. ECF, SSP and GCH performed the experiments. All authors analyzed the data. SK and JMS contributed reagents/materials/analysis tools. ECF wrote the paper.
Note Added in Proof The information credited as a personal communication from S. Morrison on page 4 can be found in the following published article: Kiel MJ, Yilmaz OH, Iwashita T, Yilmaz OH, Terhorst C, et al. (2005) SLAM family receptors distinguish hematopoietic stem and progenitor cells and reveal endothelial niches for stem cells. Cell 121: 1109–1121.
Abbreviations
Appamyloid beta precursor protein
CMLchronic myelogenous leukemia
ECMextracellular matrix
ESTexpressed sequence tag
GOGene Ontology
HSChematopoietic stem cell(s)
LTlong-term
Mmpmatrix metalloproteinase
MPPmultipotent progenitor(s)
SAMSignificance Analysis of Microarrays
STshort-term
==== Refs
References
Ivanova NB Dimos JT Schaniel C Hackney JA Moore KA 2002 A stem cell molecular signature Science 298 601 604 12228721
Ramalho-Santos M Yoon S Matsuzaki Y Mulligan RC Melton DA 2002 “Stemness”: Transcriptional profiling of embryonic and adult stem cells Science 298 597 600 12228720
Venezia TA Merchant AA Ramos CA Whitehouse NL Young AS 2004 Molecular signatures of proliferation and quiescence in hematopoietic stem cells PLoS Biol 2 e301 15459755
Terskikh AV Miyamoto T Chang C Diatchenko L Weissman IL 2003 Gene expression analysis of purified hematopoietic stem cells and committed progenitors Blood 102 94 101 12623852
Akashi K He X Chen J Iwasaki H Niu C 2003 Transcriptional accessibility for genes of multiple tissues and hematopoietic lineages is hierarchically controlled during early hematopoiesis Blood 101 383 389 12393558
Antonchuk J Sauvageau G Humphries RK 2001 HOXB4 overexpression mediates very rapid stem cell regeneration and competitive hematopoietic repopulation Exp Hematol 29 1125 1134 11532354
Park IK Qian D Kiel M Becker MW Pihalja M 2003 Bmi-1 is required for maintenance of adult self-renewing haematopoietic stem cells Nature 423 302 305 12714971
Reya T Duncan AW Ailles L Domen J Scherer DC 2003 A role for Wnt signalling in self-renewal of haematopoietic stem cells Nature 423 409 414 12717450
Varnum-Finney B Purton LE Yu M Brashem-Stein C Flowers D 1998 The Notch ligand, Jagged-1, influences the development of primitive hematopoietic precursor cells Blood 91 4084 4091 9596653
Zhang J Niu C Ye L Huang H He X 2003 Identification of the haematopoietic stem cell niche and control of the niche size Nature 425 836 841 14574412
Calvi LM Adams GB Weibrecht KW Weber JM Olson DP 2003 Osteoblastic cells regulate the haematopoietic stem cell niche Nature 425 841 846 14574413
Arai F Hirao A Ohmura M Sato H Matsuoka S 2004 Tie2/angiopoietin-1 signaling regulates hematopoietic stem cell quiescence in the bone marrow niche Cell 118 149 161 15260986
Heissig B Hattori K Dias S Friedrich M Ferris B 2002 Recruitment of stem and progenitor cells from the bone marrow niche requires MMP-9 mediated release of kit-ligand Cell 109 625 637 12062105
Wagers AJ Allsopp RC Weissman IL 2002 Changes in integrin expression are associated with altered homing properties of Lin−/lo Thy1.1lo Sca-1+ c-kit+ hematopoietic stem cells following mobilization by cyclophosphamide/granulocyte colony-stimulating factor Exp Hematol 30 176 185 11823053
Potocnik AJ Brakebusch C Fassler R 2000 Fetal and adult hematopoietic stem cells require beta1 integrin function for colonizing fetal liver, spleen, and bone marrow Immunity 12 653 663 10894165
Christensen JL Weissman IL 2001 Flk-2 is a marker in hematopoietic stem cell differentiation: A simple method to isolate long-term stem cells Proc Natl Acad Sci U S A 98 14541 14546 11724967
Zhou S Schuetz JD Bunting KD Colapietro AM Sampath J 2001 The ABC transporter Bcrp1/ABCG2 is expressed in a wide variety of stem cells and is a molecular determinant of the side-population phenotype Nat Med 7 1028 1034 11533706
Osawa M Hanada K Hamada H Nakauchi H 1996 Long-term lymphohematopoietic reconstitution by a single CD34-low/negative hematopoietic stem cell Science 273 242 245 8662508
Iwama A Hamaguchi I Hashiyama M Murayama Y Yasunaga K 1993 Molecular cloning and characterization of mouse TIE and TEK receptor tyrosine kinase genes and their expression in hematopoietic stem cells Biochem Biophys Res Commun 195 301 309 8395828
Hess DA Meyerrose TE Wirthlin L Craft TP Herrbrich PE 2004 Functional characterization of highly purified human hematopoietic repopulating cells isolated according to aldehyde dehydrogenase activity Blood 104 1648 1655 15178579
Morrison SJ Qian D Jerabek L Thiel BA Park IK 2002 A genetic determinant that specifically regulates the frequency of hematopoietic stem cells J Immunol 168 635 642 11777956
Ashburner M Ball CA Blake JA Botstein D Butler H 2000 Gene ontology: Tool for the unification of biology. The Gene Ontology Consortium Nat Genet 25 25 29 10802651
Wong RC Pebay A Nguyen LT Koh KL Pera MF 2004 Presence of functional gap junctions in human embryonic stem cells Stem Cells 22 883 889 15536180
Tazuke SI Schulz C Gilboa L Fogarty M Mahowald AP 2002 A germline-specific gap junction protein required for survival of differentiating early germ cells Development 129 2529 2539 11973283
Nasdala I Wolburg-Buchholz K Wolburg H Kuhn A Ebnet K 2002 A transmembrane tight junction protein selectively expressed on endothelial cells and platelets J Biol Chem 277 16294 16303 11847224
Hirata K Ishida T Penta K Rezaee M Yang E 2001 Cloning of an immunoglobulin family adhesion molecule selectively expressed by endothelial cells J Biol Chem 276 16223 16231 11279107
Johnson KG Ghose A Epstein E Lincecum J O'Connor MB 2004 Axonal heparan sulfate proteoglycans regulate the distribution and efficiency of the repellent slit during midline axon guidance Curr Biol 14 499 504 15043815
Wang B Xiao Y Ding BB Zhang N Yuan X 2003 Induction of tumor angiogenesis by Slit-Robo signaling and inhibition of cancer growth by blocking Robo activity Cancer Cell 4 19 29 12892710
Prasad A Fernandis AZ Rao Y Ganju RK 2004 Slit protein-mediated inhibition of CXCR4-induced chemotactic and chemoinvasive signaling pathways in breast cancer cells J Biol Chem 279 9115 9124 14645233
Bashaw GJ Goodman CS 1999 Chimeric axon guidance receptors: The cytoplasmic domains of slit and netrin receptors specify attraction versus repulsion Cell 97 917 926 10399919
Stein E Tessier-Lavigne M 2001 Hierarchical organization of guidance receptors: Silencing of netrin attraction by slit through a Robo/DCC receptor complex Science 291 1928 1938 11239147
Lapidot T 2001 Mechanism of human stem cell migration and repopulation of NOD/SCID and B2mnull NOD/SCID mice. The role of SDF-1/CXCR4 interactions Ann N Y Acad Sci 938 83 95 11458529
Wright DE Wagers AJ Gulati AP Johnson FL Weissman IL 2001 Physiological migration of hematopoietic stem and progenitor cells Science 294 1933 1936 11729320
Steigemann P Molitor A Fellert S Jackle H Vorbruggen G 2004 Heparan sulfate proteoglycan syndecan promotes axonal and myotube guidance by slit/robo signaling Curr Biol 14 225 230 14761655
Salmivirta M Jalkanen M 1995 Syndecan family of cell surface proteoglycans: Developmentally regulated receptors for extracellular effector molecules Experientia 51 863 872 7556568
Beauvais DM Rapraeger AC 2004 Syndecans in tumor cell adhesion and signaling Reprod Biol Endocrinol 2 3 14711376
Irizarry MC Deng A Lleo A Berezovska O Von Arnim CA 2004 Apolipoprotein E modulates gamma-secretase cleavage of the amyloid precursor protein J Neurochem 90 1132 1143 15312168
Snow AD Kinsella MG Parks E Sekiguchi RT Miller JD 1995 Differential binding of vascular cell-derived proteoglycans (perlecan, biglycan, decorin, and versican) to the beta-amyloid protein of Alzheimer's disease Arch Biochem Biophys 320 84 95 7793988
Kocher O Cheresh P Lee SW 1996 Identification and partial characterization of a novel membrane-associated protein (MAP17) up-regulated in human carcinomas and modulating cell replication and tumor growth Am J Pathol 149 493 500 8701988
Levy S Todd SC Maecker HT 1998 CD81 (TAPA-1): A molecule involved in signal transduction and cell adhesion in the immune system Annu Rev Immunol 16 89 109 9597125
Nakorn TN Miyamoto T Weissman IL 2003 Characterization of mouse clonogenic megakaryocyte progenitors Proc Natl Acad Sci U S A 100 205 210 12490656
Schmidt C Kunemund V Wintergerst ES Schmitz B Schachner M 1996 CD9 of mouse brain is implicated in neurite outgrowth and cell migration in vitro and is associated with the alpha 6/beta 1 integrin and the neural adhesion molecule L1 J Neurosci Res 43 12 31 8838570
Park KR Inoue T Ueda M Hirano T Higuchi T 2000 CD9 is expressed on human endometrial epithelial cells in association with integrins alpha(6), alpha(3) and beta(1) Mol Hum Reprod 6 252 257 10694273
Gutierrez-Lopez MD Ovalle S Yanez-Mo M Sanchez-Sanchez N Rubinstein E 2003 A functionally relevant conformational epitope on the CD9 tetraspanin depends on the association with activated beta1 integrin J Biol Chem 278 208 218 12411441
Tumbar T Guasch G Greco V Blanpain C Lowry WE 2004 Defining the epithelial stem cell niche in skin Science 303 359 363 14671312
Fortunel NO Otu HH Ng HH Chen J Mu X 2003 Comment on “‘Stemness': Transcriptional profiling of embryonic and adult stem cells” and “a stem cell molecular signature.” Science 302 393.
Su L Hattori M Moriyama M Murata N Harazaki M 2003 AF-6 controls integrin-mediated cell adhesion by regulating Rap1 activation through the specific recruitment of Rap1GTP and SPA-1 J Biol Chem 278 15232 15238 12590145
Miyamoto T Iwasaki H Reizis B Ye M Graf T 2002 Myeloid or lymphoid promiscuity as a critical step in hematopoietic lineage commitment Dev Cell 3 137 147 12110174
Cross MA Enver T 1997 The lineage commitment of haemopoietic progenitor cells Curr Opin Genet Dev 7 609 613 9388776
Katsube T Takahisa M Ueda R Hashimoto N Kobayashi M 1998 Cortactin associates with the cell-cell junction protein ZO-1 in both Drosophila and mouse J Biol Chem 273 29672 29677 9792678
Hou P Estrada L Kinley AW Parsons JT Vojtek AB 2003 Fgd1, the Cdc42 GEF responsible for faciogenital dysplasia, directly interacts with cortactin and mAbp1 to modulate cell shape Hum Mol Genet 12 1981 1993 12913069
Vecchione A Marchese A Henry P Rotin D Morrione A 2003 The Grb10/Nedd4 complex regulates ligand-induced ubiquitination and stability of the insulin-like growth factor I receptor Mol Cell Biol 23 3363 3372 12697834
Riedel H 2004 Grb10 exceeding the boundaries of a common signaling adapter Front Biosci 9 603 618 14766395
Venezia V Russo C Repetto E Salis S Dolcini V 2004 Apoptotic cell death influences the signaling activity of the amyloid precursor protein through ShcA and Grb2 adaptor proteins in neuroblastoma SH-SY5Y cells J Neurochem 90 1359 1370 15341520
Yano S Arroyo N Yano N 2004 Catalase binds Grb2 in tumor cells when stimulated with serum or ligands for integrin receptors Free Radic Biol Med 36 1542 1554 15182856
Ito K Hirao A Arai F Matsuoka S Takubo K 2004 Regulation of oxidative stress by ATM is required for self-renewal of haematopoietic stem cells Nature 431 997 1002 15496926
Guo L Wu C 2002 Regulation of fibronectin matrix deposition and cell proliferation by the PINCH-ILK-CH-ILKBP complex FASEB J 16 1298 1300 12060675
Wu C 1999 Integrin-linked kinase and PINCH: Partners in regulation of cell-extracellular matrix interaction and signal transduction J Cell Sci 112 ( Pt 24): 4485 4489 10574698
Pearson-White S Deacon D Crittenden R Brady G Iscove N 1995 The ski/sno protooncogene family in hematopoietic development Blood 86 2146 2155 7662963
Atanasoski S Notterpek L Lee HY Castagner F Young P 2004 The protooncogene Ski controls Schwann cell proliferation and myelination Neuron 43 499 511 15312649
Nomura T Tanikawa J Akimaru H Kanei-Ishii C Ichikawa-Iwata E 2004 Oncogenic activation of c-Myb correlates with a loss of negative regulation by TIF1beta and Ski J Biol Chem 279 16715 16726 14761981
Wilson A Murphy MJ Oskarsson T Kaloulis K Bettess MD 2004 c-Myc controls the balance between hematopoietic stem cell self-renewal and differentiation Genes Dev 18 2747 2763 15545632
Li X Perissi V Liu F Rose DW Rosenfeld MG 2002 Tissue-specific regulation of retinal and pituitary precursor cell proliferation Science 297 1180 1183 12130660
Xu PX Cheng J Epstein JA Maas RL 1997 Mouse Eya genes are expressed during limb tendon development and encode a transcriptional activation function Proc Natl Acad Sci U S A 94 11974 11979 9342347
Li X Oghi KA Zhang J Krones A Bush KT 2003 Eya protein phosphatase activity regulates Six1-Dach-Eya transcriptional effects in mammalian organogenesis Nature 426 247 254 14628042
Hosking BM Wang SC Downes M Koopman P Muscat GE 2004 The VCAM-1 gene that encodes the vascular cell adhesion molecule is a target of the Sry-related high mobility group box gene, Sox18 J Biol Chem 279 5314 5322 14634005
Fernandez-Lloris R Vinals F Lopez-Rovira T Harley V Bartrons R 2003 Induction of the Sry-related factor SOX6 contributes to bone morphogenetic protein-2-induced chondroblastic differentiation of C3H10T1/2 cells Mol Endocrinol 17 1332 1343 12677004
Hamada-Kanazawa M Ishikawa K Nomoto K Uozumi T Kawai Y 2004 Sox6 overexpression causes cellular aggregation and the neuronal differentiation of P19 embryonic carcinoma cells in the absence of retinoic acid FEBS Lett 560 192 198 14988021
Nemeth MJ Cline AP Anderson SM Garrett-Beal LJ Bodine DM 2005 Hmgb3 deficiency deregulates proliferation and differentiation of common lymphoid and myeloid progenitors Blood 105 627 634 15358624
Uittenbogaard M Chiaramello A 2002 Expression of the bHLH transcription factor Tcf12 (ME1) gene is linked to the expansion of precursor cell populations during neurogenesis Brain Res Gene Expr Patterns 1 115 121 15018808
So CW Karsunky H Wong P Weissman IL Cleary ML 2004 Leukemic transformation of hematopoietic progenitors by MLL-GAS7 in the absence of Hoxa7 or Hoxa9 Blood 103 3192 3199 15070702
Thorsteinsdottir U Mamo A Kroon E Jerome L Bijl J 2002 Overexpression of the myeloid leukemia-associated Hoxa9 gene in bone marrow cells induces stem cell expansion Blood 99 121 129 11756161
Fujiwara Y Komiya T Kawabata H Sato M Fujimoto H 1994 Isolation of a DEAD-family protein gene that encodes a murine homolog of Drosophila vasa and its specific expression in germ cell lineage Proc Natl Acad Sci U S A 91 12258 12262 7991615
De La Fuente R Viveiros MM Wigglesworth K Eppig JJ 2004 ATRX, a member of the SNF2 family of helicase/ATPases, is required for chromosome alignment and meiotic spindle organization in metaphase II stage mouse oocytes Dev Biol 272 1 14 15242786
Goodell MA Brose K Paradis G Conner AS Mulligan RC 1996 Isolation and functional properties of murine hematopoietic stem cells that are replicating in vivo J Exp Med 183 1797 1806 8666936
Morrison SJ Weissman IL 1994 The long-term repopulating subset of hematopoietic stem cells is deterministic and isolatable by phenotype Immunity 1 661 673 7541305
Spangrude GJ Johnson GR 1990 Resting and activated subsets of mouse multipotent hematopoietic stem cells Proc Natl Acad Sci U S A 87 7433 7437 1977160
Yang L Bryder D Adolfsson J Nygren J Mansson R 2005 Identification of Lin− Sca1+ kit+ CD34+ Flt3− short-term hematopoietic stem cells capable of rapidly reconstituting and rescuing myeloablated transplant recipients Blood 105 2717 2723 15572596
Wognum AW Eaves AC Thomas TE 2003 Identification and isolation of hematopoietic stem cells Arch Med Res 34 461 475 14734086
Reya T Morrison SJ Clarke MF Weissman IL 2001 Stem cells, cancer, and cancer stem cells Nature 414 105 111 11689955
Passegué E Wagner EF Weissman IL 2004 JunB deficiency leads to a myeloproliferative disorder arising from hematopoietic stem cells Cell 119 431 443 15507213
Jamieson CH Ailles LE Dylla SJ Muijtjens M Jones C 2004 Granulocyte-macrophage progenitors as candidate leukemic stem cells in blast-crisis CML N Engl J Med 351 657 667 15306667
Ikuta K Ingolia DE Friedman J Heimfeld S Weissman IL 1991 Mouse hematopoietic stem cells and the interaction of c-kit receptor and steel factor Int J Cell Cloning 9 451 460 1720154
Visnjic D Kalajzic Z Rowe DW Katavic V Lorenzo J 2004 Hematopoiesis is severely altered in mice with an induced osteoblast deficiency Blood 103 3258 3264 14726388
Hackney JA Charbord P Brunk BP Stoeckert CJ Lemischka IR 2002 A molecular profile of a hematopoietic stem cell niche Proc Natl Acad Sci U S A 99 13061 13066 12226475
Moore KA Ema H Lemischka IR 1997 In vitro maintenance of highly purified, transplantable hematopoietic stem cells Blood 89 4337 4347 9192756
Fleming WH Alpern EJ Uchida N Ikuta K Spangrude GJ 1993 Functional heterogeneity is associated with the cell cycle status of murine hematopoietic stem cells J Cell Biol 122 897 902 8349737
Cheshier SH Morrison SJ Liao X Weissman IL 1999 In vivo proliferation and cell cycle kinetics of long-term self-renewing hematopoietic stem cells Proc Natl Acad Sci U S A 96 3120 3125 10077647
Tusher VG Tibshirani R Chu G 2001 Significance analysis of microarrays applied to the ionizing radiation response Proc Natl Acad Sci U S A 98 5116 5121 11309499
|
16151515
|
PMC1200425
|
CC BY
|
2021-01-05 08:00:24
|
no
|
PLoS Genet. 2005 Sep 2; 1(3):e28
|
utf-8
|
PLoS Genet
| 2,005 |
10.1371/journal.pgen.0010028
|
oa_comm
|
==== Front
PLoS GenetPLoS GenetpgenplgeplosgenPLoS Genetics1553-73901553-7404Public Library of Science San Francisco, USA 1615151610.1371/journal.pgen.001003005-PLGE-RA-0166R2plge-01-03-01Research ArticleCell BiologyGenetics/Gene ExpressionNematodesCaenorhabditisUncoupling of Longevity and Telomere Length in C. elegans
Life Span Independent of Telomere LengthRaices Marcela 1Maruyama Hugo 12Dillin Andrew 1Karlseder Jan 1*1 The Salk Institute for Biological Studies, La Jolla, California, United States of America
2 Graduate School of Biostudies, Kyoto University, Kyoto, Japan
Kim Stuart EditorStanford University School of Medicine, United States of America*To whom correspondence should be addressed. E-mail: [email protected] 2005 2 9 2005 1 3 e3012 7 2005 1 8 2005 Copyright: © 2005 Raices et al.2005This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.The nematode Caenorhabditis elegans, after completing its developmental stages and a brief reproductive period, spends the remainder of its adult life as an organism consisting exclusively of post-mitotic cells. Here we show that telomere length varies considerably in clonal populations of wild-type worms, and that these length differences are conserved over at least ten generations, suggesting a length regulation mechanism in cis. This observation is strengthened by the finding that the bulk telomere length in different worm strains varies considerably. Despite the close correlation of telomere length and clonal cellular senescence in mammalian cells, nematodes with long telomeres were neither long lived, nor did worm populations with comparably short telomeres exhibit a shorter life span. Conversely, long-lived daf-2 and short-lived daf-16 mutant animals can have either long or short telomeres. Telomere length of post-mitotic cells did not change during the aging process, and the response of animals to stress was found independent of telomere length. Collectively, our data indicate that telomere length and life span can be uncoupled in a post-mitotic setting, suggesting separate pathways for replication-dependent and -independent aging.
Synopsis
The worm Caenorhabditis elegans has historically been used as a powerful model to study organismal aging. After a brief reproductive period, an adult worm consists of 959 post-mitotic cells. Telomeres, the natural ends of linear chromosomes, have long been implicated in the aging process of mitotic cells, and telomere shortening has been proposed to be a limiting factor in cell division. The authors show that telomere length of C. elegans chromosomes is independent of organismal aging. Worm telomeres were characterized in detail, and found to vary considerably in individual clones. These differences were conserved over several populations, suggesting that telomere length is controlled by similar mechanisms in worms and mammals, emphasizing the value of C. elegans as a model for telomere biology. However, telomere length has not been found to determine the potential life span of the animals, since worms with short telomeres lived as long as worms with long telomeres. Vice versa, worms that have an altered life span due to mutations in the insulin receptor pathway do not show any changes in telomere length. Telomere length was also found to be constant in isolated, aging worm populations, suggesting that organismal aging can be uncoupled from mitotic aging.
Citation:Raices M, Maruyama H, Dillin A, Karlseder J (2005) Uncoupling of longevity and telomere length in C. elegans. PLoS Genet 1(3): e30.
==== Body
Introduction
Telomeres, the specialized nucleoprotein structures that protect chromosome ends from degradation and fusions, have long been implicated in the replicative aging process of dividing cells [1–3]. Cellular aging in mitotic cultures is defined by initial telomere length, and the rate of telomere shortening per division, in the absence of telomerase. Human cells lose approximately 100 basepairs of telomeric repeats per cell division, and can undergo 50 to 80 doubling cycles until they enter a terminally differentiated stage, called replicative senescence.
In the mouse, it is well established that telomere maintenance and protection play a key role in the germline during early development and in highly proliferative organs [4,5]. However, telomere function in terminally differentiated cells, and the implications for post-mitotic senescence and organismal longevity remain unclear. The mouse model is not optimally suited to study the relation between organismal longevity and telomeres, considering that the cells of rapidly proliferating organs are strongly affected by telomere attrition in late generations by the targeted deletion of the telomerase RNA subunit, potentially masking the effects of short telomeres on differentiated cells. Nonetheless, it has been reported that the short telomeres in wild-derived strains of mice have no inverse effect on longevity, suggesting that organismal life span in this setting is independent of telomere length [6].
The nematode Caenorhabditis elegans does not suffer from the drawbacks of the mammalian systems, making it an optimal system to investigate a potential telomere-organismal life span relationship. Fully developed adult C. elegans hermaphrodites consist of 959 non-dividing somatic cells. To date, there is no example of cellular replication in adult C. elegans, although endoreduplication of nuclear DNA within a few hypodermal cells has been noted [7]. C. elegans therefore represents a unique model system to study the importance of telomere function in a post-mitotic setting and its implications on organismal aging.
The telomeric repeat sequence in the nematode C. elegans (TTAGGC) is similar to the telomeric sequence in mammals (TTAGGG), and this sequence has been found to be sufficient for the successful capping of chromosomes [8]. Few proteins that bind to nematode telomeres have been identified so far [9–12]. Worm telomere function has been analyzed only in the context of cell division, and, as for other higher eukaryotes, telomere maintenance has been shown to be essential for genome stability in the germline [13]. The relationship between organismal life span and telomere length in the worm has been controversial. clk-2, a gene with homology to Saccharomyces cerevisiae Tel2p, acts in the DNA damage response pathway and has been suggested to have little effect on telomere length [14]. In a similar study, clk-2 mutations have been suggested to alter telomere length [15,16] while at the same time conferring a longer life span to the animals [17]. On the other hand, it has been proposed that the overexpression of the RNP A1 homolog hrp-1 leads to telomere elongation, which in turn allows for a modest elongation in life span [12]. At this point, it is unclear whether the effects on life span are due to altered expression profiles or altered telomere length in the hrp-1 overexpressing strains.
Here, we characterize C. elegans telomeres in detail in clones derived from individual worms, observe the life term of clonal populations and worm strains with different telomere length settings, investigate the effect of life span–altering mutations on telomere length, and describe the influence of telomere length on stress resistance.
Results/Discussion
Telomere Length in C. elegans Is Clonal and Inherited
To study telomere function in C. elegans, we first determined the extent of telomere variation among wild-type worms. Telomere length was analyzed in clonal populations isolated from the Bristol N2 strain [18]. The length of terminal restriction fragments was determined by hybridizing (TTAGGC)4 and (GCCTAA)4 probes to C. elegans genomic DNA digested with AluI and MboI, restriction enzymes that liberate the chromosome ends. Telomere length ranged from 2 to 9 kilobasepairs (kb), and was heterogeneous among different clones (Figure 1A). While most isolated N2 clones contained fairly short telomeres in the size range of 2.5–3 kb (note clones A, C, D, and F), some clones showed bands of higher molecular weight, up to 9 kb (note clones B and E). Exonuclease Bal31 treatment prior to restriction fragmentation of genomic DNA samples confirmed that these bands represented terminal repeats and corresponded to telomeres (Figure 1B), as has been suggested earlier [8]. Within 10 min after treatment with Bal31, signal intensity was reduced, and the hybridization signal was almost completely removed after 90 min, indicating that the signal originated from DNA termini (Figure 1B), and pointing out the background signal resulting from internal cross-hybridizing sequences. These results are consistent with mammalian telomeres of similar length [19].
Figure 1 Clonal Variation and Conservation of Telomere Length in C. elegans
(A) AluI/MboI digests of genomic DNA from six Bristol N2 clones was separated by gel electrophoresis and hybridized with (TTAGGC)4 or (GCCTAA)4 probes as described [28]. Fragment size is indicated in kb, and internal repeats are indicated by an asterisk.
(B) DNA from six individual N2 clones was incubated with Bal31 for the indicated time, and then processed as in (A).
(C) Six individual N2 clones were propagated for ten generations, and telomere length was determined as described.
(D) Seven different C. elegans strains were examined for their telomere length as described in (A).
Given the telomere length differences between individual clonal worm populations, we next examined whether variations in telomere length were stably maintained throughout generations. Single worms were isolated to create individual, isolated colonies of worms. Because C. elegans exists almost exclusively as a self-fertile hermaphrodite, we grew each clonal population for several generations. Each of the clones was expanded and split into two. One part was immediately harvested and used for genomic DNA extraction for telomere length analysis; the other part was subsequently passaged for ten more population doublings, and the resulting worms were harvested for genomic DNA extraction. We found that following this regiment, even after ten generations the overall telomere length of each clonal population remained relatively constant, suggesting that individual telomere length regulation mechanisms are stably maintained and genetically conserved (Figure 1C).
The results of our last experiment suggested that telomere length among a population of worms was variable, yet stable. Consistent with this hypothesis, single telomere analysis of the left arm of Chromosome five (VL) between different, genetically isolated, wild-type C. elegans strains exhibited strong variations in telomere length, suggesting that length heterogeneity among strains could also exist at other chromosome ends [20]. We expanded this study to analyze the bulk telomere length in mixed populations of seven different wild-type worm strains, including the standard N2 worms. Telomere length distributions were different in all the strains examined (Figure 1D and Table 1). The average telomere length ranged from 2.6 kb in wild-type N2 worms to 16 kb in TR403 worms, with maximal telomere length reaching 7 kb in N2 worms and 17.8 kb in TR403 worms. Bal31 digests confirmed that the signal resulted from terminal repeats (unpublished data). These data suggest that both Southern blotting and single telomere analysis are appropriate methods for telomere length measurement in C. elegans. Altogether, these results demonstrate that there is strong heterogeneity in telomere length within individual C. elegans clones, as well as between different C. elegans strains.
Table 1 Survival, Average and Maximal Telomere Length of Different Worm Strains
Post-Mitotic Life Span Is Independent of Telomere Length
Because it has been suggested that telomere length may play a role in organismal aging [5] and, more recently, post-mitotic aging of C. elegans [12], we tested whether the large variations in telomere length in the different wild-type worm strains analyzed were reflected in the life span of these animals. Despite the observed differences in telomere length, no significant variations in life span were detected among any of the seven worm strains (Table 1). Even TR403 worms with very long telomeres in the range of 16 kb displayed life spans that were not significantly different (p = 0.14) from those of standard Bristol N2 worms (Table 1), suggesting that long telomeres do not convey a noticeable survival advantage for the nematodes.
To further establish if there was any correlation between life span regulation and telomere length, we analyzed whether variations in the expression of daf-2 or daf-16, two key components of the insulin/IGF-1 signaling pathway known to be involved in C. elegans life span regulation [21,22], affected telomere length. To control for possible variation due to genetic drift that could accompany daf-2 and daf-16 mutant worms, we utilized a single wild-type strain, N2, and RNAi directed toward either daf-2 or daf-16 [23]. This approach ensured genetic identity among the worms examined in our experiment. Single worms were isolated to create individual colonies. Each clonal population was grown on the different bacteria expressing dsRNA of either daf-2 or daf-16 and divided into three parts. One part was immediately harvested and used for genomic DNA extraction, one part was used for life span analysis, and another part was subsequently passaged to analyze telomere length at the end of the life span. N2 worms grown on control bacteria displayed a mean life span of 19.8 + 0.5 d. However, in all clones analyzed, daf-2 (RNAi) increased the mean life span to 42.3 + 2.1 d (p < 0.0001), and daf-16 (RNAi) led to a significant decrease in life span (14.9 + 0.5 d, p < 0.0001) (Figure 2A and Table 2). Telomere length in nine individual clones was determined at the beginning (F1) and at the end of every experiment (Figure 2B–D). Since the life span of control daf-16 (RNAi) and daf-2 (RNAi) worms differs considerably, the end of the experiment correlates to F5 for control, F4 for daf-16, and F8 for daf-2 animals. Despite the significant changes in survival, no correlation between life span and telomere length could be observed (Table 2). Average telomere length between control worms (2.5 kb), daf-2 (RNAi) worms (2.5 kb), and daf-16 (RNAi) worms (2.7 kb) did not vary, and the maximal telomere length settings did not correlate with long life span (Table 2). The longest clonal telomere length setting (> 6.5 kb), as well as the shortest maximal telomere length setting (3.5 kb) were both observed in short-lived daf-16 (RNAi) treated worms (daf-16 [RNAi] N2 clone 2 and clone 9, respectively).
Figure 2 Organismal Life Span Is Independent of Telomere Length
(A) Life span of N2 worms under control (green), daf-2 (blue), and daf-16 (red) RNAi suppression conditions. Worms were grown on bacteria expressing RNAi constructs targeting the indicated genes, and life span was assessed as described [21].
(B) Telomere length of nine individual control N2 clones was assessed as described. Samples were collected at the beginning (B) and end (E) of the experiment, and fragment size is indicated in kb.
(C) Telomere length of nine individual daf-2 (RNAi) N2 clones was assessed as described in (B).
(D) Telomere length of nine individual daf-16 (RNAi) N2 clones was assessed as described in (B).
Table 2 Survival, Average and Maximal Telomere Length of Control, daf-2 (RNAi) and daf-16 (RNAi) Clones
To verify whether the results gathered by suppression of daf-2 and daf-16 by RNAi could be confirmed in a mutant background, we analyzed nine individual clones from long-lived daf-2(e1370) and short-lived daf-16(mu86) null mutant animals for life span and telomere length. Wild-type N2 clones displayed an average life span of 18 + 0.9 d, all the different clones isolated from the daf-2(e1370) strain had an extended average life span of 37.7 + 1.7 d (p < 0.0001), and all the daf-16(mu86) null mutant clones lived significantly shorter than wild-type N2 worms, with an average life span of 12.9 + 0.5 d (p < 0.0001) (Figure S1A and Table S1). Telomere length analysis of each individual clone was performed, and the average and maximal telomere length settings were determined (Figure S1B–D). Despite the significant differences in life span, no obvious correlation between life span regulation and telomere length could be found (Table S1). Long-lived daf-2(e1370) mutant animals had an average telomere length of 3.8 kb, and short-lived daf-16(mu86) null mutant animals had an average telomere length of 3.7 kb. Control animals were observed to have slightly shorter telomeres than daf-2(e1370) and daf-16(mu86) mutant animals (Table S1). This is most likely traceable to the original clones used to generate the null mutants. Collectively, these data indicate two important aspects of aging and telomere length regulation. One, life span of nematodes can be manipulated independently of telomere length. Two, telomere length settings do not influence life span in the nematode.
Telomere Length Remains Constant during Aging
Next, we took advantage of CF512 fer-15(b26);fem-1(hc17) temperature-sensitive sterile worms [24] to test whether telomere length changes during organismal aging of an adult worm population. CF512 larvae were plated at the restrictive temperatures (25 °C) to block progeny production, and adult worms were harvested at 0, 5, 8, 10 and 13 d post reaching adulthood (Figure 3A). DNA was extracted from these samples, and telomere length was monitored by hybridization. In this model system that is free of contamination from embryos, no significant changes in telomere length were observed during the natural aging process of the animals (Figure 3B), again suggesting that organismal aging is independent of telomere shortening.
Figure 3 Telomere Length Is Independent from Organismal Life Span and Thermo Tolerance
(A) CF512 temperature-sensitive sterile worms were grown at 25 °C, and life span was monitored. Samples were taken at indicated times.
(B) DNA from CF512 temperature-sensitive sterile worms was extracted at the indicated time points, and telomere length was monitored in mixed and synchronous populations (for details, see Materials and Methods). Telomere length is indicated in kb. The asterisk points out a variable band-like signal resulting from gel drying.
(C) Seven different worm strains and daf-2 and daf-16 RNAi suppressed worms (indicated in different colors) with genetically determined telomere length (Figure 1D) were subjected to heat treatment as described [31]. Survival was assessed at the indicated time points.
(D) Two worm strains with long telomeres (CC2, TR403) and N2 worms with short telomeres were subjected to daf-16 suppression by RNAi. Life spans were determined as described.
Worms usually exhibit an age-related failure to respond to heat shock, whereas long-lived worms are more resistant to thermal stress [25,26]. Furthermore, short telomere length in peripheral blood mononuclear cells has been correlated with high levels of stress in women [27]. We therefore tested whether there could be any associated stress resistance qualities that could be correlated with increased telomere length. Consequently, we studied thermal stress resistance in wild-type strains with varying telomere length settings, such as strain TR403, which has exceptionally long telomeres, as well as long-lived daf-2(e1370) mutant animals and short-lived daf-16(mu86) null mutant animals. No obvious correlation between thermo tolerance and telomere length could be found (Figure 3C). All the worm strains analyzed displayed similar survival rates at 35 °C, including the TR403 strain carrying 6-fold longer telomeres than the Bristol N2 laboratory strain. Only the daf-2(e1370) mutant worms, used as positive control, displayed significantly greater heat-shock resistance (Figure 3C, p < 0.0001); however, these animals do not have increased telomere length. In summary, these data indicate that long telomeres do not supply an advantage during times of thermal stress, and that thermo tolerance, supplied by longevity, does not correlate with telomere length. Two of these strains with long telomeres (CC2 and TR403), and the standard short telomere containing N2 worms were subsequently tested for their response to lower daf-16 expression, and found to respond with a significantly shorter life span, independent of their telomere length, again suggesting that genetically controlled telomere length is uncoupled from genetically controlled organismal life span (Figure 3D).
Collectively our data argue for an uncoupling of organismal aging and telomere length. We have shown that both telomere length as well as longevity seem genetically controlled, but at this point our experiments do not exclude the possibility of epigenetic differences involved in the control of telomere length. However, both long-lived and short-lived worms can have either short or long telomeres. On the other hand, long telomeres do not contribute to increased longevity or stress resistance, as demonstrated in the TR403 or CC2 strains.
Telomerase knockout mice have progressively short telomeres after generational passaging. However, age-related effects in these mice are difficult to assess due to the increased rate of telomere loss in highly proliferative cells, compared to terminally differentiated cells. Our results suggest that telomere length does not influence the aging process in terminally differentiated, post-mitotic cells of an entire organism, C. elegans. Therefore, while telomere length control is essential for maintenance of the germline and other highly proliferative cells, telomere length control is dispensible for aging of non-dividing cells. We believe that other genetic pathways, such as the insulin/IGF-1 pathway, are required for the proper aging of post-mitotic cells.
Recently, it has been reported that elongation of telomeres by overexpression of hrp-1, a putative RNA binding protein that has been suggested to also associate with single stranded telomeric DNA, increased life span in a daf-16 dependent manner. Furthermore, overexpression of hrp-1 resulted in worms with longer telomeres and increased resistance to heat stress [12]. Here we show that telomere length can be uncoupled from organismal life span, is not dependent upon the insulin/IGF-1 signaling pathway, and is independent of resistance to thermal stress, suggesting that the effects that hrp-1 overexpression has on nematode longevity are not due to telomere elongation. Therefore, hrp-1 may be bifunctional, having one role in organismal aging and a separate, independent role in telomere length regulation.
By now, it is well established that telomere length and telomere shortening rates during mitotic cell division limit the replicative life span of mammalian cells. However, fibroblasts that enter a terminally differentiated state due to critically short telomeres usually do not die, but remain in a metabolically active state. In light of the experiments presented here, we suggest a model where telomere length and life span in post-mitotic organisms and cells are independent of each other.
Materials and Methods
Strains.
Worms were grown at 20 °C and maintained as described [18]. The strains used in this study include the following: Wild-type strains N2, CB3191, CB3192, CB4852, CB4856, CC2, TR403 (obtained from the Caenorhabditis Genetics Center, Minneapolis, Minnesota, United States) and the daf-16(mu86)I and daf-2(e1370)III mutant strains [22], as well as the temperature-sensitive sterile CF512 fer-15(b26)II; fem-1(hc17)IV strain [24].
Telomere length analysis.
For genomic DNA extraction, nematodes were floated off the bacterial lawns with M9 buffer, centrifuged, and washed again twice with M9 buffer. After harvesting, worms were lysed for 1 h at 50 °C in buffer containing 100 mM Tris HCl (pH 8.5), 100 mM NaCl, 50 mM EDTA, 1% SDS, 100 μg/ml proteinase K, and 1% β-mercaptoethanol. DNA was isolated by phenol-chloroform extraction in Phase Lock Gel tubes (Eppendorf, Hamburg, Germany) and isopropanol precipitation. DNA was then treated with 10 μg/ml RNAse A for 2 h. After that, the proteinase K incubation and extraction were repeated once. DNA was dissolved in 10 mM Tris HCl (pH 7.5) and 0.1 mM EDTA and stored at 4 °C. Telomere blots were carried out as described [28]. (TTAGGC)4 and (GCCTAA)4 end-labeled oligonucleotides were used as probes. Blots were exposed to PhosphorImager screens, and mean telomeric restriction fragment lengths were determined using ImageQuant software (both from GE Healthcare, Little Chalfont, United Kingdom).
Bal 31 treatments.
C. elegans DNA (50–100μg) was treated with 10–20 U of Bal 31 exonuclease (New England Biolabs, Beverly, Massachusetts, United States) at 30 °C in 500 μl of 1× Bal31 buffer. Reactions were stopped with the addition of EGTA (25 mM final concentration) at 0, 0.5, and 2 h. Samples were phenol extracted, and the DNA was collected by isopropanol precipitation. DNA from each time point was digested with AluI and MboI, quantified using Hoechst fluorimetry, fractionated on 0.7% agarose gels, and blotted as described above.
RNAi.
HT115 bacteria expressing dsRNA were grown at 37 °C in LB with 10 μg/ml tetracycline and 50 μg/ml carbenicillin, then were seeded onto NG-carbenicillin plates and supplemented with 100 μl 0.1 M IPTG (1.5 mM final concentration). Eggs were added to plates, and animals were transferred to new plates every 3 d. Vectors used in this study were: pAD12 (control vector Bluescript with opposing T7 promoters), pAD48 (2.3 kb of daf-2 coding sequence cloned into pAD12), and pAD43 (1.1 kb of daf-16 coding sequence cloned into pAD12).
Survival analysis.
Life span assays were performed at 20 °C and were initiated at the L1 larvae stage [21]. Animals (60–120 per experiment) were transferred away from their progeny to new plates every other day until the end of the reproductive period. We used Statview 4.5 (SAS, Cary, North Carolina, United States) software to carry out statistical analysis and to determine mean life spans and 75th percentiles. The logrank (Mantel-Cox) statistics were used to evaluate the hypothesis that animals in experimental and control groups behaved similarly. Animals that crawled off the plate, “exploded” (i.e., had a gonad extruding through their vulva), or “bagged” (i.e., died from internal hatching) were censored at the time of the event and were incorporated into the dataset as described [29].
Survival analysis of temperature-sensitive sterile animals.
Age-synchronized worms were obtained at 25 °C using the CF512 fer-15(b26)II;fem-1(hc17)IV temperature-sensitive sterile strain [24]. Embryos were isolated using an alkaline hypochlorite solution [30] from CF512 worms grown at 15 °C. Synchronized L1 larvae were isolated by gently shaking the embryos overnight at room temperature in M9 buffer. L1 larvae were placed on fresh OP50 plates and incubated at 25 °C until the end of the experiment. Adult worms were harvested at different time points, and telomere length was analyzed as described. At each time point, all worms were transferred to fresh OP50 plates. In parallel, a small fraction of the synchronized L1 larvae was placed on fresh OP50 plates and incubated at 20 °C to produce a mixed population of worms that was maintained throughout the experiment and used as control. Survival analyses of CF512 were performed at 25 °C as described.
Thermo tolerance assays.
All the worms were maintained under standard C. elegans growth condition at 20 °C. Response to heat shock was assayed on age-synchronized groups, produced by placing reproductive adults onto fresh agar plates and permitting the eggs laid to develop for 4 d into adults [31]. Survival time at elevated temperatures was determined by shifting synchronized d 4 adults (10–15 per plate) to 35 °C. The first time point was taken approximately 6 h after the shift, with subsequent time points taken every 3 h.
Supporting Information
Figure S1 Organismal Life Span Is Independent of Telomere Length
(A) Life span of control N2 worms (green), daf-2(e1370) mutants (blue), and daf-16(mu86) mutants (red). Life span was assessed as described [21].
(B) Telomere length of nine individual control N2 clones was assessed as described. Samples were collected at the beginning (B) and end (E) of the experiment, and fragment size is indicated in kb.
(C) Telomere length of nine individual daf-2(e1370) mutant clones was assessed as described in (B).
(D) Telomere length of nine individual daf-16(mu86) mutant clones was assessed as described in (B).
(6.4 MB TIF)
Click here for additional data file.
Table S1 Survival, Average and Maximal Telomere Length of Control daf-2(e1370) Mutants and daf-16(mu86) Mutant Clones
(52 KB DOC)
Click here for additional data file.
We thank members of the Karlseder and Dillin labs, V. Lundblad, and V. Bres for comments on the manuscript. This work was supported by the Mathers Foundation (JK), the Ellison Medical Foundation (AD), and the National Institutes of Health (RO1 GM069525 to JK) and (RO1 AG024365 to AD).
Competing interests. The authors have declared that no competing interests exist.
Author contributions. MR, AD, and JK conceived and designed the experiments. MR and HM performed the experiments. MR, HM, AD, and JK analyzed the data. AD contributed reagents/materials/analysis tools. AD and JK wrote the paper.
A previous version of this article appeared as an Early Online Release on August 1, 2005 (DOI: 10.1371/journal.pgen.0010030.eor).
Abbreviation
kbkilobasepair
==== Refs
References
de Lange T 2002 Protection of mammalian telomeres Oncogene 21 532 540 11850778
Shay JW Wright WE 2000 Hayflick, his limit, and cellular ageing Nat Rev Mol Cell Biol 1 72 76 11413492
Bodnar AG Ouellette M Frolkis M Holt SE Chiu CP 1998 Extension of life-span by introduction of telomerase into normal human cells Science 279 349 352 9454332
Blasco MA Lee HW Hande MP Samper E Lansdorp PM 1997 Telomere shortening and tumor formation by mouse cells lacking telomerase RNA Cell 91 25 34 9335332
Rudolph KL Chang S Lee HW Blasco M Gottlieb GJ 1999 Longevity, stress response, and cancer in aging telomerase-deficient mice Cell 96 701 712 10089885
Hemann MT Greider CW 2000 Wild-derived inbred mouse strains have short telomeres Nucleic Acids Res 28 4474 4478 11071935
Flemming AJ Shen ZZ Cunha A Emmons SW Leroi AM 2000 Somatic polyploidization and cellular proliferation drive body size evolution in nematodes Proc Natl Acad Sci U S A 97 5285 5290 10805788
Wicky C Villeneuve AM Lauper N Codourey L Tobler H 1996 Telomeric repeats (TTAGGC)n are sufficient for chromosome capping function in Caenorhabditis elegans
Proc Natl Acad Sci U S A 93 8983 8988 8799140
Kim SH Hwang SB Chung IK Lee J 2003 Sequence-specific binding to telomeric DNA by CEH-37, a homeodomain protein in the nematode Caenorhabditis elegans
J Biol Chem 278 28038 28044 12711598
Im SH Lee J 2003 Identification of HMG-5 as a double-stranded telomeric DNA-binding protein in the nematode Caenorhabditis elegans
FEBS Lett 554 455 461 14623111
Yi SY Joeng KS Kweon JU Cho JW Chung IK 2001 A single-stranded telomere binding protein in the nematode Caenorhabditis elegans
FEBS Lett 505 301 306 11566194
Joeng KS Song EJ Lee KJ Lee J 2004 Long lifespan in worms with long telomeric DNA Nat Genet 36 607 611 15122256
Ahmed S Hodgkin J 2000 MRT-2 checkpoint protein is required for germline immortality and telomere replication in C. elegans
Nature 403 159 164 10646593
Ahmed S Alpi A Hengartner MO Gartner A 2001
C. elegans RAD-5/CLK-2 defines a new DNA damage checkpoint protein Curr Biol 11 1934 1944 11747819
Benard C McCright B Zhang Y Felkai S Lakowski B 2001 The C. elegans maternal-effect gene clk-2 is essential for embryonic development, encodes a protein homologous to yeast Tel2p and affects telomere length Development 128 4045 4055 11641227
Lim CS Mian IS Dernburg AF Campisi J 2001
C. elegans clk-2, a gene that limits life span, encodes a telomere length regulator similar to yeast telomere binding protein Tel2p Curr Biol 11 1706 1710 11696330
Lakowski B Hekimi S 1996 Determination of life-span in Caenorhabditis elegans by four clock genes Science 272 1010 1013 8638122
Brenner S 1974 The genetics of Caenorhabditis elegans
Genetics 77 71 94 4366476
van Steensel B Smogorzewska A de Lange T 1998 TRF2 protects human telomeres from end-to-end fusions Cell 92 401 413 9476899
Cheung I Schertzer M Baross A Rose AM Lansdorp PM 2004 Strain-specific telomere length revealed by single telomere length analysis in Caenorhabditis elegans
Nucleic Acids Res 32 3383 3391 15247331
Apfeld J Kenyon C 1998 Cell nonautonomy of C. elegans daf-2 function in the regulation of diapause and life span Cell 95 199 210 9790527
Kenyon C Chang J Gensch E Rudner A Tabtiang R 1993 A C. elegans mutant that lives twice as long as wild type Nature 366 461 464 8247153
Dillin A Hsu AL Arantes-Oliveira N Lehrer-Graiwer J Hsin H 2002 Rates of behavior and aging specified by mitochondrial function during development Science 298 2398 2401 12471266
Garigan D Hsu AL Fraser AG Kamath RS Ahringer J 2002 Genetic analysis of tissue aging in Caenorhabditis elegans : A role for heat-shock factor and bacterial proliferation Genetics 161 1101 1112 12136014
Lithgow GJ White TM Hinerfeld DA Johnson TE 1994 Thermotolerance of a long-lived mutant of Caenorhabditis elegans
J Gerontol 49 B270 276 7963273
Lithgow GJ White TM Melov S Johnson TE 1995 Thermotolerance and extended life-span conferred by single-gene mutations and induced by thermal stress Proc Natl Acad Sci U S A 92 7540 7544 7638227
Epel ES Blackburn EH Lin J Dhabhar FS Adler NE 2004 From the cover: Accelerated telomere shortening in response to life stress Proc Natl Acad Sci U S A 101 17312 17315 15574496
Karlseder J Smogorzewska A de Lange T 2002 Senescence induced by altered telomere state, not telomere loss Science 295 2446 2449 11923537
Lawless JF 1982 Models and methods for lifetime data New York Wiley & Sons 592 p.
Hope I 1999 C. elegans : A practical approach Oxford Oxford University Press 281 p.
Hsu AL Murphy CT Kenyon C 2003 Regulation of aging and age-related disease by DAF-16 and heat-shock factor Science 300 1142 1145 12750521
|
16151516
|
PMC1200426
|
CC BY
|
2021-01-05 07:59:36
|
no
|
PLoS Genet. 2005 Sep 2; 1(3):e30
|
utf-8
|
PLoS Genet
| 2,005 |
10.1371/journal.pgen.0010030
|
oa_comm
|
==== Front
PLoS GenetPLoS GenetpgenplgeplosgenPLoS Genetics1553-73901553-7404Public Library of Science San Francisco, USA 1615151710.1371/journal.pgen.001003205-PLGE-RA-0109R2plge-01-03-04Research ArticleEpidemiology - Public HealthStatisticsGenetics/Population GeneticsGenetics/Complex TraitsCryptic RelatednessGenomic ControlAssociation StudiesComplex DiseaseConfounding from Cryptic Relatedness in Case-Control Association Studies Confounding from Cryptic RelatednessVoight Benjamin F *Pritchard Jonathan K Department of Human Genetics, University of Chicago, Chicago, Illinois, United States of AmericaAbecasis Goncalo EditorUniversity of Michigan, United States of America*To whom correspondence should be addressed. E-mail: [email protected] 2005 2 9 2005 1 3 e3218 5 2005 2 8 2005 Copyright: © 2005 Voight and Pritchard.2005This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.Case-control association studies are widely used in the search for genetic variants that contribute to human diseases. It has long been known that such studies may suffer from high rates of false positives if there is unrecognized population structure. It is perhaps less widely appreciated that so-called “cryptic relatedness” (i.e., kinship among the cases or controls that is not known to the investigator) might also potentially inflate the false positive rate. Until now there has been little work to assess how serious this problem is likely to be in practice. In this paper, we develop a formal model of cryptic relatedness, and study its impact on association studies. We provide simple expressions that predict the extent of confounding due to cryptic relatedness. Surprisingly, these expressions are functions of directly observable parameters. Our analytical results show that, for well-designed studies in outbred populations, the degree of confounding due to cryptic relatedness will usually be negligible. However, in contrast, studies where there is a sampling bias toward collecting relatives may indeed suffer from excessive rates of false positives. Furthermore, cryptic relatedness may be a serious concern in founder populations that have grown rapidly and recently from a small size. As an example, we analyze the impact of excess relatedness among cases for six phenotypes measured in the Hutterite population.
Synopsis
There has long been concern in the human genetics community that case-control association studies may be subject to high rates of false positives if there is unrecognized population structure. After being considered rather suspect in the 1990s for this reason, case-control studies are regaining popularity, and will no doubt be used widely in future genome-wide association studies.
Therefore, it is important to fully understand the types of factors that can lead to excess rates of false positives in case-control studies. Virtually all of the previous discussion in the literature of excess false positives (confounding) in case-control studies has focused on the role of population structure. Yet a widely cited 1999 paper by Devlin and Roeder (that introduced the genomic control concept) argued that, in fact, “cryptic relatedness” (referring to the idea that some members of a case-control sample might actually be close relatives, unbeknownst to the investigator) is likely to be a far more important confounder than population structure. Moreover, one of the two main types of statistical approaches for dealing with confounding in case-control studies (i.e., structured association methods) does not correct for cryptic relatedness.
This work provides the first careful model of cryptic relatedness, and outlines exactly when cryptic relatedness is and is not likely to be a problem. The authors provide simple expressions that predict the extent of confounding due to cryptic relatedness. Surprisingly, these expressions are functions of directly observable parameters. The analytical results show that, for well-designed studies in outbred populations, the degree of confounding due to cryptic relatedness will usually be negligible. However, in contrast, studies where there is a sampling bias toward collecting relatives may indeed suffer from excessive rates of false positives.
Citation:Voight BF, Pritchard JK (2005) Confounding from cryptic relatedness in case-control association studies. PLoS Genet 1(3): e32.
==== Body
Introduction
Case-control association studies are a popular, convenient, and potentially powerful strategy for identifying genes of small effect that contribute to complex traits [1]. However, case-control studies may be susceptible to high rates of false positives if the underlying statistical assumptions are not satisfied. In particular, it has long been a source of concern that population structure might cause confounding in such studies [2,3], and a number of statistical methods have been developed to detect and correct for unrecognized population structure [4–9].
However, in their 1999 paper, Devlin and Roeder argued that another source of confounding, “cryptic relatedness,” might actually be a more serious source of error for case-control studies. Cryptic relatedness refers to the idea that some members of a case-control sample might actually be close relatives, in which case their genotypes are not independent draws from the population frequencies. When that happens, the allele frequency estimates in the case and control samples are unbiased but may have greater variance than expected, and tests of association that ignore the excess relatedness have inflated type-1 error rates. Devlin and Roeder [4] pointed out that if one is doing a genetic association study, then one surely believes that the disease has an underlying genetic basis that is at least partially shared among affected individuals. If the cases share a set of genetic risk factors then, presumably, this means that the cases will be somewhat more closely related to each other, on average, than they are to control individuals. Devlin and Roeder then presented some numerical examples that suggested that cryptic relatedness may be an important effect in practice. However, it is difficult to know how realistic those examples are because they were constructed artificially, and were not based on a population genetic model.
At this time, there are few empirical data that bear on whether cryptic relatedness is a serious problem in practice. One study of association mapping in a founder population concluded that in that population, cryptic relatedness did have a significant impact on tests of association [10]. Methods exist that can incorporate kinship relationships into the test for association if such information is known [11–14]. If relationships are not known in advance, then genomic control methods can correct for cryptic relatedness [4,6,8], while structured association methods (developed for the population structure problem) cannot [7,9].
In this article, we aim to address the question of whether, and when, cryptic relatedness is likely to be a serious issue for case-control association studies. Our approach is to develop a formal model of cryptic relatedness within a population framework. We show that a natural measure of the impact of cryptic relatedness, that we will denote δ, depends on the population size, the genetic model parameterized by the recurrence risk ratio [15], and the number of sampled cases and controls. Our initial model assumes that studies are “well designed” in the sense that they do not have serious sampling biases, such as a bias toward enrolling related cases into a study. For that model, our results indicate that for association studies in large outbred populations, the confounding effect due to cryptic relatedness is expected to be negligible, but that it may well be a more serious issue in small, growing populations. We also consider two simple scenarios in which the sampling is biased toward collecting relatives among the cases. Such sampling can lead to non-trivial inflation.
Results
A Model of Cryptic Relatedness
Consider a study in which m cases affected with a disease and m random controls are genotyped at a single bi-allelic locus with alleles B and b that are at frequencies p and 1 − p, respectively. We aim to model the impact of cryptic relatedness on a test of association at this locus, assuming that the locus is not in fact linked to any disease-associated genes. The starting point for our notation and modeling is taken, with some modification, from [4].
We suppose that cases and controls are sampled from a single population (i.e., without population structure) of finite size, with discrete generations, and that mating is independent of the phenotype of interest. All individuals are sampled from the current generation. Since the impact of cryptic relatedness is due to alleles that are identical by descent, it will be necessary to model the coalescence times of chromosomes. We will use T ∈ {1, 2, 3, …} to denote the random time at which a particular pair of chromosomes in the current generation coalesces. (That is, T is the number of generations before the present at which the copies of the marker locus on each of the two chromosomes in question trace their ancestry back to a single ancestral chromosome.) According to standard models, for randomly chosen chromosomes (i.e., unconditional on phenotype)
, where Nx is the number of diploid individuals in generation x [16].
We will also assume that affected individuals have the same distribution of family sizes as do unaffected individuals, and that selection against the disease phenotype is negligible. Hence, chromosomes from affected individuals coalesce with chromosomes from random individuals at the same rate as do chromosomes from pairs of random individuals. To be precise, let T
(i,a)(i′,a′) denote the coalescence time between chromosomes a and a′ from individuals i and i′. (Here, a and a′ denote one of the two copies of each chromosome, chosen at random in individuals i and i′, respectively.) Then by assumption,
where Φi = aff and Φi = rand indicate that individuals i and i′ carry affected and random (unknown) phenotypes, respectively. In contrast, we will show that chromosomes from pairs of affected individuals have an excess probability of very recent coalescence. The extra relatedness of cases occurs because they share a heritable trait, and not from average differences in the family sizes of affected and unaffected individuals. Under the assumption in Equation 1, it follows that P[Φi = aff|T
(i,a)(i′,a′) = t,Φi = rand] = Kp, where Kp denotes the overall population prevalence of the disease of interest. This is reasonable, because simply knowing that individual i has a relative i′ whose affection status is unknown, should not alter the probability that i is affected.
We also define a quantity Kt that is analogous to the standard relative recurrence risk Kr [15]. Specifically, for a pair of individuals i and i′, where i is affected, Kt is defined as the probability that i′ is also affected, given that a specific pair of alleles from the two individuals coalesces to a common ancestral chromosome t generations before the present (where the alleles are at a locus unlinked to any disease loci): Kt = P[Φi′ = aff|Φi = aff, T
(i,a)(i′,a′) = t].
Notice, however, that the definition of Kt implies some ambiguity in the actual relationship between the two individuals in question: e.g., T can be 1 either for siblings or for half-siblings, and 2 for cousins or half-cousins. Therefore, to evaluate Kt, it will be necessary to be specific about mating patterns in the population. Later in the paper, we describe results for two particular models of random mating.
The ratio Kt/Kp will be denoted λt. This is closely related to the standard recurrence risk ratio λr [15], and measures the proportional increase in risk for an individual given that one of his/her chromosomes coalesces with the chromosomes of an affected individual t generations before the present. Due to shared genetic or environmental factors, λr (and hence λt) is often >>> 1 for close relatives; this means that even random sampling of affected individuals can lead to a sample that contains an excess of related cases.
Let
be an indicator variable for the presence (
) or absence (
) of the B allele on the ath copy of this locus in affected individual i. (Here, a ∈ {1, 2} labels the two homologous copies of a marker in a diploid individual.) Similarly,
denotes the analogous indicator variable for the ath copy in control individual j.
Then we define a test statistic, D, which measures the difference in the overall allele counts between case and control samples at a given marker:
When appropriately normalized, D forms the basis of familiar tests of association. Under the null hypothesis, D
2/Var[D] is χ
2 distributed with one degree of freedom [4]. D is proportional to both the trend test [17] and to the allele test [18].
Under the standard null hypothesis, an allele copy at a given marker is type B with probability p, independently for all allele copies in the sample. The independence assumption implies that there is no population structure, no inbreeding, and that all cases and controls are mutually unrelated. If all alleles are mutually independent, then the variance of D is 4mp(1 − p). If, however, cryptic relatedness exists in the sample, then the actual variance of the test—call this Var
*[D]—will exceed the variance predicted under the null hypothesis. We will measure the deviation from the null variance using the “inflation factor” δ, defined as follows:
In the absence of true association between the marker and the genotype, the commonly used test of association, D
2/[4mp(1 − p)], has a distribution that is the product of δ and a χ
2 random variable [4].
Values of the inflation factor, δ, near 1.0 imply that the standard test of association is correctly calibrated, or nearly so. Values of δ substantially larger than 1.0 indicate that there will be an excess of false positive signals. Our target here is to derive an expected value for δ under a model of cryptic relatedness. These general results do not rely on a particular genetic model, but we do present examples using an additive model. We consider models of constant population size and of recent population expansion.
Theory
We now characterize the extra variance that is caused by relatedness within a given case-control study, and use this to compute the expected inflation factor δ. Starting from the definition of D, in Equation 2, we can write Var
*[D] as
where i ≠ i′, j ≠ j′. We now need to determine how the value of this expression depends on cryptic relatedness.
Since Gi and Hj are Bernoulli trials, we have:
The following two terms in Equation 4 account for the possibility of departures from Hardy-Weinberg equilibrium in the sample. Assuming that these factors are independent of case-control status, we can write these as
where F measures the extent of the departure from Hardy-Weinberg equilibrium [4,19]. If, in fact, there is a different average level of inbreeding in cases than in controls [20], then we would replace F in Equation 7 and thereafter, with an average F across the cases and controls. (Notice that, unlike here, the inflation factor used by Devlin and Roeder was defined relative to the trend test, so that Hardy-Weinberg departures cancel out in their formulation.)
In our model, the controls are sampled randomly from the population. This means that the terms
and
are zero. This follows because, conditional on p, the fact that a random allele in the population is B, or b, provides no additional information about the genotype of another case or control in the sample. The assumption that controls are sampled randomly will usually be a good approximation, even if controls are specifically ascertained as not having the disease. As we will show below, the size of these covariance terms depends on the recurrence risk ratio for the phenotype, and the recurrence risk ratio for being unaffected is typically near one.
Next, since case alleles Gi are each similarly distributed, we can reduce Equation 4 by characterizing a single covariance between case alleles and then collecting the sum of all covariance terms that contain only case alleles. Given this, the Hardy-Weinberg equilibrium terms, and Equation 5, Equation 4 simplifies to:
where i ≠ i′. And now, finally, we need to evaluate
under a model of cryptic relatedness. In order to do this, we first need to evaluate the probability that alleles in affected individuals share a common ancestor in generation t before the present. This will allow us to calculate the extra relatedness in cases due to the phenotype.
Recall that Kp is the population prevalence of the disease; Kt is the probability that a relative of an affected individual is also affected, given that the two individuals share a common ancestor t generations before the present; and that λt = Kt/Kp is the corresponding ratio of risks [15]. Next, let T
(i,a)(i′,a′) denote the coalescent time of allele copies a and a′ from individuals i and i′. In a slight abuse of notation, we will abbreviate T
(i,a)(i′,a′) as Tii′. In what follows, individuals i and i′ are random (unphenotyped) draws from the population, except when specifically noted (e.g., Φi = aff indicates that i is affected). Then, using Bayes' rule, we can compute the coalescence rates for two chromosomes sampled from affected cases in the population as follows:
where P[Tii′ = t] denotes the prior probability of coalescence in generation t, for random (unphenotyped) individuals. Next, using the assumption that affected and unaffected individuals coalesce with random chromosomes at the same rate (Equation 1), it follows that P[Φi = aff|T
ii′ = t] = Kp, and hence
Equation 9 produces a pleasingly simple result: the coalescence rate for chromosomes from affected individuals is increased by a factor that is closely related to the standard recurrence risk ratio.
Recurrence Risk for Relatives
The recurrence risk ratio is an important quantity in genetic epidemiology, and is widely measured [1]. For siblings, typical recurrence risk ratios for complex diseases range from around 2 to 50. For more distant relationships, the risk ratio declines approximately geometrically toward 1 as the number of meioses separating two relatives increases.
In our theoretical development, we will assume that disease inheritance is governed by a single additive gene [15], unlinked to the marker locus of interest. Other genetic models, including more complex models, behave similarly to this, except that the rate of decay of λt with increasing t may differ somewhat [15], leading to different coefficients in the cryptic relatedness term in Equation 16 below.
For the additive model, [15] obtained an expression for the recurrence risk ratio, λr, for any possible relationship, r, in terms of the recurrence risk ratio for full siblings, λs:
where Φr is the kinship coefficient between rth-degree relatives. For example, Φr = 1/4 between sibs, and decays by 1/2 for each increment to r. To connect λr to our model—which is written in terms of coalescent time t instead of r—we need to be more explicit about the mating patterns in the population model.
For example, under the standard Wright-Fisher model where individuals select their parents independently at random, most relatives are “half-relations”: half-siblings, half-first cousins, half-second cousins, etc. In that case, for t = 1, 2, 3, … , the corresponding kinship coefficients are Φr = 1/8, 1/32, 1/128, and so on. Then for example, for t = 2, λt − 1 = 4(λs − 1)/32. If instead, mating is purely monogamous, but partners are still chosen at random, then all relationships are “full”: full siblings, full cousins, etc. That is, for t = 1, 2, 3, … , the corresponding kinship coefficients are Φr = 1/4, 1/16, 1/64,… .
In summary, λt may be much larger than 1 for the closest relatives, but it becomes approximately 1 if the common ancestor is more than just a few generations ago (> 10 or 15, say). This qualitative conclusion does not depend strongly on the assumed genetic model. Referring to Equation 9, this means that chromosomes from affected individuals have an excess probability of coalescing extremely rapidly (within the past few generations). If they do not, then they behave essentially like random chromosomes, for which coalescence takes place on timescales of thousands of generations in typical populations (Figure 1).
Figure 1 Coalescence Rates for Pairs of Random Chromosomes (Red) and for Pairs of Chromosomes from Affected Individuals (Green)
Notice that chromosomes from affected individuals have a small excess probability of coalescing very rapidly (i.e., in the most recent ten generations or so). Otherwise, their coalescence rates are essentially like those of random chromosomes. The region at the left-hand side of the graph between the red and green lines represents the excess probability of very recent coalescence among case chromosomes (denoted R in the text). This is what gives rise to the effect of cryptic relatedness. For larger t, the line for cases drops slightly below the line for random individuals, since both distributions integrate to 1. These plots assume an additive genetic model, with λs = 60, the “half”-relationships mating model, and a population size of 2,000. The line for cases was generated under the approximation that the excess relatedness is completely limited to the first n = 10 generations. In this case, the maximum coalescent probability for case chromosomes is 0.00275, when t = 1; R ≈ 0.00334. As expected, the mean coalescence time is ≈ 4,000 generations for both distributions. Alterations in n yield similar results (unpublished data).
The dynamics of this process are reminiscent of structured coalescent models with many demes [21–23]. In those models, two chromosomes from the same deme either coalesce with each other very quickly or escape into the population at large, and coalesce on a much longer time scale. These two phases have been described by John Wakeley as the “scattering phase” and the “collecting phase,” respectively [24]. An extreme example of this type of process (with selfing) was illustrated by Rousset [25].
Calculating the Inflation Factor
As described above, ancestral chromosomes of affected individuals coalesce at an increased rate during the most recent few generations (Figure 1), and otherwise behave essentially like random chromosomes. We now provide a heuristic derivation of the inflation factor δ; later we show that our expression closely approximates the results obtained in simulations. For simplicity, we consider the following approximation.
Let R be the excess probability of very recent coalescence for affected chromosomes relative to random chromosomes. That is,
where n might be taken as 10 or 15, say. Then write:
To evaluate
, notice that there are two cases. With probability R, the two chromosomes coalesce very rapidly due to their shared phenotype. In that case, they share such a recent common ancestor that they are almost certainly identical by descent. In the second case, with probability 1 − R, the two chromosomes behave as random chromosomes, and their genotypes are independent Bernoulli draws from the population frequencies:
And finally, substituting Equations 11, 13, and 7 into Equation 3, we obtain
Equation 14 is worthy of discussion. When the simplest model of independence among sampled alleles holds, then δ = 1. The term containing F corresponds to Hardy-Weinberg departures, due to inbreeding for instance. The summation term corresponds to the effect of cryptic relatedness; the sum itself can be thought of as calculating the excess probability of identity by descent between chromosomes from affected individuals. Overall, the effect of cryptic relatedness increases linearly with the sample size m (for a given population size and λt).
Applications to Specific Models
In this section, we evaluate Equation 14 under a range of specific models, in order to determine when cryptic relatedness is likely to have a substantial impact on case-control studies. The models presented assume an additive genetic model, as described above. At first, we will assume that the population is of constant size N, so that the probability of coalescence in generation t, P[Tii′ = t], is (1 − 1/2N)t−1/(2N). After that, we turn to models with population growth. For simplicity, we set F = 0.
The Inflation Factor in Populations of Constant Size
Recall from Equation 10 that λt − 1 = 4Φr(λs − 1). Recall also, that when individuals select their parents independently at random, as in the standard Wright-Fisher model, that most relatives are “half-relations” (e.g., half-siblings, half-cousins, etc.), and then the kinship coefficients Φr are 1/8, 1/32, 1/128, … for t = 1, 2, 3, etc. Using δ
half to indicate this situation where individuals are related via “half-relationships,” it follows that
Noting that
for small t (provided that N is not small), and that ∑ 2−2t+1 converges quickly to 2/3; Equation 15 can be further approximated as
If instead, mating is purely monogamous, but partners are still chosen at random, then all relationships are “full”—e.g., full siblings, full cousins, etc., and the kinship coefficients are two-fold higher. The corresponding inflation factor, δ
full, is
indicating that the impact of cryptic relatedness is approximately doubled when there is fully monogamous pairing of parents, compared to when there is independent pairing of parents for each offspring.
Simulations
To check the accuracy of our analytical results, we generated population histories via Wright-Fisher simulation and estimated the inflation factor, δ, for a given disease and population genetic model, as described in the Materials and Methods section. Results are presented in Table 1, and compared to predicted results from Equation 16. The results show close agreement between the analytical prediction and the simulation results. In some cases, the analytical results slightly overestimate the inflation factor, probably due to the approximations used in relating Equation 9 to δ.
Table 1 Values of the Inflation Factor as a Function of Model Parameters, and a Comparison of the Simulated (δ̂mean) and Analytical (δA) Predictions, for Populations of Constant Size
While the choice of an additive model for the phenotype (i.e., a heterozygote has exactly one-half the penetrance for the phenotype as a homozygote for the risk allele does) is mathematically convenient, alternative modes of inheritance (including multilocus models, or models with dominance components) are certainly likely in practice. Such models will have the impact of changing the rate of decay of λt, and hence the coefficient of the cryptic relatedness term in Equations 16 or 17. While we do not present a complete exploration of such models, we have performed a modest number of additional simulations under non-additive models. We have found that those results are qualitatively similar to the results presented above (unpublished data).
Intrinsic Constraints on δ
Table 1 shows the predicted impact of cryptic relatedness for a range of possible disease parameters. The magnitude of the inflation factor is fairly small for all parameter combinations shown, with a maximum value of 1.07. To make this more concrete, an inflation factor of 1.07 implies a quite modest excess of false positives: for instance, a fraction 1.5 × 10−3 of tests would be significant at the p = 10−3 level. As another example, consider a genetic model based loosely on a study of autism [26], where λs = 75, and Kp of 0.0004. Assuming the full-sibling model of relatedness, a sample size of 1,000, and a population size of 2.5 million (i.e., the number required to find that many cases), δ is just 1.02.
These examples notwithstanding, however, Equations 16 and 17 seem to suggest that δ can be made arbitrarily large simply by increasing the sample size m. But in fact, the space of sensible models is actually rather constrained. Since m cannot exceed Kp times the population size, there is a practical limit on m for a given λs and population size. Because of this constraint, it is difficult to construct biologically plausible parameter combinations that result in substantial inflation factors for randomly mating populations of constant size.
To be more specific, let Ks be the rate of disease in full siblings of an affected proband, i.e., Ks = λs Kp. Furthermore, let f be the fraction of all affected individuals in the population that are included in the sample. Then, noting that f = m/NKp, Equation 17 can be rewritten as
Therefore, since f ≤ 1, for diseases where Ks is smaller than, say, about 1%, the inflation factor is negligible. The only way to get large values of δ is to have high values of Ks − Kp and nearly complete ascertainment of cases (high f). For instance, if Ks were 0.2 and λs were 4, then the inflation factor could be as large as 1.1, producing a small excess of false positives. But the latter calculation assumes complete sampling of affected individuals (f = 1), which would usually be difficult for a common disease.
In summary, in populations of constant size, the impact of cryptic relatedness is generally very small, unless (1) Ks is quite large—more than 0.2, say, and (2) f is near 1, meaning that there is nearly complete ascertainment of cases from the population. Hence, cryptic relatedness should not be a serious concern for most complex trait studies in stable populations, assuming random sampling of cases. As we will show in the next section, the situation is more serious for models with population growth.
The Inflation Factor with Changes in Population Size
We now consider a model that allows for changes in population size. Let Nt represent the population size at time t. Then, provided that the coalescent probability 1/2Nt is not especially large in any of the recent generations, and since λt − 1 decays as t increases, we can rewrite and simplify Equation 14 to
where again λt refers to the recurrence risk ratio for coalescence time t. Because (λt − 1) decays quickly toward zero, it is apparent that only changes in population size during the last few generations will impact δ. Moreover, for given values of m and λt, smaller population sizes in the past will produce higher inflation factors.
To check the accuracy of our results regarding demographic expansion, we modified the forward simulation procedure used above such that instead of a single N, we simulated exponential growth that began at time t
onset in the recent past starting at an initial population size NA. For each subsequent generation t, the population size was determined by the equation Nt
+1 = Nt · eα for a growth rate α such that the population size in the final generation is Nf. We performed at least 10,000 repetitions for each parameter combination, and the 95% standard error about the mean for each estimated δ was no greater than 0.01. In our analytic calculation, we assumed the “half” relationships model, as in Equation 15 and 16.
Results of the simulations, for a range of parameter values, are summarized in Table 2. Again, the theoretical prediction in Equation 19 is close to the simulated values. Under very recent growth models, δ̂ can be substantial (as much as 2.5 for the extreme growth scenario shown). Under more realistic models of population growth, the effect of cryptic relatedness is smaller, but still non-trivial. Based on these results, it seems clear that the magnitude of growth is an important factor for determining δ. In populations that have grown rapidly from small size in the past few generations, cryptic relatedness may indeed lead to high inflation factors. It should be noted that many of the models presented have extreme growth; hence, the higher levels of cryptic relatedness shown here are likely to exceed anything seen in practice in human populations.
Table 2 Values of the Inflation Factor in Very Recently Expanded Populations
Figure 2 Cumulative Probability of Coalescence within the Last n Meioses in the Hutterite Founder Population
Each line plots the estimated probability that two chromosomes drawn at random, from different individuals affected with a given phenotype, or from two random control individuals, descend from a single ancestral chromosome within the last n meioses. These estimates are based on the recorded Hutterite genealogy. The x-axis plots the average number of meioses along the two lineages back to the common ancestor. Notice that in the most recent generations, the case samples coalesce at higher rates than do random controls.
The qualitative difference between the equilibrium model and the population-growth model can be understood as follows. Consider two studies in which m affected individuals are sampled from each of two populations that have the same current size. If one population is of fixed size, while the other has grown rapidly from a smaller size, then the probability that two individuals are closely related is much higher in the growing population than in the equilibrium population. It follows from Equation 19 that this produces a higher inflation factor in the growing population than in the stable one.
Cryptic Relatedness with Biased Sampling
Thus far, we have considered models that assume “good” sampling design, in the sense that the sample of cases represents a random sample of the affected individuals in a population. We now consider the impact of sampling schemes that bias toward enrolling close relatives as cases in a study. For the previous models, we showed that with random ascertainment of cases, the inflation factor δ is maximized with complete ascertainment of cases from a population. The following models are instead motivated by the scenario in which a study enrolls only a small fraction of the affected individuals in a large population but, due to sampling biases, tends to recruit close relatives. Such situations might arise in practice if, for example, a patient at a clinic or in a study encouraged affected family members to visit the same clinic, or also to enroll in the study.
As an extreme, but simple example, consider first the situation in which the case sample consists of m(1 − σ) unrelated affected individuals, plus mσ/2 pairs of affected siblings (σ ∈ [0, 1]). The controls are all unrelated to anyone else. Assume furthermore that there is not inbreeding, so that F = 0 and the probability of recent identity-by-descent for chromosomes in siblings is 0.5. (For simplicity, we assume both in this and the next model that the sampling is from a sufficiently large population relative to m that we can approximately ignore the impact of cryptic relatedness apart from that induced by the biased sampling of siblings.) Then recall from Equation 14 that δ ≈ 1 + F + (m − 1)R where R is the (average) excess probability of recent coalescence, computed across all pairs of case chromosomes. In this model, a fraction σ/(m − 1) of the pairs of individuals are siblings. The probability that a randomly selected chromosome a in one sibling and a′ in the other sibling descend from the same parental chromosome is R = 1/4. Hence, for this model we obtain δ ≈ 1 + σ/4. At most, if the entire case sample is made up of sibling pairs, δ = 1.25. Any relatedness among the controls would further increase δ.
As a second simple example, suppose that a study recruits only a small fraction of affected individuals from a large population, but that recruits sometimes then encourage their siblings to enroll. Let the number of siblings of a recruited individual be Poisson with mean g, and let h be the probability that an affected sibling goes on to enroll in the study, independently for each affected sibling. Then the number of siblings of the initial recruit who enroll as patients in the study is Pois(ghKs). After some algebra, it follows that the expected fraction of pairs of case individuals in the sample who are siblings is γ(γ + 2)/[(m − 1)(γ + 1)], where γ = ghKs. Hence (again taking F = 0), we obtain
From these examples, it seems that biased sampling of cases can have a substantial effect on inflating the test statistics—though this is less dramatic perhaps than might have been expected. For example, suppose that index cases have an average of g = 2 siblings, that they refer affected siblings with probability h = 0.5, and that Ks = 0.4. Then the inflation factor δ ≈ 1.17.
Cryptic Relatedness in the Hutterites
We have used data collected from a founder population, the Schmiedeleut (S-leut) Hutterites of South Dakota, to illustrate the impact of cryptic relatedness on association studies for phenotypes measured in that population [27]. The S-leut Hutterite population consists of 13,000 members connected by a single, known, multigenerational pedigree that goes back to 64 founder individuals about 12–13 generations ago. Approximately 800 members of this population have been phenotyped for many traits and genotyped at a large number of microsatellite markers [27,28]. We considered six phenotypes: asthma, atopy, diabetes, hypertension, obesity (> 33% body fat for males, > 38% body fat for females), and stuttering (ever stuttered), all of which we treated as binary traits. We are grateful to C. Ober, who kindly allowed us access to these data.
It has previously been reported that naïve tests of association produce an excess of false positive signals in this population [10,14]. Our aim in this section is to further explore the impact of relatedness among cases in the context of the theory developed here. In particular, we set out to determine (1) whether we could detect excess relatedness among affected individuals, (2) the empirical level of confounding at random markers, and (3) whether we could predict the observed level of confounding based on the pedigree.
The fact that we have complete genealogical information for the Hutterites allows us to estimate the coalescence probabilities for pairs of alleles in any two individuals at any time since the founding of this population. These probabilities were estimated as described in the Materials and Methods section. The data do not provide information about coalescent events more than about 12 generations before the present, but the theory presented above suggests that the impact of cryptic relatedness is due to very recent coalescent events (and this is supported by our results, as follows).
The results of this analysis are presented in Figure 2. For all six phenotypes, there is an excess rate of coalescence within the pedigree, relative to random controls. Moreover, most of the increased probability of coalescence is due to rather close relatedness among cases (i.e., mainly for ≤ 4 meioses). This is consistent with the theoretical prediction that λt − 1 declines rapidly to zero.
We next used the genotype data to obtain an empirical estimate of δ for each phenotype, under the assumption that most random markers are not genuinely associated with disease loci. We considered 437 microsatellite markers typed in approximately 800 members of this population and estimated δ as described in the simulation methods above. The procedure for estimating δ in this data is described in the Materials and Methods section.
Table 3 summarizes the results from this analysis. For all six phenotypes, there is a non-trivial inflation to the test for association under the null hypothesis, in the range of about 1.2–1.3. This is consistent with the previous report by Newman et al. of an excess of positive signals at a set of microsatellite markers in this population [10]. An inflation factor of 1.2 implies a rejection rate that is ≈ 1.5-fold too high at the 5% level, and ≈ 2.7-fold too high at the 0.001 level. A δ of 1.3 implies a rejection rate that is ≈ 1.7-fold too high at the 0.05 level, and ≈ 3.8-fold too high at the 0.001 level. In a majority of cases, the predicted level of inflation matches empirical estimates, and the analytical result in all cases predicts a non-trivial inflation factor for each phenotype. For related subsets of phenotypes (asthma/atopy and obesity/hypertension/diabetes), the observed inflation factor appears similar. However, this is partly coincidental: δ depends on both the coalescent time and the sample size, which are different for each phenotype.
Table 3 Observed (δ̂obs) and Predicted (δA) Inflation Factors for Six Phenotypes Measured in the Hutterite Founder Population
Discussion
Should one be concerned about confounding from cryptic relatedness in association studies? To address this question, we have developed theory to predict the amount of cryptic relatedness expected in a random-mating population. Our results demonstrate that confounding effects of this kind are expected to be substantial only under rather special conditions. The bulk of the effect is due to the occurrence of quite close relationships among sampled individuals. Except in small populations, random pairs of affected individuals are unlikely to be closely related. Our results in Equation 14 show that for a given genetic model and population size, the impact of cryptic relatedness grows linearly with sample size. However, this obscures the fact that in practice, the maximum number of cases m that can be sampled from a given population size, N, is constrained by the population prevalence (Kp), and hence is inversely related to λr. That is to say, assuming constant population size, it is difficult to construct examples in which cryptic relatedness has an appreciable effect.
In contrast, studies of populations in which there has been rapid and recent population growth, and where the total study population is small, should indeed be concerned about cryptic relatedness. This scenario produces higher levels of relatedness than are possible for the same values of m and λr in stable populations. Studies in populations that meet these conditions—especially founder populations—should use pedigree-based methods or genomic control to minimize false positives due to cryptic relatedness [4,10,12].
Another situation in which cryptic relatedness may be important is when there is extensive inbreeding. A model in which individuals are likely to mate with relatives will increase δ relative to the models analyzed in this paper. When there is inbreeding, if two individuals share one recent common ancestor, they are likely to share other recent ancestors. That is, conditional on having a recent common ancestor, the expected kinship coefficient between two individuals would be higher than modeled in Equations 16 and 17. With modest inbreeding, this is likely to be a small effect, but the effect may be important in some populations with extensive inbreeding. Indeed, population structure may be viewed as a strong form of inbreeding, and that is often suspected to be a non-trivial source of confounding [29]. In contrast, sampling schemes that draw both cases and controls equally from just a segment of a population (e.g., from part of a city) should not induce particular problems. Even if there is extra covariance among sampled individuals, this should occur both within and between cases and controls equally, and thus cancel (Equation 4).
It should be noted that our results assume that the disease phenotype is selectively neutral (see discussion surrounding Equation 1). If, in fact, affected individuals or mutation carriers have fewer offspring than normal, then this will mean that affected individuals tend to have fewer close relatives than do random individuals. This effect would in many cases lower the probability of recent coalescence of case chromosomes, thus reducing the size of δ. This situation would reduce the level of cryptic relatedness relative to the models presented here. Conversely, a phenotype that increased fitness (perhaps in carriers of genes responding to selection only) might lead to increased δ.
Lastly, it should be noted that our primary model assumed a “good” epidemiological design in which individuals are ascertained randomly from the population. However, cryptic relatedness can also result from the non-random ascertainment of family members in a case-control study. For instance, affected family members might be more likely to seek treatment in the same clinic, or affected individuals might encourage their affected relatives to enroll in a study. These types of situations may be difficult to detect at the time of enrollment, but can have non-trivial consequences even in large outbred populations. We have shown that these situations indeed result in excess false positive rates. After data collection, we recommend the use of techniques for identifying cryptic relative pairs based on genetic data [30–33]. Genomic control [4] can then be helpful for identifying any residual inflation.
Materials and Methods
Simulations.
To check the accuracy of our initial analytical results, we generated population histories via Wright-Fisher simulation and estimated the inflation factor, δ. A population of size N was advanced forward in time 4N generations, with non-overlapping generations and random pairing of parents, independently for each offspring. For each simulation, 1,000 bi-allelic sites separated by a recombination fraction of 0.5 (i.e., freely recombining) were simulated with a mutation rate of θ = 4Nμ = 1. After 4N generations, a random site with the desired allele frequency was selected as the true disease locus, and affection status was assigned to all members of the population based on an additive genetic model. To shorten the computational time, we initiated the simulations such that a smaller population with proportionally higher mutation rate was advanced forward in time until a given point in the distant past, and then the population size and mutation rate were rescaled to the desired levels. Samples of m random controls and m affected cases were then drawn from the simulated population. Then, for each marker, apart from the disease locus, we constructed the 2 × 2 contingency table containing the allele counts for cases and controls, respectively; provided that the expected count for each cell in the table was at least five, we computed the standard Pearson's χ
2 test statistic. We then estimated the inflation factor δ using estimators based on both the mean and median values of the χ
2 statistics [4,6]. For each estimated δ, 95% standard errors about the mean were based on 10,000 replicate simulations.
Estimating coalescent probabilities in the Hutterites.
We estimated the coalescent probabilities for pairs of alleles in two individual Hutterites by the following. Starting from the affected individuals in the population, or from a matched random sample of individuals from the current population, we simulated the inheritance of a pair of randomly chosen chromosomes from different individuals, backward through time, from the present to the founders of the population. If the two chromosomes coalesced to a common ancestral chromosome within the pedigree, we counted the number of meioses back to that common ancestor, reporting the average number if the number of meioses was different on the two lineages. We repeated this procedure until we observed at least 500,000 coalescence events within the simulation. To estimate the mean inbreeding coefficient (F) in this sample, we used the same procedure as above except that we picked the two chromosomes from the same random individual, traced them backward in time, and determined how frequently those two chromosomes coalesced within the pedigree.
Calculating the inflation factor in the Hutterites.
For each marker, we constructed a 2 × k contingency table, where k was the number of alleles for this marker. Then, we pooled the smallest allele counts in the table with the second smallest allele counts until a 2 × 2 contingency table was formed. These artificial 2 × 2 tables should mimic the results that would be obtained using bi-allelic markers. The depth of the pedigree is short enough that mutation within the pedigree should have minimal impact on δ. For each phenotype, we selected a random sample of controls with data collected for the analyzed phenotype and then treated the remaining affected individuals in the sample as cases. The list of random controls was then truncated (randomly) so that the sample sizes were equal in the two groups. For this set of cases and controls, we estimated δ based on the mean of tests from these 437 markers. This procedure was performed 1,000 times.
To be more careful about the possibility that some loci might be genuinely associated with a phenotype or in various degrees of linkage, we repeated the analysis using approximately 40 microsatellite markers, unlinked either to one another or to candidate gene regions showing evidence of linkage. The resulting δ̂s based on the mean were almost identical for all phenotypes to the larger marker sample (unpublished data). Finally, we generated a semi-analytical result for the phenotype by plugging the coalescent probabilities estimated from the pedigree, along with estimated inbreeding coefficients, and the average number of cases selected across all replicates, into Equation 14.
We thank Carole Ober for providing the marker, phenotypic, and genealogical data used for the Hutterite data analysis and for comments on the manuscript; Rebecca Anderson and Natasha Phillips for additional assistance in organizing and interpreting the phenotype data; and Catherine Bourgain, Graham Coop, William Wen, Sebastian Zöllner, and the anonymous reviewers for helpful comments or discussion. This work was supported in part by the National Institutes of Health (HG002772) and a Hitchings-Elion award from Burroughs Wellcome Fund to JKP; BFV received support from the above grant to JKP as well as NIH DK55889 to Nancy J. Cox and from a Genetics Regulation Training Grant NIH/NIGMS NRSA 5 T32 GM07197.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. BFV and JKP both conceived of and designed the model, and wrote the paper. In addition, BFV also performed the simulations and analyzed the data.
A previous version of this article appeared as an Early Online Release on August 2, 2005 (DOI: 10.1371/journal.pgen.0010032.eor).
==== Refs
References
Risch NJ 2000 Searching for genetic determinants in the new millennium Nature 405 847 856 10866211
Knowler WC Williams RC Pettitt DJ Steinberg AG 1989 Gm3;5,13,14 and type 2 diabetes mellitus: An association in American Indians with genetic admixture Am J Hum Genet 43 520 526
Lander ES Schork NJ 1994 Genetic dissection of complex traits Science 265 2037 2048 8091226
Devlin B Roeder K 1999 Genomic control for association studies Biometrics 55 997 1004 11315092
Pritchard JK Rosenberg NA 1999 Use of unlinked genetic markers to detect population stratification in association studies Am J Hum Genet 65 220 228 10364535
Bacanu SA Devlin B Roeder K 2000 The power of genomic control Am J Hum Genet 66 1933 1944 10801388
Pritchard JK Stephens M Rosenberg NA Donnelly PJ 2000). Association mapping in structured populations Am J Hum Genet 67 170 181 10827107
Reich D Goldstein D (2001 Detecting association in a case-control study while correcting for population stratification Genet Epidemiol 20 4 16 11119293
Satten GA Flanders WD Yang Q 2001 Accounting for unmeasured population substructure in case-control studies of genetic association using a novel latent-class model Am J Hum Genet 68 466 477 11170894
Newman DL Abney M McPeek MS Ober C Cox NJ 2001 The importance of genealogy in determining genetic associations with complex traits Am J Hum Genet 69 1146 1148 11590549
Slager SL Schaid DJ 2001 Evaluation of candidate genes in case-control studies: A statistical method to account for related subjects Am J Hum Genet 68 1457 1462 11353403
Abney MA McPeek MS Ober C 2001 Narrow and broad heritabilities of quantitative traits in a founder population Am J Hum Genet 68 1302 1307 11309690
Abney MA Ober C McPeek MS 2002 Quantitative trait homozygosity and association mapping and empirical genomewide significance in large, complex pedigrees: Fasting serum-insulin level in the Hutterites Am J Hum Genet 70 920 934 11880950
Bourgain C Hoffjan S Nicolae R Newman D Steiner L 2003 Novel case-control test in a founder population identifies P-selectin as an atopy-susceptibility locus Am J Hum Genet 73 612 626 12929084
Risch NJ 1990 Linkage strategies for genetically complex traits. I. Multilocus models Am J Hum Genet 46 222 228 2301392
Hudson RR 1990 Oxford surveys in evolutionary biology Oxford Oxford University Press
Armitage P 1955 Test for linear trends in proportions and frequencies Biometrics 11 375 386
Sasieni PD 1997 From genotypes to genes: Doubling the sample size Biometrics 53 1253 1261 9423247
Gillespie JH 1998 Population genetics: A concise guide Baltimore Johns Hopkins University Press 174 p.
Rudan I Smolej-Narancic N Campbell H Carothers A Wright A 2003 Inbreeding and the genetic complexity of human hypertension Genetics 163 1011 1021 12663539
Hey J 1991 A multi-dimensional coalescent process applied to multi-allelic selection models and migration models Theor Popul Biol 39 30 48 2024230
Nei M Takahata N 1993 Effective population size, genetic diversity, and coalescence time in subdivided populations J Mol Evol 37 240 244 8230248
Nordborg M Donnelly P 1997 The coalescent process with selfing Genetics 146 1185 1195 9215919
Wakeley J 1999 Nonequilibrium migration in human history Genetics 153 1863 1871 10581291
Rousset F 2002 Inbreeding and relatedness coefficients: What do they measure? Heredity 88 371 380 11986874
Risch N Spiker D Lotspeich L Nouri N Hinds D 1999). A genomic screen of autism: Evidence for a multilocus etiology Am J Hum Genet 65 493 507 10417292
Abney MA McPeek MS Ober C (2000 Estimation of variance components of quantitative traits in inbred populations Am J Hum Genet 66 629 650 10677322
Ober C Tsalenko A Parry R Cox NJ 2000 A second-generation genomewide screen for asthma-susceptibility alleles in a founder population Am J Hum Genet 67 1154 1162 11022011
Thomas DC Witte JS 2002 Point: Population stratification: A problem for case-control studies of candidate-gene associations? Cancer Epidemiol Biomarkers Prev 11 513 520 12050091
Thompson E 1975 The estimation of pairwise relationships Ann Hum Genet 39 173 188 1052764
Lynch M Ritland K 1999 Estimation of pairwise relatedness with molecular markers Genetics 152 1753 1766 10430599
Ritland K 2000 Marker-inferred relatedness as a tool for detecting heritability in nature Mol Ecol 9 1195 1204 10972759
Milligan BG 2003 Maximum-likelihood estimation of relatedness Genetics 163 1153 1167 12663552
|
16151517
|
PMC1200427
|
CC BY
|
2021-01-05 08:00:24
|
no
|
PLoS Genet. 2005 Sep 2; 1(3):e32
|
utf-8
|
PLoS Genet
| 2,005 |
10.1371/journal.pgen.0010032
|
oa_comm
|
==== Front
Filaria JFilaria Journal1475-2883BioMed Central London 1475-2883-4-81608683810.1186/1475-2883-4-8ResearchRepeated high doses of avermectins cause prolonged sterilisation, but do not kill, Onchocerca ochengi adult worms in African cattle deC Bronsvoort Barend M [email protected] Alfons [email protected]é Virginia [email protected] Vincent N [email protected] David [email protected] Alexander J [email protected] Veterinary Parasitology, Division of Parasite and Vector Biology, Liverpool School of Tropical Medicine/Faculty of Veterinary, Science, University of Liverpool, Pembroke Place Liverpool,UK2 Fachgebiet Parasitologie, Universität Hohenheim, Germany3 Institut de Récherches Agricole pour le Développement, Wakwa, B.P. 65 Ngaoundere Cameroon4 Centre for Tropical veterinary Medicine, Royal (Dick) School of Veterinary Studies, University of Edinburgh, Easter Bush Veterinary Centre, Roslin, Midlothian, EH25 9RG, UK5 Universität Tübingen, Friedhofstrasse 73, 72074 Tübingen, Germany6 Institute of Agricultural Research for Development, Regional Centre of Bambui, B.P. 51 or 80, Bamenda, Cameroon2005 8 8 2005 4 8 8 10 1 2005 8 8 2005 Copyright © 2005 deC Bronsvoort et al; licensee BioMed Central Ltd.2005deC Bronsvoort et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Ivermectin (Mectizan™, Merck and CO. Inc.) is being widely used in the control of human onchocerciasis (Onchoverca volvulus) because of its potent effect on microfilariae. Human studies have suggested that, at the standard dose of 150 μg/kg an annual treatment schedule of ivermectin reversibly interferes with female worm fertility but is not macrofilaricidal. Because of the importance of determining whether ivermectin could be macrofilaricidal, the efficacy of high and prolonged doses of ivermectin and a related avermectin, doramectin, were investigated in cattle infected with O. ochengi.
Methods
Drugs with potential macrofilaricidal activity, were screened for the treatment of human onchocerciasis, using natural infections of O. ochengi in African cattle. Three groups of 3 cows were either treated at monthly intervals (7 treatments) with ivermectin (Ivomec®, Merck and Co. Inc.) at 500 μg/kg or doramectin (Dectamax®, Pfizer) at 500 μg/kg or not treated as controls. Intradermal nodules were removed at 6 monthly intervals and adult worms were examined for signs of drug activity.
Results
There was no significant decline in nodule diameter, the motility of male and female worms, nor in male and female viability as determined by the ability to reduce tetrazolium, compared with controls, at any time up to 24 months from the start of treatments (mpt). Embryogenesis, however, was abrogated by treatment, which was seen as an accumulation of dead and dying intra-uterine microfilariae (mf) persisting for up to 18 mpt. Skin mf densities in treated animals had fallen to zero by <3 mpt, but by 18 mpt small numbers of mf were found in the skin of some treated animals and a few female worms were starting to produce multi-cellular embryonic stages. Follow-up of the doramectin treated group at 36 mpt showed that mf densities had still only regained a small proportion of their pre-treatment levels.
Conclusion
These results have important implications for onchocerciasis control in the field. They suggest that ivermectin given at repeated high does may sterilise O. volvulus female worms for prolonged periods but is unlikely to kill them. This supports the view that control programmes may need to continue treatments with ivermectin for a period of decades and highlights the need to urgently identify new marcofiliaricidal compounds.
==== Body
Background
An estimated 17.5 million people in Africa and a further 140,000 in the Americas are infected by the filarial parasite Onchocerca volvulus, the cause of human river blindness [1]. The disease and its vector, Simulium damnosum s.l., have been the target of the largest successful and sustained vector-control programme the world has known, the Onchocerciasis Control Programme in West Africa (OCP) [2]. However in 1987 the veterinary endectocide, ivermectin, was registered as a safe and effective microfilaricide for human use against O. volvulus as an oral preparation, Mectizan™. The manufacturers, Merck and Co. Inc., via their Mectizan™ Donation Programme are committed to provide ivermectin free of charge to the port of entry of endemic countries 'for as long as is necessary' to aid the control of onchocerciasis. The African Programme for Onchocerciasis Control (APOC) [3] utilises ivermectin as its main tool to eliminate onchocerciasis as a public health problem. It aims to distribute ivermectin once a year through a sustainable, community-directed distribution system (CDTS).
The macrocyclic lactones are effective against a wide range of ecto- and endoparasites of animals and man. They are grouped into two major chemical families: the avermectins (including ivermectin and doramectin) and the milbemycins (including moxidectin). Ivermectin is derived from the actinomycete, Streptomyces avermitilis, and was originally developed as a veterinary endectocide against endo- and ectoparasites. Doramectin is a fermentation product from a mutant S. avermectilis strain. Its spectrum of activity is nearly the same as ivermectin [4] but its lipophilic cyclohexyl moiety at the 25-position produces a longer tissue half-life of 89 hours [5].
The efficacy of ivermectin against the skin dwelling larval stage of O. volvulus which causes blindness and skin lesions is well established [6-8]. However, its action against adult male and female worms is the subject of debate. Human studies with O. volvulus using a single dose at 150 μg/kg [9] produced no increased mortality of the adult worms and reversible affects on their fertility. Repeated 6-monthly [10-13] or 3-monthly treatments [14] produced conflicting results, but a proportion of female worms recovered their fertility 12–18 months after the last treatment and mf loads were 6–7% of pre-treatment loads 12 months after the last treatment, rising to between 9–32% at 18 months [15]. Epidemiological analysis and mathematical simulations of annual ivermectin treatments in man suggest that there is approximately a 30% reduction in mf production [16]. These authors, however, were not able to differentiate between the possible biological mechanisms such as a general loss of fecundity, or death of a proportion of the adult worms, or ivermectin facilitated immunity.
In this paper, we describe a controlled therapeutic trial in which the effects of high, repeated doses of ivermectin or doramectin were investigated against the African cattle parasite O. ochengi for up to 36 months after the first treatment. This species is phylogenetically closely related to O. volvulus for which it provides an excellent analogue [17] and was formerly adopted by the World Health Organisation (WHO) tertiary drug screening system for macrofilaricides until 2000. Previous studies have indicated both its practical merits and its predictive value [18-20].
Materials and methods
Between February and March 1994, nine 4–6 year old Gudali cows (Bos indicus), each with more than 20 palpable O. ochengi nodules in the ventral abdominal skin, were purchased from local markets in the Adamoua province of Cameroon. They were kept at pasture on the research station of the Institut de Récherches Agricole pour le Développement (IRAD) (formerly the Institit de Récherches Zootechniques et Vétérinaires), Wakwa, 10 km S.W. of Ngaoundere, at an altitude of 1000 m where the annual transmission potential (ATP, measured in L3s/person/year) for O. ochengi is low, around 600 L3's (Renz and Bronsvoort unpublished observations). Treatments began in October 1994, and nodulectomies and 3-monthly skin biopsies were continued for 24 mpt. In July of 1997 (36 mpt) the experiment was terminated.
Treatment regime
The cows were cast in lateral recumbency and the nodule distribution on each side of the ventral abdomen was mapped on a standardised recording sheet to allow tracking of individual nodules. The cows (mean weight = 379 kg) were ranked in order of their number of nodules and assigned to one of 3 groups, giving an even distribution of nodules between groups (Table 2). The cows were treated with either a subcutaneous inject of 500 μg/kg ivermectin, or a subcutaneous inject of 500 μg/kg doramectin or not treated to act as controls. Treatments were repeated at monthly intervals for 7 treatments.
Table 2 Nodule counts (including all removed nodules at 24 mpt), diameter (means ± standard deviations) and median age scores of female worms before and 24 months after the start of treatment in cattle, following 7 monthly treatments with ivermectin 500 μg/kg subcutaneously, doramectin 500 μg/kg subcutaneously or no treatment. Nodule numbers were adjusted for nodules removed for assay. (the number of animals in each group, n = 3 except ‡ n = 2; number of nodules per group, n = 6 except * where n = 4).
Timepoint 0 mpt 24 mpt
count1 Diameter2 (mm) ♀age score3 count Diameter (mm) ♀age score
Control 75 ± 26 5.9 ± 0.6 1.5 61 ± 25 ‡ 6.7 ± 0.7* 2*
Ivermectin 73 ± 25 7.4 ± 1.2 1 72 ± 19 6.5 ± 0.4 3
Doramectin 73 ± 35 6.9 ± 1.1 2 66 ± 36 6.9 ± 0.9 3
1 Two-way ANOVA: effect of interaction of treatment and mpt p = 0.630; effect of treatment p = 0.891; effect of mpt p = 0.510
2 Two-way ANOVA: effect of interaction of treatment and mpt p = 0.270;effect of treatment p = 0.403; effect of mpt p = 0.996
3 Not tested statistically
Sampling regime
At 3-monthly intervals 3 superficial skin biopsies (between 10–20 mg) per animal were taken from a shaved patch of skin on the ventral mid-line between the umbilicus and the udder. They were incubated in Roswell Park Institute Medium (RPMI) for 24 hours followed by collagenase at 37°C for 24 hours, using the protocol of Renz et al. [18]. The majority of mf escaped from the skin in the first 24 hours of incubation with the remaining mf found after digestion of the skin with collagenase. The geometric mean number of mf per 100 mg of skin from the three biopsies for each animal using the log+1 transformation. The mean per group was calculated in a similar way. All animals were sampled on the same morning.
O. ochengi nodules numbers were assessed by careful palpation of the skin of each cow. The distribution of nodules was recorded prior to treatment and about 20 nodules per cow were permanently marked with three dots around the nodule using green tattooing ink. This ensured identification of sites where nodules had been should they disappear completely following treatment. Nodule numbers were re-counted at 6 monthly intervals until 24 mpt (8/9 animals). After recording nodule distribution, four nodules per animal were removed under local anaesthetic and the wounds closed with a catgut suture and treated with Negasunt® (Bayer plc) powder to control fly strike. Two nodules per cow were preserved in 10% formal saline for histological examination. The remaining 2 nodules per cow were examined the same day by the triple assay described by Renz et al. [18], with the following slight modifications.
Triple assay of live female and male worms
In brief, the nodules were trimmed of any host tissue and measured along their longest axis to give the 'nodule diameter'. Each nodule was then cut open on a hollow slide in phosphate buffered saline and the worm contents carefully expressed. The male worms were teased free and placed individually in the wells of a 24 well plate with RPMI. The anterior 5–10 mm of each female worm was amputated and placed in a separate well and classed as either 1, young (transparent white, little incrustations on the cuticle); 2, mature (yellow-brownish, with incrustations); 3, old (dark brownish, strong incrustations); or 4, calcified (black and calcified).
After incubation (30 minutes at 37°C), the motility of male worms and anterior ends of female worms were scored after 30 seconds of observation on a 3 point scale (0 = no movement; 1 = weak, slow or intermittent movement; 2 = vigorous movement). Individual worm viability was then assessed by the MTT/formazan assay [18,21]. The optical density of the formazan reduction product was determined at 492 nm (OD492, 'Cambridge Life Instruments photo-spectrometer'). Mean OD492 readings for the total male worm or anterior ends of female worms were used until 18 mpt after which the mean OD492 / 10 mm of worm was calculated.
The remaining (posterior) part of each female worm was then washed into a mortar with 1.8 ml of PBS (final volume 2 ml), gently crushed with a pestle and a sample transferred to a Fuchs-Rosenthal chamber (3.2 μl) to determine an embryogram following the method of Schulz-Key [22]. A count was made of the normal and abnormal/pathologic embryonic stages and intra-uterine mf. Oocytes were also counted, but the repeatability of their counts was not reliable and they were not included in the final analysis.
Statistical Analyses
Statistical analyses were performed using 'Minitab'1 and BMDP2 computer packages. The nodule counts, nodule diameters and the MTT reduction/formazan formation results were analysed using a two-way repeat measures ANOVA between groups and timepoints. Worm age, motility scores and the percentage pathological stages at each timepoint were compared using a Kruskal-Wallis non-parametric one-way ANOVA. All statistical tests were interpreted at the 5% level of significance.
Results
Of the original nine cows at the start of the study, one control animal, no. 17, had to be slaughtered after the 12 mpt sampling on welfare grounds. Two more animals, one control and one ivermectin treated animal, were removed for different reasons after the 24 mpt sampling and regular observations were terminated at this point. However, where possible skin mf loads were monitored and nodules collected from the remaining six animals until 51 mpt.
Microfilariae
Following treatment, the mf density in the skin declined sharply and no skin mf were found at 3 mpt (Table 1). The control animals maintained substantial mf loads throughout the study, although their loads fluctuated considerably. Small numbers of mf began reappearing in 4/6 of the treated animals between 15 and 21 mpt.
Table 1 Geometric mean (3 skin snips)O. ochengi mf density+1 per 100 mg skin in cattle following 7 monthly treatments with ivermectin 500 μg/kg subcutaneously, doramectin 500 μg/kg subcutaneously or untreated controls. Animals lost to follow-up are indicated in the table.
ID 0 mpt 3 mpt 6 mpt 9 mpt 12 mpt 15 mpt 18 mpt 21 mpt 24 mpt 36 mpt
Control 17 3803 635 203 0 18 a
19 841 260 1515 774 1135 604 1698 185 257 c
22 4578 1583 911 712 767 291 982 975 1462 281
Ivermectin 14 509 0 0 0 0 1 3 0 8 2
18 671 0 0 0 0 0 2 0 0 b
20 240 0 0 0 0 0 0 0 0 0
Doramectin 12 6901 0 0 0 0 0 0 5 16 75
15 1607 0 0 0 0 0 0 0 0 0
23 146 0 0 0 0 0 3 2 0 0
a: Slaughtered following tetracycline therapy for dermatophilus infection, b: slaughtered due to septic arthritis, c lost to follow-up
Nodule counts, diameter and female age score
Table 2 gives the mean nodule counts, the mean nodule diameter and the median female age score (male age scores were not included) for the 3 groups before and 24 months after the start of treatment. There was no significant difference in nodule counts or nodule diameters between the groups or time points.
Motility, number of males and females per nodule and viability as measured by formazan formation
Male Onchocerca worms are mobile and although the number per female varied during the experiment, the ratio of males to females remained around one (with one female worm to each nodule). The male and female median motility scores are summarised in Table 3 and Figure 1. The median motility of male worms remained at 2 throughout the study except at 18 mpt for the ivermectin group. Similarly, for female anterior ends, their motility was 2 except at 24 mpt when this dropped. A Kruskal-Wallis ANOVA comparing the group medians for this timepoint was not significant (p = 0.20).
Table 3 The ratio of the number of male/female (♂/♀) worms examined for each group at each timepoint (one female worm was found per nodule) and the median male (♂MOT) and (♀MOT) motility score before and up to 24 months the after the start of treatment in cattle following 7 monthly treatments with ivermectin 500 μg/kg subcutaneously, doramectin 500 μg/kg subcutaneously or untreated controls.
mpt 0 6 12 18 24
Control ♂/♀ 5/6 5/5 8/5 4/4 3/4
♂MOT 2 2 2 2 2
♀MOT 2 2 2 2/6 2
Ivermectin ♂/♀ 3/6 5/6 6/6 0.3 3/6
♂MOT 2 2 2 1.5 2
♀MOT 2 2* 2 2* 1.5*
Doramectin ♂/♀ 11/6 5/6 9/6 6/6 9/6
♂MOT 2 2 2 2 2
♀MOT 2 2 2 2 0.5**
*= 1 calcified dead worm, **= 2 calcified dead worms
Figure 1 Mean (with standard deviation) OD492 of female (a) and male (b) O. ochengi formazan formation after MTT reduction before and up to 24 months after treatment (from 18 months after the start of treament OD492 is standardised for 10 mm of length). ■ = control, □ = ivermectin, = doramectin. (N.B. The mean OD492 reading for males was first calculated for each nodule and this was then used to calculate the group means given in Figure 1b) Foot notes. (a) Two-way ANOVA for effect of timepoint and treatment group: Interaction effect between group and timepoint p = 0.44; group effect p = 0.62; effect of timepoint p = 0.02. (b) Two-way ANOVA for effect of timepoint and treatment group: Interaction effect between group and timepoint p = 0.69; group effect p = 0.79; effect of timepoint p = 0.02.
From Figures 1a and 1b there is clearly no drug effect on worm viability/OD492 of either males or females. There is an effect of time with mean OD492 readings dropping after 18 mpt. This effect is associated with the change to a standardisation of the viability/OD492 per 10 mm of worm length.
Embryograms
The results of the embryograms are given in Figure 2. Normal embryonic stages (small morulae, large morulae, horseshoes, pretzels or coiled microfilariae) disappeared from the two treated groups by 6 mpt and failed to reappear in the female worms. Intrauterine mf accumulated and finally degenerated in the uteri, failing to be expelled for more than 24 months from the start of treatment. Beginning with approximately 10,000 normal intra-uterine mf stages pre-treatment, the number of pathological stages peaked around 30–40,000 at 6 to 12 mpt. The intra-uterine mf disintegrated steadily, first becoming vacuolated and non-motile, and eventually becoming ghost forms. This sterilizing effect persisted in 4 of the 6 treated animals for up to 24 months from the start of treatment. In two of the doramectin treated animals a nodule (i.e. 2/6 nodules), classed as old, was recovered at both the 18 and 24 mpt examinations, with large numbers of normal embryonic and intra-uterine mf amongst the pathological stages still in the uteri. In the ivermectin-treated animals, no fertile nodules were found, but the small number of mf in the skin in one animal (no. 14) indicates the presence of at least one fertile worm couple.
Figure 2 Mean number (× 1000) of embryonic stages 'o' (small and large morulae, horseshoe and pretzel stages) and intra-uterine mf of O. ochengi per nodule, over-laid with the number of pathological stages '■ 'calculated from the median percent of pathological forms before and up to 24 months following treatment with ivermectin or doramectin.
Discussion
Previous drug trials [18,23] have demonstrated the advantages of the O. ochengi screen in cattle in which individual animals are sequentially sampled and worm viability is determined.
As in previous studies in cattle [18], the skin mf density declined to zero following treatment with ivermectin and the same result was found with doramectin. Although mf began reappearing around 18 mpt only one animal, no. 12, showed a consistent increase in mf density and therefore the presence of at least one fertile worm couple. An interesting observation was the disappearance of mf from the skin of a control animal, no. 17. This was associated with repeated treatments of oxytetracyclines for a severe mycotic drematitis (Dermatophilus congolensis) skin infection. Treatment failed to control the skin infection and the animal was culled on welfare grounds after the 12-mpt sampling. The macrofilaricidal effect of oxytetracyclines was subsequently demonstrated [23].
In spite of being given at 2.5 times the recommended dose for cattle and at monthly intervals for 7 treatments, neither ivermectin nor doramectin reduced nodule diameter, male or female worm number or male motility compared with controls. Female motility appeared to be compromised in the doramectin treated group by 24 mpt, though this was not statistically significant. The viability/OD492 of female worms, however, did not appear to have been reduced significantly. Taken together these results support the conclusions of previous studies in humans taking invermectin once or twice annually that even using high and prolonged dosing, the avermectins are not macrofilaricidal against Onchocerca spp. [10,12,13]. The failure to kill the adult worm is, however, not due to a failure of the drug to penetrate the O. ochengi intra-dermal nodule [24,25].
The main effect of both ivermectin and doramectin on the adult female worm was on the development of the multicellular embryonic stages and the intra-uterine mf, as in man [10,12,13] where repeated treatments of ivermectin at 150 μg/kg every 6–12 months were reversible after 9–18 months though fecundity had been reduced. Interestingly, the number of intra-uterine mf rose sharply in both treated groups as dead and dying mf accumulated in the uteri of the worms. Over the same period the embryonic stages disappeared from the uteri. Certainly some of these must have continued to develop through to mf in order to explain these changes [26]. Why the degradation and disappearance of dead mf from the uteri is so protracted, is unclear, but it may be important in preventing fertilisation of the female [11,27], as this is the stimulation for a new cycle of oogenesis [28]. Alternatively, the uteri may be paralysed as happens to the pharyneal muscles of Haemonchus contortus [29].
Observations in man noted a disappearance of O. volvulus males from nodules following multiple treatments with ivermectin [14,30]. In addition, after treatment with other macrofilaricidal drugs, fewer males were recovered from O. ochengi nodules [18] and this is considered a useful, early indicator of macrofilaricidal activity. Our study, however, is in agreement with observations on O. volvulus in man [11,12] as we found male numbers to vary over time with no clear decline. Further, the male worms maintained their motility and viability (OD492) throughout the 24 months of the study. The effects of avermectins on male fertility were not examined here.
Our study has demonstrated that with higher doses of ivermectin or doramectin, repeated at monthly intervals, embryogenesis can be disrupted and possibly permanently impaired. Even 18 months after the last treatment none of the removed ivermectin treated nodules showed signs of recommencing embryogenesis. At 24 mpt, 2/6 nodules in the doramectin group did appear to regain some reproductive capability.
In previous studies of age-specific nodule counts [31] it has been suggested that nodules continue to be acquired throughout the life of the cow while their mf loads level off around 4–6 years of age. From the prophylactic studies in cattle [19,32] it is clear that ivermectin and moxidectin can prevent the development of L3 larvae in the host for at least 3 months after a single treatment. The corrals at Wakwa have a known low ATP of around 600 L3's per animal per year as opposed to 24,000 L3's per year near the Simulium breeding sites (Renz and Bronsvoort unpublished observations). In our study although there was a low level of transmission in the area, the nodule burden of the animals did not alter significantly up until 24 mpt after allowing for the removal of 20 nodules per animal. An alternative hypothesis could be that there is some form of density-dependent regulation of the adult worms with only very low turnover of the adult population. This theory is supported by the observation that control animals in the current experiment failed to acquire new nodules in the first year as determined by the regular 6 monthly counts. In addition different trials where L3's were experimentally inoculated into calves have failed to produce a clear positive correlation between the number of infective larvae inoculated and the number of nodules that subsequently developed [33,34]. If in fact there is some form of density dependent regulation of the adult population with individual worm burdens determined by some host-parasite balance then it may actually be more beneficial to sterilise the female worm, thus preventing the majority of new infections from developing.
The sterilising effect of high doses of ivermectin, combined with their prophylactic potential [19], will ensure the usefulness of this drug and/or related compounds. With community-directed distributions being set up under the APOC scheme there will be a much improved potential for timing treatments and/or increasing the frequency of treatments to maximise these other useful effects. High doses and increased frequency of treatment has been shown here to reduce female worm fertility in O. ochengi and similar results have been demonstrated in humans for O. volvulus [35]. This potentially could help to reduce the length of control programmes though it is doubtful whether high coverage levels could be sustained long enough to achieve worldwide eradication [36]. In addition, work on O. ochengi suggests that people receiving ivermectin are unlikely to develop immunity to O. volvulus and are therefore highly susceptible should drug distribution cease [32]. However, WHO itself recognises that ivermectin by itself can not completely interrupt transmission of O. volvulus [37-39] and therefore work should continue to find and develop an effective macrofilaricide that would be used in combination with ivermectin.
In conclusion, although this experiment has failed to demonstrate that high and repeated doses of ivermectin or doramectin are macrofilaricidal, it has shown that they both have a profound effect on embryogenesis producing a prolonged period of sterility in female worms. However, simulation modelling suggests that ivermectin alone will not interrupt transmission [36]. This appears to be born out by epidemiological studies from the field where the conclusion was that repeated ivermectin mass annual treatment will not lead to the elimination of transmission of onchocerciasis from West Africa though data on 6-monthly treatments was not sufficient to draw a conclusion [39]. There is, therefore, still an urgent need to continue the search for a safe effective macrofilaricidal compound.
Authors' contributions
BB: Carried out the field work and project management in Cameroon and was responsible for data analysis and preparation of the manuscript
AR: Responsible for project inception and design and preparation of the manuscript
VNT: Responsible for field work, project set-up and design and overall project coordination in Cameroon
VT: Responsible for field work and project set-up
DE: Responsible for analysis and interpretation of field samples in the laboratory
AT: Responsible for project inception and design and preparation of the manuscript
Note
1Minitab Inc, http:\\www.minitab.com
2BMDP Statistical Software, Inc., Cork Technology Park, Cork, Ireland.
Acknowledgements
This investigation was funded by the MACROFIL programme of the UNDP/World Bank/ WHO Special Programme for Research and Training in Tropical Diseases (ID no. 910583). Histological sections were prepared by Mrs. H. Böggering, University of Tübingen. Pfizer (Germany) are thanked for donating doramectin. We thank Dr. Banser, director of IRAD, for his support. We acknowledge Jean Ebene for his technical support. We are most grateful to Abbo Bakar, Abbo Soudi and Yakoubou for their livestock skills, dedication and great enthusiasm. The authors would also like to thank Dr. Brian Duke for his helpful comments.
==== Refs
WHO Onchocerciasis and its control WHO Technical Report Series 1995 852 Geneva, WHO
Molyneux DH Davies JB Onchocerciasis control: Moving towards the millennium Parasitol Today 1997 13 418 425 15275143 10.1016/S0169-4758(97)00142-7
Remme JHF The African Program for Onchocerciasis Control - Preparing to Launch Parasitol Today 1995 11 403 406 10.1016/0169-4758(95)80017-4
Shoop NL Mrozik H Fisher HM Structure and activity of avermectins and milbemycins in animal health Veterinary Parasitology 1995 59 139 156 7483237 10.1016/0304-4017(94)00743-V
Goudie AC Evans NA Gration FAF Bishop BF Gibson SP Holdom KS Kaye B Wicks SR Lewis P Weatherley AJ Bruce CI Herbert A Seymour DJ Doramectin - a potent novel endecocide Veterinary Parasitology 1993 49 5 15 8236738 10.1016/0304-4017(93)90218-C
Awadzi K Opoku NO Addy ET Quartey BT The Chemotherapy of Onchocerciasis.19. The Clinical and Laboratory Tolerance of High-Dose Ivermectin Trop Med Parasitol 1995 46 131 137 8525285
Richards FO Flores RZ Duke BOL Dynamics of Microfilariae of Onchocerca-Volvulus over the 1st 72 Hours after Treatment with Ivermectin Trop Med Parasitol 1989 40 299 303 2617037
Greene BM Modern Medicine Versus an Ancient Scourge - Progress toward Control of Onchocerciasis J Infect Dis 1992 166 15 21 1607688
Duke BOL Zeaflores G Munoz B The Embryogenesis of Onchocerca-Volvulus over the 1st Year after a Single Dose of Ivermectin Trop Med Parasitol 1991 42 175 180 1801140
Duke BOL Zeaflores G Castro J Cupp EW Munoz B Comparison of the Effects of a Single Dose and of 4 6-Monthly Doses of Ivermectin on Adult Onchocerca-Volvulus Am J Trop Med Hyg 1991 45 132 137 1867345
Chavasse DC Post RJ Lemoh PA Whitworth JAG The Effect of Repeated Doses of Ivermectin on Adult Female Onchocerca-Volvulus in Sierra-Leone Trop Med Parasitol 1992 43 256 262 1293732
Klager S Whitworth JAG Post RJ Chavasse DC Downham MD How Long Do the Effects of Ivermectin on Adult Onchocerca- Volvulus Persist Trop Med Parasitol 1993 44 305 310 8134772
Klager SL Whitworth JAG Downham MD Viability and fertility of adult Onchocerca volvulus after 6 years of treatment with ivermectin Trop Med Int Health 1996 1 581 589 8911442
Duke BOL Zeaflores G Castro J Cupp EW Munoz B Effects of 3-Month Doses of Ivermectin on Adult Onchocerca- Volvulus Am J Trop Med Hyg 1992 46 189 194 1539754
Whitworth JAG Maude GH Downham MD Clinical and parasitological responses after up to 6.5 years of ivermectin treatment for onchocerciasis Trop Med Int Health 1996 1 786 793 8980590 10.1046/j.1365-3156.1996.d01-3.x
Plaisier AP Alley ES Boatin BA Vanoortmarssen GJ Remme H Devlas SJ Bonneux L Habbema JDF Irreversible Effects of Ivermectin on Adult Parasites in Onchocerciasis Patients in the Onchocerciasis Control Program in West-Africa J Infect Dis 1995 172 204 210 7797912
Trees AJ Onchocerca-Ochengi - Mimic, Model or Modulator of Onchocerca- Volvulus Parasitol Today 1992 8 337 339 15463530 10.1016/0169-4758(92)90068-D
Renz A Trees AJ Achukwi D Edwards G Wahl G Evaluation of Suramin, Ivermectin and Cgp-20376 in a New Macrofilaricidal Drug Screen, Onchocerca-Ochengi in African Cattle Trop Med Parasitol 1995 46 31 37 7631125
Tchakoute VL Bronsvoort M Tanya V Renz A Trees AJ Chemoprophylaxis of Onchocerca infections: in a controlled, prospective study ivermectin prevents calves becoming infected with O-ochengi Parasitology 1999 118 195 199 10028534 10.1017/S0031182098003680
Trees AJ Wood VL Bronsvoort M Renz A Tanya VN Animal models - Onchocerca ochengi and the development of chemotherapeutic and chemoprophylactic agents for onchocerciasis Ann Trop Med Parasitol 1998 92 S175 S179 9861291 10.1080/00034989859799
Comley JCW Rees MJ Turner CH Jenkins DC Colorimetric quantification of filarial viability International Journal of Parasitology 1989 19 77 83 2707965 10.1016/0020-7519(89)90024-6
Schulz-Key H The collagenase technique: how to isolate and examine adult Onchocerca volvulus for the evaluation of drug effects Tropical Medical Parasitology 1988 39 423 440
Langworthy NG Renz A Mackenstedt U Henkle-Duhrsen K Bronsvoort MBD Tanya VN Donnelly MJ Trees AJ Macrofilaricidal activity of tetracycline against the filarial nematode Onchocerca ochengi: elimination of Wolbachia precedes worm death and suggests a dependent relationship Proc R Soc Lond Ser B-Biol Sci 2000 267 1063 1069 10.1098/rspb.2000.1110
Cross HF Bronsvoort BM Wahl G Renz A AchuKwi D Trees AJ The entry of ivermectin and suramin into Onchocerca ochengi nodules Ann Trop Med Parasitol 1997 91 393 401 9290846 10.1080/00034989761012
Cross HF Renz A Trees AJ In-vitro uptake of ivermectin by adult male Onchocerca ochengi Ann Trop Med Parasitol 1998 92 711 720 9924550 10.1080/00034989859177
Schulz-Key H Klager S Awadzi K Diallo S Greene BM Lariviere M Aziz MA Treatment of human onchocerciasis: the efficacy of ivermectin on the parasite Tropical Medical Parasitology 1985 36 20
Chavasse DC Post RJ Davies JB Whitworth JAG Absence of Sperm from the Seminal Receptacle of Female Onchocerca-Volvulus Following Multiple Doses of Invermectin Trop Med Parasitol 1993 44 155 158 8256088
Schulz-Key H Karam M Periodic reproduction of Onchocerca volvulus Parasitology Today 1986 2 284 286 15462735 10.1016/0169-4758(86)90138-9
Geary TG Sims SM Thomas EM Vanover L Davis JP Winterrowd CA Klein RO Ho NFH Thompson DP Haemonchus contortus: Ivermectin-induced paralysis of the pharynx Experimental Parasitology 1993 77 88 96 8344410 10.1006/expr.1993.1064
Duke BOL Zeaflores G Castro J Cupp EW Munoz B Effects of Multiple Monthly Doses of Ivermectin on Adult Onchocerca-Volvulus Am J Trop Med Hyg 1990 43 657 664 2267970
Trees AJ Wahl G Klager S Renz A Age-Related Differences in Parasitosis May Indicate Acquired- Immunity against Microfilariae in Cattle Naturally Infected with Onchocerca-Ochengi Parasitology 1992 104 247 252 1594291
Njongmeta LM Nfon CK Gilbert J Makepeace BL Tanya VN Trees AJ Cattle protected from onchocerciasis by ivermectin are highly susceptible to infection after drug withdrawal Int J Parasit 2004 34 1069 1074 10.1016/j.ijpara.2004.04.011
Achukwi Daiber WH Renz A Wahl G Wanji S Prepatency Period and Some Aspects of the Epizootiology of Onchocerca-Ochengi Infestation in Cattle in the Adamawa Plateau, Cameroon Parasite-J Soc Fr Parasitol 1994 1 10 12
Trees AJ Graham SP Renz A Bianco AE Tanya V Onchocerca ochengi infections in cattle as a model for human onchocerciasis: recent developments Parasitology 2000 120 S133 S142 10874716 10.1017/S0031182099005788
Gardon J Boussinesq M Kamgno J Gardon-Wendel N Demanga N Duke BOL Effects of standard and high doses of ivermectin on adult worms of Onchocerca volvulus: a randomised controlled trial Lancet 2002 360 203 210 12133654 10.1016/S0140-6736(02)09456-4
Winnen M Plaisier AP Alley ES Nagelkerke NJD van Oortmarssen G Boatin BA Habbema JDF Can ivermectin mass treatments eliminate onchocerciasis in Africa? Bull World Health Organ 2002 80 384 390 12077614
Samba EM Public Health in Action 1: the Onchocerciasis Control Programme in West Africa 1994 Geneva, WHO
Seidenfaden R Fischer A Bonow I Ekale D Tanya V Renz A Combined benefits of annual mass treatment with ivermectin and cattle zooprophylaxis on the severity of human onchocerciasis in northern Cameroon Trop Med Int Health 2001 6 715 725 11555439 10.1046/j.1365-3156.2001.00771.x
Borsboom G Boatin B Nagelkerke N Agoua H Akpoboua K Alley EW Bissan Y Renz A Yameogo L Remme J Habbema JD Impact of ivermectin on onchocerciasis transmission: assessing the empirical evidence that repeated ivermectin mass treatments may lead to elimination/eradication in West-Africa Filaria Journal 2003 2 8 12769825 10.1186/1475-2883-2-8
|
16086838
|
PMC1200428
|
CC BY
|
2021-01-04 16:38:25
|
no
|
Filaria J. 2005 Aug 8; 4:8
|
utf-8
|
Filaria J
| 2,005 |
10.1186/1475-2883-4-8
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-571595524110.1186/1465-9921-6-57ResearchExpression of HSP47 in Usual Interstitial Pneumonia and Nonspecific Interstitial Pneumonia Kakugawa Tomoyuki [email protected] Hiroshi [email protected] Tomayoshi [email protected] Hiroshi [email protected] Seiko [email protected] Noriho [email protected] Sumako [email protected] Kanako [email protected] Mariko [email protected] Yohei [email protected] Shigeru [email protected] Second Department of Internal Medicine, Nagasaki University School of Medicine, Nagasaki, Japan2 Department of Pathology, Nagasaki University Hospital, Nagasaki, Japan3 Biostatistics Section, Division of Scientific Data Registry, Atomic Bomb Disease Institute, Nagasaki University Graduate School of Biomedical Sciences, Nagasaki; Japan2005 14 6 2005 6 1 57 57 8 2 2005 14 6 2005 Copyright © 2005 Kakugawa et al; licensee BioMed Central Ltd.2005Kakugawa et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Heat shock protein (HSP) 47, a collagen-specific molecular chaperone, is involved in the processing and/or secretion of procollagens, and its expression is increased in various fibrotic diseases. The aim of this study was to determine whether quantitative immunohistochemical evaluation of the expression levels of HSP47, type I procollagen and α-smooth muscle actin (SMA) allows the differentiation of idiopathic usual interstitial pneumonia (UIP) from UIP associated with collagen vascular disease (CVD) and idiopathic nonspecific interstitial pneumonia (NSIP).
Methods
We reviewed surgical lung biopsy specimens of 19 patients with idiopathic UIP, 7 with CVD-associated UIP and 16 with idiopathic NSIP and assigned a score for the expression of HSP47, type I procollagen and α-SMA in type II pneumocytes and/or lung fibroblasts (score 0 = no; 1 = weak; 2 = moderate; 3 = strong staining).
Results
The expression level of HSP47 in type II pneumocytes of idiopathic UIP was significantly higher than in CVD-associated UIP and idiopathic NSIP. The expression of HSP47 in fibroblasts was significantly higher in idiopathic UIP and idiopathic NSIP than in CVD-associated UIP. The expression of type I procollagen in type II pneumocytes was significantly higher in idiopathic UIP than in idiopathic NSIP. The expression of type I procollagen in fibroblasts was not different in the three groups, while the expression of α-SMA in fibroblasts was significantly higher in idiopathic UIP than in idiopathic NSIP.
Conclusion
Our results suggest the existence of different fibrotic pathways among these groups involved in the expression of HSP47 and type I procollagen.
usual interstitial pneumonianonspecific interstitial pneumoniacollagen vascular diseaseheat shock protein 47type I procollagens
==== Body
Background
Heat shock protein (HSP) 47 is a collagen-binding, stress-inducible protein localized in the endoplasmic reticulum and is never released into the extracellular matrix. HSP47 plays a specific role in the intracellular processing of procollagen production as a collagen-specific molecular chaperone [1-4]. HSP47 expression has been shown to be upregulated in experimental animal models of fibrosis, including murine bleomycin-induced pulmonary fibrosis [5,6], rat peritoneal sclerosis [7] and carbon tetrachloride-induced rat liver cirrhosis [8]. In addition, we previously reported increased expression of human HSP47 in the fibrotic lesions of idiopathic pulmonary fibrosis (IPF) [9], fibrotic transplanted kidney [10], and peritoneal sclerosis [11]. Recent reports have demonstrated that HSP47 expression is highly tissue- and cell-specific, and is restricted to most phenotypically altered collagen-producing cells, and correlates well with that of collagen [9-11]. These findings suggest the important role of HSP47 in collagen synthesis in various fibrotic disorders. Furthermore, it was demonstrated that inhibition of HSP47 by antisense oligodeoxynucleotides markedly suppressed the production of collagen in 3T6 cells [4], in experimental proliferative glomerulonephritis [12] and in experimental peritoneal fibrosis [7]. These findings suggest that HSP47 might be a promising target for the treatment of fibrotic diseases.
Increased numbers of myofibroblasts in pulmonary fibrosis has been documented in both human lung tissues and those of animal models [5,6,13,14]. Myofibroblasts participate in remodeling and progressive destruction of the lung parenchyma through several mechanisms including increased extracellular matrix synthesis and contractility of the lung parenchyma [15]. We also showed previously that bleomycin treatment induced a marked increase in myofibroblasts in the active fibrotic areas of the lung from the early fibrotic stage in mice [5,6]. Activation of alpha-smooth muscle actin (α-SMA)-positive myofibroblasts is believed to play a key role in the progression of pulmonary fibrosis.
Patients with usual interstitial pneumonia (UIP) associated with a collagen vascular disease (CVD) have an improved prognosis compared with patients with idiopathic UIP [16]. Therefore, whether CVD-associated UIP and idiopathic UIP are the same pathological entity remains unresolved. Although identifying the pattern of UIP on surgical lung biopsy provides important prognostic data in distinction from other patterns of interstitial pneumonia [17-19], there are few studies that analyzed the individual histopathologic features comprising this pattern [20]. Although Flaherty et al [16] recently reported that patients with CVD-associated UIP have fewer fibroblastic foci in the lung and better prognosis than those with idiopathic UIP, the phenotypic difference of lung cells (e.g., fibroblasts and alveolar epithelial cells) between these two diseases and its association with the prognosis are poorly understood.
The recognition that lung biopsy samples of some patients with idiopathic interstitial disease do not fit into any well-defined histopathologic patterns of idiopathic interstitial pneumonia led to the proposal of the terms "unclassified interstitial pneumonia" by Kitaichi in 1990 [21] and nonspecific interstitial pneumonia (NSIP) by Katzenstein and Fiorelli in 1994 [22]. NSIP is characterized by varying degrees of inflammation and fibrosis within the alveolar walls. However, the most distinctive feature of NSIP is the temporal uniformity, which is in sharp contrast to the temporal heterogeneity seen in UIP. Among idiopathic interstitial pneumonias, NSIP has a more favorable response to corticosteroids and a better prognosis than UIP [17-19,23,24]. Based on these studies, NSIP is currently accepted as a clinicopathological entity in idiopathic interstitial pneumonias. Based on this background, we speculated that the pathogenesis of NSIP is different from UIP. In this regard, the phenotypic difference in fibroblasts and epithelial cells between these diseases is also not well elucidated.
The main hypothesis of the present study was that the expression levels of HSP47 in alveolar epithelial cells and lung fibroblasts are different in idiopathic UIP, CVD-associated UIP and idiopathic NSIP, thus allowing differentiation of these conditions. To test our hypothesis, we determined the expression levels of HSP47, type I procollagen and α-SMA in fibroblasts and type II pneumocytes in surgical lung biopsy specimens by immunohistochemistry.
Materials and methods
Study populations
The subjects of this study were patients enrolled in the hospitals of Nagasaki University School of Medicine. The study protocol was approved by the institutional review board, and informed consent was obtained from each patient. They included 19 patients with idiopathic UIP, 7 with CVD-associated UIP and 16 with idiopathic NSIP. Patients with idiopathic UIP and idiopathic NSIP had neither shown any signs nor positive serological and other markers of CVD. The associated diagnosis in the CVD-associated UIP patients were systemic sclerosis (n = 3), systemic sclerosis with Sjögren syndrome (n = 1), polymyositis with Sjögren syndrome (n = 1), primary Sjögren syndrome (n = 1) and mixed connective tissue disease (n = 1). Idiopathic NSIP patients included 8 with cellular and fibrosing pattern and 8 with fibrosing pattern. None of these patients had received steroids or other immunosuppressants therapy at the time of clinical sample collection. Patients characteristics before lung biopsy including age, smoking history, period from the onset to the surgical lung biopsy, results of pulmonary function tests and arterial blood gas analysis were collected from either the hospital medical records or those of the general practitioners. In the patients in this study, a suspicion of interstitial pneumonia was based on symptoms, physiologic abnormalities, and HRCT findings. The diagnosis was pathologically confirmed by open lung biopsy or video-assisted thoracoscopic surgery from multiple lobes in all patients and classified according to the American Thoracic Society/European Respiratory Society consensus criteria [25]. Control lung tissues were obtained from normal areas of lungs surgically removed for lung cancer (11 men and 9 women; median age, 62 years; range, 47 to 81 years).
Antibodies
The primary antibodies used for the immunohistochemical studies included anti-human procollagen type I (Chemicon, Temecula, CA), anti-α-SMA (Neomarkers, Fremont, CA), anti-human cytokeratin (clone MNF116; Dako Corporation, Carpinteria, CA) and anti-HSP47 (Biotechnologies Corp., Victoria, BC, Canada). α-SMA and cytokeratin were used as markers of myofibroblasts and epithelial cells, respectively. Negative control studies were performed by using irrelevant immunoglobulin G with the same subclass of the first antibodies instead of the primary antibodies.
Immunohistochemistry
Immunohistochemistry was performed with the conventional avidin-biotin-peroxidase histochemical technique using Vecstain Elite ABC Kit (Vector Laboratories, Burlingame, CA). Briefly, sequential paraffin sections (4-μm-thick) were deparaffinized with toluene and rinsed thoroughly with ethanol. The sections were then soaked in 0.3% H2O2 with absolute methanol for 20 minutes at room temperature to inactivate the endogenous peroxidase activity. They were incubated with blocking serum for 30 minutes, and then covered with primary antibodies and incubated for 1 hour. After washing in phosphate-buffered saline (PBS), the sections were processed further using the kits according to the instructions provided by the manufacturer, and then developed with 3,3'-diaminobenzidine and H2O2, followed by the Mayer's hematoxylin staining method.
Pathologic Assessment
The profusion of fibroblastic foci in idiopathic UIP and CVD-associated UIP was graded semiquantitatively using a grading system of 0 to 3 (0 = absent; 1 = mild; 2 = moderate; 3 = marked) according to the method described previously [16]. The staining intensity and distribution of HSP47, α-SMA and type I procollagen in fibroblasts in active fibrotic areas (fibroblastic foci in UIP cases; alveolar septal interstitium expanded by fibrosis and intra-alveolar organizing fibrosis in NSIP cases) was also scored semiquantitatively using a grading system of 0 to 3 (0 = no staining; 1 = weak staining; 2 = moderate staining; 3 = strong staining). The expression levels of HSP47 and type I procollagen in type II pneumocytes were scored in a similar manner. The fibroblastic foci score and immunohistochemical score for each patient was calculated by averaging the score of each lobe. Histological sections were assessed independently, twice by each of two observers who were blind to the groups. The results were reproducible for interobserver and intraobserver variability. Statistical analysis was applied to representative results of one observer.
Statistical Analysis
All values were expressed as median (range). Differences in categorical data between two groups were assessed by the Fisher's exact probability test for data of two categories and by chi-square test for those of three or more categories. Differences in continuous data between two groups were assessed by using the Wilcoxon rank-sum test and among three groups by the Kruskal-Wallis rank test. In addition, logistic regression was used to examine the relationships between the presence/absence of CVD-associated UIP, idiopathic NSIP and other variables. Statistical significance was defined by a p value of less than 0.05. Statistical analyses were carried out using SAS (SAS Institute, Inc., Cary, NC).
Results
Patient Characteristics
Table 1 shows the characteristics of patients enrolled in this study. Baseline demographic and physiologic characteristics were similar among the groups, except for sex and percentage of predicted total lung capacity.
Histopathological and Immunohistochemical Findings
Photomicrographs of histological and immunohistochemical studies of representative surgical lung biopsy specimens are shown in Figure 1 (A-E; idiopathic UIP, F-J; CVD-associated UIP, K-O; idiopathic NSIP). Figure 1A–E, F–J and 1K–O represent sequential sections, respectively. Histopathological examination revealed fibroblastic foci in both idiopathic UIP (Fig. 1A) and CVD-associated UIP (Fig. 1F) and fibroblast proliferation in idiopathic NSIP (Fig. 1K). In all diseases, distinct aggregates of closely spaced fibroblasts with little intervening collagen deposition were seen and hyperplastic alveolar lining cells covered their luminal surface. Cuboidal epithelial cells were stained with cytokeratin, indicating that these cells were type II pneumocytes (Fig. 1B, G and 1L). Strong expression of HSP47 was noted predominantly in fibroblasts and most of type II pneumocytes in idiopathic UIP (Fig. 1C). In contrast, weak or no expression of HSP47 was noted in fibroblasts and type II pneumocytes in CVD-associated UIP (Fig. 1H). In idiopathic NSIP, strong expression of HSP47 was noted in fibroblasts, but not in type II pneumocytes (Fig. 1M). Type I procollagen was strongly expressed predominantly in fibroblasts and most of type II pneumocytes in idiopathic UIP (Fig. 1D), but neither in CVD-associated UIP (Fig. 1I) nor idiopathic NSIP (Fig. 1N). Expression of α-SMA was noted in some of fibroblasts in all three diseases, indicating that these cells were myofibroblasts (Fig. 1E, J and 1O). Negative control studies using non-specific immunoglobulin-G revealed no positive cells (data not shown).
Figure 1 Photomicrographs of histopathological and immunohistochemical studies of representative surgical lung biopsy specimens (A-E; idiopathic UIP, F-J; CVD-associated UIP, K-O; idiopathic NSIP, scale bar = 100 μm). Histopathological examination (hematoxylin-eosin staining) revealed fibroblastic foci in both idiopathic UIP (A) and CVD-associated UIP (F), and fibroblast proliferation in idiopathic NSIP (K). Hyperplastic cuboidal epithelial cells were stained with cytokeratin, indicating that these cells were type II pneumocytes (B, G and L). Strong expression of HSP47 was noted predominantly in fibroblasts and type II pneumocytes in idiopathic UIP (C). Weak or no expression of HSP47 was noted in fibroblasts and type II pneumocytes in CVD-associated UIP (H). In idiopathic NSIP, strong expression of HSP47 was noted in fibroblasts, but not in type II pneumocytes (M). Type I procollagen was strongly expressed predominantly in fibroblasts and type II pneumocytes in idiopathic UIP (D), but neither in CVD-associated UIP (I) nor idiopathic NSIP (N). Expression of α-SMA was noted in some of fibroblasts, indicating that these cells were myofibroblasts, in all three diseases (E, J and O). Insets c, d, h, i, m and n are pictures taken at high power magnification (scale bar = 20 μm) of corresponding C, D, H, I, M and N sections to clearly show the phenotypic difference of type II pneumocytes. α-SMA = α-smooth muscle actin; CVD = collagen vascular disease; HSP47 = heat shock protein 47; NSIP = nonspecific interstitial pneumonia; UIP = usual interstitial pneumonia.
Table 1 Patient Characteristics
Idiopathic UIP CVD-associated UIP Idiopathic NSIP p value
Age (years) 64 (34–72) 44 (38–63) 58 (28–75) 0.06
Sex (male/female) 17/2 1/6 5/11 0.0002
Smoking (none/ex/smoke) 6/7/6 5/2/0 10/3/2 0.15
Symptom onset (months) 9 (5–38) 12 (4–35) 6 (1–74) 0.44
Spirometry:
VC (L) 2.63 (1.28–3.81) 2.01 (1.34–2.63) 2.50 (1.11–3.39) 0.10
predicted VC (%) 78.6 (55.8–117.0) 67.8 (53.5–112.3) 79.9 (50.0–108.1) 0.59
FEV1 (L) 2.18 (1.08–2.81) 1.67 (1.13–2.28) 1.72 (0.99–2.98) 0.05
predicted FEV (%) 84.10 (65.5–90.6) 84.3 (78.1–94.6) 80.79 (51.0–96.2) 0.19
Gas exchange:
DLco (ml/min/mmHg) 9.13 (6.5–14.65) 9.94 (3.96–20.27) 11.45 (6.0–19.85) 0.35
predicted DLco (%) 49.00 (37.2–97.3) 43.80 (25.2–61.4) 57.85 (44.7–109.3) 0.05
Lung volume:
predicted TLC (%) 62.45 (45.2–96.7) 79.90 (58.9–107.6) 71.55 (62.9–112.4) 0.04
TLC (L) 3.52 (1.78–5.39) 3.07 (2.0–4.0) 3.83 (1.8–8.8) 0.41
Arterial blood gases:
PaO2 (mmHg) 82.5 (58.3–108.0) 89.2 (40.5–97.5) 83.4 (67.0–92.9) 0.43
Data are median (range).
α-SMA = α-smooth muscle actin; CVD = collagen vascular disease; DLco = diffusing capacity for carbon monoxide; HSP47 = heat shock protein 47; NSIP = nonspecific interstitial pneumonia; Symptom onset = period from the onset to the surgical lung biopsy; TLC = total lung capacity; UIP = usual interstitial pneumonia; VC = vital capacity.
Results of Pathologic Assessment
Patients with idiopathic UIP had higher profusion of fibroblastic foci (median, 2.0 [range, 0.67–3]) than those with CVD-associated UIP (1.0 [0–2]), albeit non-significantly (p = 0.06). Patients with idiopathic UIP (P < 0.01), CVD-associated UIP (P < 0.05) and idiopathic NSIP (P < 0.01) had a significantly higher expression of HSP47 in fibroblasts than that in control. Patients with idiopathic UIP (P < 0.01) and idiopathic NSIP (P < 0.05) had a significantly higher expression of HSP47 in fibroblasts than that in CVD-associated UIP (Fig. 2A). Patients with idiopathic UIP (P < 0.01) and idiopathic NSIP (P < 0.01) had a significantly higher expression of HSP47 in type II pneumocytes than that in control, while there was no significant difference between CVD-associated UIP and control. Expression of HSP47 in type II pneumocytes of patients with idiopathic UIP was significantly higher than those of CVD-associated UIP (P < 0.05) and idiopathic NSIP (P < 0.01, Fig. 2B). Patients with idiopathic UIP (P < 0.01) and idiopathic NSIP (P < 0.05) had a significantly higher expression of type I procollagen in fibroblasts compared with control, while there was no significant difference between CVD-associated UIP and control (1.0 [0–3], 1.0 [0–1.5], 1.0 [0–1.5] and 0.0 [0–1] for idiopathic UIP, CVD-associated UIP, idiopathic NSIP and control, respectively). Patients with idiopathic UIP (P < 0.01), CVD-associated UIP (P < 0.05) and idiopathic NSIP (P < 0.01) had a significantly higher expression of type I procollagen in type II pneumocytes than that in control (Fig. 2C). There was no significant difference in the expression of type I procollagen in fibroblasts among idiopathic UIP, CVD-associated UIP and idiopathic NSIP, while that in type II pneumocytes in patients with idiopathic UIP was significantly higher than in idiopathic NSIP (P < 0.01, Fig. 2C). Patients with idiopathic UIP (P < 0.01), CVD-associated UIP (P < 0.01) and idiopathic NSIP (P < 0.01) had a significantly higher expression of α-SMA in fibroblasts than that in control (Fig. 2D). Patients with idiopathic UIP had a significantly higher expression of α-SMA in fibroblasts than those with idiopathic NSIP (P < 0.05, Fig. 2D).
Figure 2 Box-and-whisker plots of HSP47, type I procollagen and α-SMA in patients with idiopathic UIP, CVD-associated UIP and idiopathic NSIP. (A) Patients with idiopathic UIP and idiopathic NSIP (P < 0.05) had a significantly higher expression of HSP47 in fibroblasts than those with CVD-associated UIP. (B) Expression of HSP47 in type II pneumocytes of patients with idiopathic UIP was significantly higher than that of CVD-associated UIP and idiopathic NSIP. (C) Expression of type I procollagen in type II pneumocytes of patients with idiopathic UIP was significantly higher than in idiopathic NSIP. (D) Patients with idiopathic UIP had a significantly higher expression of α-SMA in fibroblasts than those with idiopathic NSIP. *P < 0.01, compared with control. †P < 0.05, compared with control. The boxes represent the 25th to 75th percentiles, the solid lines within the boxes show the median values, the whiskers are the 10th and 90th percentiles, and the points represent outliers. For abbreviations, see Figure 1.
We further examined the differences between groups by univariate analysis based on the results of the pathological assessment (Table 2). The expression of HSP47 in fibroblasts was the most discriminative baseline feature for separating idiopathic UIP from CVD-associated UIP. The odds ratio of having idiopathic UIP compared with CVD-associated UIP was 12.68 for a unit increase in the mean score of the expression of HSP47 in fibroblasts. The expression of HSP47 in type II pneumocytes was the most discriminative baseline feature for separating idiopathic UIP from idiopathic NSIP. The odds ratio of having idiopathic UIP compared with idiopathic NSIP was 6.66 for a unit increase in the mean score of the expression of HSP47 in type II pneumocytes.
Table 2 Univariate Analysis of the Results of Pathologic Assessment.
Predictor Odds Ratio 95% confidence interval p Value
Idiopathic UIP vs CVD-associated UIP
Fibroblastic foci score 3.73 0.94–14.79 0.06
HSP47 in fibroblasts 12.68 1.53–105.19 0.02
HSP47 in type II pneumocytes 3.39 1.10–10.45 0.03
α-SMA in fibroblasts 4.78 0.44–51.27 0.20
type I procollagen in fibroblasts 2.73 0.61–12.23 0.19
type I procollagen in type II pneumocytes 3.43 0.77–15.32 0.11
Idiopathic UIP vs Idiopathic NSIP
HSP47 in fibroblasts 1.91 0.74–4.95 0.17
HSP47 in type II pneumocytes 6.66 1.76–25.15 0.01
α-SMA in fibroblasts 4.46 0.94–21.17 0.06
type I procollagen in fibroblasts 3.60 0.94–13.77 0.06
type I procollagen in type II pneumocytes 6.10 1.43–25.93 0.01
For abbreviations, see Table 1.
Discussion
It is important to know the cell type(s) that contributes to lung fibrogenesis. The histopathologic pattern of UIP is well-characterized, with the cardinal feature being the patchy distribution of temporally heterogeneous fibrosis, comprising areas of established fibrosis with adjacent foci of fibroblastic proliferation, so-called fibroblastic foci. Recent studies have emphasized the importance of the fibroblastic focus, a manifestation of ongoing lung injury in patients with established fibrosis [13,15,26-29]. A number of investigators have stressed the prognostic value of quantifying fibroblastic foci in patients with idiopathic interstitial pneumonia [28,29]. Furthermore, a previous study showed that patients with CVD-associated UIP have fewer fibroblastic foci and better prognosis compared to patients with idiopathic UIP [16]. In idiopathic NSIP, which exhibits a more favorable response to corticosteroids and has a better prognosis than idiopathic UIP [17-19,23,24], fibroblastic foci with dense fibrosis are inconspicuous or absent [25]. Thus, fibroblasts are believed to play an important role in the progression of chronic pulmonary fibrosis. However, previous studies did not fully elucidate the phenotypic differences in fibroblasts among idiopathic UIP, CVD-associated UIP and idiopathic NSIP.
We identified higher expression of HSP47 in fibroblasts in patients with idiopathic UIP and idiopathic NSIP compared with CVD-associated UIP. We also showed that patients with idiopathic UIP had a significantly higher expression of α-SMA in fibroblasts than those with idiopathic NSIP. The differential expression of HSP47 and α-SMA in fibroblasts suggests that the underlying pathogenic process of fibrosis in these diseases may be distinctly different. The results emphasized that the expression of HSP47 in fibroblasts is the most discriminative features between idiopathic UIP and CVD-associated UIP, as identified by logistic regression. These findings imply that the expression of HSP47 in fibroblasts could also have important prognostic implications in interstitial pneumonias.
Our results also emphasized the importance of the phenotypic difference of epithelial cells among idiopathic UIP, CVD-associated UIP and idiopathic NSIP. Type II pneumocytes hyperplasia in areas of inflammation and fibrosis is commonly seen in both UIP and NSIP [25]. Although several studies performed in experimental models emphasized the importance of the alveolar epithelium in normal repair [30,31], the prevailing hypothesis regarding the pathogenesis of interstitial pulmonary fibrosis has been that the disease is due to a chronic unsolved inflammatory response, and in this context, the possible role of the epithelium has been largely neglected. However, there is increasing evidence supporting the notion that epithelial injury in the absence of ongoing inflammation is sufficient to cause the development of pulmonary fibrosis [15,30,31] and that alveolar epithelial cells are the primary source of cytokines and growth factors involved in fibroblast migration and proliferation [15]. In spite of these findings, the precise biological link between alveolar epithelial injury and fibrosis represent a challenging puzzle in which several pieces remain to be assembled.
We recently showed in rodent bleomycin-induced pulmonary fibrosis model that type II pneumocytes start to express HSP47 with the progression of fibrosis [5,6]. The present study extended this finding by showing that hyperplastic type II pneumocytes in idiopathic UIP express both HSP47 and type I procollagen. These findings suggest that increased number of type II pneumocytes in addition to fibroblasts produce type I procollagen through the induction of HSP47 and play an important role in the development of fibrosis. Our study also demonstrates a lower expression of HSP47 and type I procollagen in type II pneumocytes in CVD-associated UIP and idiopathic NSIP compared with idiopathic UIP. Interestingly, the expression level of HSP47 in type II pneumocytes was much higher in idiopathic UIP than in idiopathic NSIP, while there was no significant difference in the expression level of HSP47 in fibroblasts between them. Furthermore, the expression of type I procollagen in type II pneumocytes was different among idiopathic UIP, CVD-associated UIP and idiopathic NSIP in spite of the similar expression level of type I procollagen in fibroblasts. These findings suggest that the phenotypic difference of type II pneumocytes in interstitial pneumonias is more important biological feature than that of fibroblasts. In addition, we previously reported that treatment with pirfenidone, an anti-fibrotic agent, decreased HSP47-positive type II pneumocytes in bleomycin-induced pulmonary fibrosis in mice [6]. This finding further suggests that HSP47-positive type II pneumocytes might be promising target for therapeutic strategies designed for idiopathic UIP.
Several lines of evidence suggest that in some forms of tissue fibrosis, an epithelial-mesenchymal transition actively participates in the local formation of interstitial fibroblasts [32-34]. This process has been particularly studied in renal fibrosis. According to these findings, epithelial cells that detach from their basement membrane enter into epithelial-mesenchymal transition, express HSP47 and type I procollagen and divide as fibroblasts [32-34]. Whether this process takes place in the lung in IPF or in other forms of pulmonary fibrosis is largely unknown, but awaits further evaluation both in vitro and in vivo. The present findings that type II pneumocytes express HSP47 and type I procollagen in idiopathic UIP imply that a similar process could take place in idiopathic UIP. Further studies are warranted in order to elucidate the precise mechanism(s) involved in this process.
In this study, we also investigated whether the expression of HSP47, type I procollagen and α-SMA in fibroblasts and/or type II pneumocytes correlated with clinical course. However, no pathologic feature was associated with survival nor clinical data such as results of pulmonary function tests and arterial blood gas analysis (data not shown). This is probably because only small number of patients was assessed in this study. Further studies of a large number of patients are required.
Conclusion
In summary, we have shown that type II pneumocytes and/or lung fibroblasts of patients with idiopathic UIP, CVD-associated UIP and idiopathic NSIP express different levels of HSP47 and type I procollagen. Our findings support the concept that these diseases are different pathophysiological entities with different fibrotic pathways. We speculate that careful immunohistochemical evaluation of HSP47, α-SMA and type I procollagen in the lung may be a useful method to understand differences in the underlying pathogenic mechanisms of these diseases. Further studies of a large number of patients are required to determine the prognostic and therapeutic values of HSP47 expression.
Abbreviations
α-SMA: alpha-smooth muscle actin
CVD: collagen vascular disease
HSP: heat shock protein
IPF: idiopathic pulmonary fibrosis
NSIP: nonspecific interstitial pneumonia
PBS: phosphate-buffered saline
UIP: usual interstitial pneumonia
Authours' contributions
TK, HM, TH and MM have made substantial contribution to acquisition and analysis of data.
TK, HM, TH and SK have made substantial contributions to conception and design.
TK, HM, TH and SK have been involved in drafting the article.
HI, SN, NS, SY, KS and YM have been involved in revising it critically for important intellectual content.
Acknowledgements
The authors thank Dr. M Kitaichi (Department of Laboratory Medicine and Pathology, Kinki-chuo Chest Medical Center) for the valuable advice regarding the pathological diagnosis, Prof. T Nagayasu (Division of Surgical Oncology, Department of Translational Medical Sciences, Nagasaki University Graduate School of Biomedical Sciences) for undertaking the surgical lung biopsies and providing the tissues which made this work possible and Mr. A Yokoyama for the excellent technical support.
==== Refs
Nakai A Satoh M Hirayoshi K Nagata K Involvement of the stress protein HSP47 in procollagen processing in the endoplasmic reticulum J Cell Biol 1992 117 903 914 1577863 10.1083/jcb.117.4.903
Nagata K Saga S Yamada KM A major collagen-binding protein of chick embryo fibroblasts is a novel heat shock protein J Cell Biol 1986 103 223 229 3722264 10.1083/jcb.103.1.223
Saga S Nagata K Chen WT Yamada KM pH-dependent function, purification, and intracellular location of a major collagen-binding glycoprotein J Cell Biol 1987 105 517 527 3038929 10.1083/jcb.105.1.517
Sauk JJ Smith T Norris K Ferreira L Hsp47 and the translation-translocation machinery cooperate in the production of alpha 1(I) chains of type I procollagen J Biol Chem 1994 269 3941 3946 7905876
Ishii H Mukae H Kakugawa T Iwashita T Kaida H Fujii T Hayashi T Kadota J Kohno S Increased expression of collagen-binding heat shock protein 47 in murine bleomycin-induced pneumopathy Am J Physiol Lung Cell Mol Physiol 2003 285 L957 963 12842808
Kakugawa T Mukae H Hayashi T Ishii H Abe K Fujii T Oku H Miyazaki M Kadota J Kohno S Pirfenidone attenuates expression of HSP47 in murine bleomycin-induced pulmonary fibrosis Eur Respir J 2004 24 57 65 15293605 10.1183/09031936.04.00120803
Nishino T Miyazaki M Abe K Furusu A Mishima Y Harada T Ozono Y Koji T Kohno S Antisense oligonucleotides against collagen-binding stress protein HSP47 suppress peritoneal fibrosis in rats Kidney Int 2003 64 887 896 12911538 10.1046/j.1523-1755.2003.00169.x
Masuda H Fukumoto M Hirayoshi K Nagata K Coexpression of the collagen-binding stress protein HSP47 gene and the alpha 1(I) and alpha 1(III) collagen genes in carbon tetrachloride-induced rat liver fibrosis J Clin Invest 1994 94 2481 2488 7989606
Iwashita T Kadota J Naito S Kaida H Ishimatsu Y Miyazaki M Ozono Y Kohno S Involvement of collagen-binding heat shock protein 47 and procollagen type I synthesis in idiopathic pulmonary fibrosis: contribution of type II pneumocytes to fibrosis Hum Pathol 2000 31 1498 1505 11150375 10.1053/hupa.2000.20378
Abe K Ozono Y Miyazaki M Koji T Shioshita K Furusu A Tsukasaki S Matsuya F Hosokawa N Harada T Taguchi T Nagata K Kohno S Interstitial expression of heat shock protein 47 and alpha-smooth muscle actin in renal allograft failure Nephrol Dial Transplant 2000 15 529 535 10727549 10.1093/ndt/15.4.529
Shioshita K Miyazaki M Ozono Y Abe K Taura K Harada T Koji T Taguchi T Kohno S Expression of heat shock proteins 47 and 70 in the peritoneum of patients on continuous ambulatory peritoneal dialysis Kidney Int 2000 57 619 631 10652040
Sunamoto M Kuze K Tsuji H Ohishi N Yagi K Nagata K Kita T Doi T Antisense oligonucleotides against collagen-binding stress protein HSP47 suppress collagen accumulation in experimental glomerulonephritis Lab Invest 1998 78 967 972 9714184
Kuhn C McDonald JA The roles of the myofibroblast in idiopathic pulmonary fibrosis. Ultrastructural and immunohistochemical features of sites of active extracellular matrix synthesis Am J Pathol 1991 138 1257 1265 2024710
Pache JC Christakos PG Gannon DE Mitchell JJ Low RB Leslie KO Myofibroblasts in diffuse alveolar damage of the lung Mod Pathol 1998 11 1064 1070 9831203
Selman M King TE Pardo A Idiopathic pulmonary fibrosis: prevailing and evolving hypotheses about its pathogenesis and implications for therapy Ann Intern Med 2001 134 136 151 11177318
Flaherty KR Colby TV Travis WD Toews GB Mumford J Murray S Thannickal VJ Kazerooni EA Gross BH Lynch JP 3rdMartinez FJ Fibroblastic foci in usual interstitial pneumonia: idiopathic versus collagen vascular disease Am J Respir Crit Care Med 2003 167 1410 1415 12615630 10.1164/rccm.200204-373OC
Bjoraker JA Ryu JH Edwin MK Myers JL Tazelaar HD Schroeder DR Offord KP Prognostic significance of histopathologic subsets in idiopathic pulmonary fibrosis Am J Respir Crit Care Med 1998 157 199 203 9445300
Travis WD Matsui K Moss J Ferrans VJ Idiopathic nonspecific interstitial pneumonia: prognostic significance of cellular and fibrosing patterns: survival comparison with usual interstitial pneumonia and desquamative interstitial pneumonia Am J Surg Pathol 2000 24 19 33 10632484 10.1097/00000478-200001000-00003
Nicholson AG Colby TV du Bois RM Hansell DM Wells AU. The prognostic significance of the histologic pattern of interstitial pneumonia in patients presenting with the clinical entity of cryptogenic fibrosing alveolitis Am J Respir Crit Care Med 2000 162 2213 2217 11112140
Hyde DM King TE JrMcDermott T Waldron JA JrColby TV Thurlbeck WM Flint WM Ackerson L Cherniack RM Idiopathic pulmonary fibrosis. Quantitative assessment of lung pathology. Comparison of a semiquantitative and a morphometric histopathologic scoring system Am Rev Respir Dis 1992 146 1042 1047 1416394
Kitaichi M Pathologic features and the classification of interstitial pneumonia of unknown etiology Bull Chest Dis Res Inst Kyoto Univ 1990 23 1 18 2134041
Katzenstein AL Fiorelli RF Nonspecific interstitial pneumonia/fibrosis. Histologic features and clinical significance Am J Surg Pathol 1994 18 136 147 8291652
Daniil ZD Gilchrist FC Nicholson AG Hansell DM Harris J Colby TV du Bois RM A histologic pattern of nonspecific interstitial pneumonia is associated with a better prognosis than usual interstitial pneumonia in patients with cryptogenic fibrosing alveolitis Am J Respir Crit Care Med 1999 160 899 905 10471616
Flaherty KR Toews GB Travis WD Colby TV Kazerooni EA Gross BH Jain A Strawderman RL 3rdPaine R Flint A Lynch JP 3rdMartinez FJ Clinical significance of histological classification of idiopathic interstitial pneumonia Eur Respir J 2002 19 275 283 11866008 10.1183/09031936.02.00182002
American Thoracic Society/European Respiratory Society International Multidisciplinary Consensus Classification of the Idiopathic Interstitial Pneumonias Am J Respir Crit Care Med 2002 165 277 304 11790668
Katzenstein AL Myers JL Idiopathic pulmonary fibrosis: clinical relevance of pathologic classification Am J Respir Crit Care Med 1998 157 1301 1315 9563754
Cooper JA Jr Pulmonary fibrosis: pathways are slowly coming into light Am J Respir Cell Mol Biol 2000 22 520 523 10783121
King TE JrSchwarz MI Brown K Tooze JA Colby TV Waldron JA JrFlint A Thurlbeck W Cherniack RM Idiopathic pulmonary fibrosis: relationship between histopathologic features and mortality Am J Respir Crit Care Med 2001 164 1025 32 11587991
Nicholson AG Fulford LG Colby TV du Bois RM Hansell DM Wells AU The relationship between individual histologic features and disease progression in idiopathic pulmonary fibrosis Am J Respir Crit Care Med 2002 166 173 177 12119229 10.1164/rccm.2109039
Adamson IY Young L Bowden DH Relationship of alveolar epithelial injury and repair to the induction of pulmonary fibrosis Am J Pathol 1988 130 377 383 3341452
Haschek WM Witschi H Pulmonary fibrosis – a possible mechanism Toxicol Appl Pharmacol 1979 51 475 487 538759 10.1016/0041-008X(79)90372-7
Zeisberg M Bonner G Maeshima Y Colorado P Muller GA Strutz F Kalluri R Renal fibrosis: collagen composition and assembly regulates epithelial-mesenchymal transdifferentiation Am J Pathol 2001 159 1313 1321 11583959
Ng YY Huang TP Yang WC Chen ZP Yang AH Mu W Nikolic-Paterson DJ Atkins RC Lan HY Tubular epithelial-myofibroblast transdifferentiation in progressive tubulointerstitial fibrosis in 5/6 nephrectomized rats Kidney Int 1998 54 864 876 9734611 10.1046/j.1523-1755.1998.00076.x
Iwano M Plieth D Danoff TM Xue C Okada H Neilson EG Evidence that fibroblasts derive from epithelium during tissue fibrosis J Clin Invest 2002 110 341 350 12163453 10.1172/JCI200215518
|
15955241
|
PMC1200429
|
CC BY
|
2021-01-04 16:23:25
|
no
|
Respir Res. 2005 Jun 14; 6(1):57
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-57
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-591596702610.1186/1465-9921-6-59ReviewATP-binding cassette (ABC) transporters in normal and pathological lung van der Deen Margaretha [email protected] Vries Elisabeth GE [email protected] Wim [email protected] Rik J [email protected] Hetty [email protected] Dirkje S [email protected] University Medical Center Groningen, Department of Internal Medicine, Medical Oncology, Hanzeplein 1, 9713 GZ, Groningen, The Netherlands2 Department of Pulmonary Medicine, Hanzeplein 1, 9713 GZ, Groningen, The Netherlands3 Department of Pathology and Laboratory Medicine, Hanzeplein 1, 9713 GZ, Groningen, The Netherlands4 Free University, Department of Pathology, Boelelaan 1117, 1081 HV, Amsterdam, The Netherlands2005 20 6 2005 6 1 59 59 4 4 2005 20 6 2005 Copyright © 2005 van der Deen et al; licensee BioMed Central Ltd.2005van der Deen et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ATP-binding cassette (ABC) transporters are a family of transmembrane proteins that can transport a wide variety of substrates across biological membranes in an energy-dependent manner. Many ABC transporters such as P-glycoprotein (P-gp), multidrug resistance-associated protein 1 (MRP1) and breast cancer resistance protein (BCRP) are highly expressed in bronchial epithelium. This review aims to give new insights in the possible functions of ABC molecules in the lung in view of their expression in different cell types. Furthermore, their role in protection against noxious compounds, e.g. air pollutants and cigarette smoke components, will be discussed as well as the (mal)function in normal and pathological lung. Several pulmonary drugs are substrates for ABC transporters and therefore, the delivery of these drugs to the site of action may be highly dependent on the presence and activity of many ABC transporters in several cell types. Three ABC transporters are known to play an important role in lung functioning. Mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) gene can cause cystic fibrosis, and mutations in ABCA1 and ABCA3 are responsible for respectively Tangier disease and fatal surfactant deficiency. The role of altered function of ABC transporters in highly prevalent pulmonary diseases such as asthma or chronic obstructive pulmonary disease (COPD) have hardly been investigated so far. We especially focused on polymorphisms, knock-out mice models and in vitro results of pulmonary research. Insight in the function of ABC transporters in the lung may open new ways to facilitate treatment of lung diseases.
==== Body
Introduction
The prime role of the airways (trachea, bronchi, bronchioles and terminal bronchioles) is to conduct air into and out of the lung and to form a first line of defence against undesired constituents of inhaled air. The airways are continuously exposed to pathogens, irritants, pollutants and agents that produce oxidative stress and therefore, the composition of the respiratory tract surface is very important. The upper airways contain specialised cell types such as ciliated cells and mucous secreting goblet cells. The lower conducting airways (respiratory bronchioles, alveolar ducts and alveolar sacs) participate in gas exchange by diffusion. The alveolar epithelial surface comprises essentially two cell types, the alveolar epithelial type I cell and the cuboidal alveolar epithelial type II cell. Type I cells flatten out and in this way constitute approximately 95% of the total alveolar surface, whereas type II cells are more numerous and produce surfactant [1].
ABC (ATP-binding cassette) transporters are a family of transmembrane proteins that can transport a wide variety of substrates across biological membranes in an energy-dependent manner. They are phylogenetically classified in seven distinct subfamilies of transporters (ABCA to ABCG), which are again divided into subgroups. To date, 48 ABC transporters have been detected in the human body [2,3]. Overexpression of ABC transporters such as P-glycoprotein (P-gp) and the multidrug resistance-associated protein 1 (MRP1) were initially detected in tumour cell lines. Their overexpression is associated with increased efflux of chemotherapeutic drugs such as anthracyclines, epipodophyllotoxins and vinca-alkaloids, and this can result in so-called multidrug resistance (MDR). Many MDR proteins can act as drug efflux pumps, resulting in decreased intracellular concentrations of toxic compounds at the site of action. MRP1 and P-gp expression in the lung have especially been studied in the context of small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC). Currently we know that ABC transporters are present in virtually every cell of all species and play central roles in physiology. The prominent expression of P-gp and MRP1 in the human lung [4] suggests that these transporters may be pivotal in the protection against endogenous or exogenous toxic compounds entering the lung.The delivery of pulmonary drugs to reach the site of action may also depend on the presence and activity of many ABC transporters [5]. Langmann et al. developed quantitative real-time RT-PCR expression profiling of 47 ABC transporters in 20 different human tissues. Tissues with a barrier function such as lung and trachea were identified to have high transcriptional activity for many ABC transporters [6]. There are already clear proofs of important functions of ABC transporters in the lung. The ABC transporter most widely investigated is the cystic fibrosis transmembrane conductance regulator (CFTR), because mutations in this gene are responsible for the development of cystic fibrosis [7]. Mutations in ABCA1 are causative for Tangier disease [8]. In newborns it was shown that mutations in ABCA3 cause aberrant production of surfactant which can be lethal [9]. The role of altered function of ABC transporters in highly prevalent pulmonary diseases such as asthma or chronic obstructive pulmonary disease (COPD) have hardly been investigated so far [10].
This review aims to give new insights in the possible functions of ABC molecules in the lung in view of their expression in different cell types. Furthermore, their roles in protection against noxious compounds will be discussed as well as the (mal)function in normal and pathological lung. We especially focused on the members of the ABCC subfamily of transporters (to which the MRPs and CFTR belong), the ABCB subfamily (a.o. MDR1/Pgp and MDR3/Pgp) and the ABCA subfamily (a.o. ABCA1 and ABCA3), because these are a selection of the best characterised human transporters and because they have been investigated in human lungs. Results of in vitro studies in pulmonary research are being reviewed and an update will be given about what is known about pulmonary functional changes in ABC transporter knock-out mice models. In addition, the (potential) effects of polymorphisms in ABC transporters on lung functioning are discussed.
MDR1/P-gp
MDR1 localisation and function
The MDR1 (ABCB1) gene is located on chromosome 7q21.12 and encodes for P-gp. P-gp plays a role in cell defence against environmental attacks such as generated by xenobiotics. It transports many hydrophobic substrates and anti-cancer drugs including etoposide, doxorubicin and vinblastine (for review, see [11]) and is mainly apically expressed in organs involved in excretion such as liver and intestine. In the lung, P-gp is expressed at the apical side of ciliated epithelial cells or ciliated collecting ducts, and on apical and lateral surfaces of serous cells of bronchial glands but not in mucus-secreting goblet cells (Figure 1 and 2) (Table 1) [12]. Epithelial cells of the trachea and major bronchi stain strongly while staining of the smaller bronchi is patchy or absent. In addition, P-gp is present in the lateral membranes of normal nasal respiratory mucosa [13]. In human and rat type I alveolar epithelium, P-gp is located at the lumenal side whereas freshly isolated type II cells lack P-gp [14]. In another study, pneumocytes did not stain for P-gp [15]. Some antibodies visualised P-gp in endothelial cells of blood vessels (Figure 1) [12,15,16]. In addition, alveolar and peripheral blood monocyte-derived macrophages stained positive but variably for P-gp [4,16,17]. The observed apical epithelial expression may signify that P-gp plays a role in transport of compounds from the interstitium into the lumen. However, the precise function of P-gp in the lung is as yet unknown. Interestingly, P-gp was found to play a role in the regulation of cell volume activated chloride channels and to possess channel activities [18,19]. However, Cl- and K+ conductances were not affected by the level of P-gp expression and the physiological role of P-gp in volume-activated chloride currents is still unclear [20-22]. Many lipophilic amines, such as fentanyl, highly accumulate in the lung. It was suggested that P-gp plays a role in the disposition of these pulmonary amine drugs [23]. Cellular uptake in lung microvascular endothelial cells of fentanyl was inhibited by the P-gp substrate verapamil, but not by the P-gp blocking antibody UIC2. The authors suggested that this results from an inhibition of fentanyl uptake. This is however unlikely, since P-gp is primarily responsible for removal of drugs out of cells.
Figure 1 Expression of ATP-binding cassette proteins in several cell types of human lung. ABC, ATP-binding cassette; BM, basement membrane; BCRP, breast cancer resistance protein; CFTR, cystic fibrosis transmembrane conductance regulator; MRP, multidrug resistance-associated protein; P-gp, P-glycoprotein. *Conflicting results exist in literature.
Figure 2 Immunohistochemical staining of ATP-binding cassette proteins in human lung. A. apical expression of P-gp in bronchial epithelium (COPD patient; frozen section, antibody C219), B. basolateral expression of MRP1 in bronchial epithelium (COPD patient; antibody MRPr1), C. MRP1 expression in bronchoalveolar lavage cells (healthy individual; antibody MRPr1). Lu, lumen. Scale bar = 25 μM.
Table 1 Summary of features of ABC transporters in human lung.
ABC gene Gene location Functional role or substrates Linked to disease Protein expression in lung cells
(+ localisation$) References#
MDR1/P-gp (ABCB1) 7q21.12 Drug resistance, Hydrophobic organic cations Unknown Bronchial epithelium (ap), mucinous glands (ap), alveolar type I cells* (ap), endothelium* (ap), alv. macroph. [4,12,14-16]
MDR3/P-gp (ABCB4) 7q21.12 Phosphatidyl choline Unknown Absent [4]
MRP1 (ABCC1) 16p13.12 Drug resistance, Organic anions (e.g. GSH conjugates, LTC4) COPD? Bronchial epithelium (ba/la), goblet cells (ba), peripheral epithelial cells* (ba/la), seromucinous glands (ba/la), alv. macroph. [4,53,85]
MRP2 (ABCC2) 10q24.2 Drug resistance, Organic anions Dubin-Johnson syndrome Bronchial epithelium* (ap), primary bronchial and peripheral epithelial cells* (in) [4,85,90]
MRP3 (ABCC3) 17q21.33 Drug resistance, Organic anions Unknown Primary bronchial and peripheral epithelial cells* (ba/la) [85]
MRP4 (ABCC4) 13q32.1 Nucleoside analogues, Prostaglandin E1, E2 Unknown Primary bronchial and peripheral epithelial cells* (in) [85]
MRP5 (ABCC5) 3q27.1 Nucleoside analogues, Hyaluron Unknown Primary bronchial and peripheral epithelial cells* (in) [85]
MRP6 (ABCC6) 16p13.12 Unknown Pseudoxanthoma elasticum Unknown
MRP7 (ABCC10) 6p21.1 Drug resistance Unknown Unknown
MRP8 (ABCC11) 16q12.1 Conjugated steroids, Nucleoside analogues, bile acids Unknown Unknown
MRP9 (ABCC12) 16q12.1 Unknown Unknown Unknown
CFTR (ABCC7) 7q31.31 Chloride ion channel Cystic fibrosis Bronchial epithelium (ap), seromucinous glands (ap), Clara cells*, Alv. type I cells* [117-120]
BCRP (ABCG2) 4q22 Drug resistance, Protection food toxins Unknown Bronchial epithelium (ba/la), endothelium, seromucinous glands [4]
ABCA1 9q31.1 Cholesterol and phospholipids Tangier disease Alv. type II cells [151]
ABCA3 16p13.3 Surfactant secretion Surfactant deficiency Alv. type II cells [155]
$Cellular localisation: ap, apical; ba, basal; la, lateral; in, intracellular; alv., alveolar; macroph., macrophages; #References are mentioned that demonstrate protein localisation histologically; *Conflicting results exist in literature.
P-gp mRNA expression was not increased in smokers (n = 11) compared to ex-and non-smokers (n = 7) [12]. Whether P-gp expression levels may play a defensive role towards tobacco-derived agents remains to be investigated.
MDR1 in tumours
High P-gp expression can imply chemotherapeutic resistance due to increased chemotherapeutic drug efflux. In cancer therapy, many attempts have been made to reverse MDR mechanisms. However, in a randomised double-blind trial in 130 SCLC patients no positive effects were seen with the P-gp modulator megestrol acetate in addition to chemotherapeutic drugs, suggesting that levels of P-gp expression in lung tumours were not relevant or that modulation of P-gp activity was not complete in this treatment [24]. Some studies show higher P-gp expression at the invasion front of lung tumours and it was suggested that P-gp expression augments invasion properties of tumour cells [25]. Only two out of 22 NSCLC samples (both adenocarcinomas) stained positive with three P-gp antibodies [15] and no P-gp was detected on pulmonary carcinoids. Other studies revealed a relation between P-gp and glutathione S-transferase-pi (GST-pi) expression in NSCLC that were exposed in vitro to doxorubicin [26], suggesting that these two factors play a role in doxorubicin resistance. There was also a correlation between current smoking and doxorubicin resistance of NSCLC. Forty-two out of 72 NSCLC smokers expressed P-gp, whereas only two out of 22 tumours of non-smokers were P-gp positive [27].
MDR1 polymorphisms
MDR1 polymorphisms were first described by Hoffmeyer et al. [28] who found a correlation between lower intestinal expression of P-gp and a polymorphism in exon 26. Many single nucleotide polymorphisms (SNPs) have been recognised in the MDR1 gene (see reference [29] for recent review about clinical aspects). The impact of these polymorphisms on lung diseases is still speculative. It was proposed that polymorphisms in the MDR1 gene may have clinical consequences in patients with cystic fibrosis, since MDR1 plays a role in CFTR regulation. Rodents contain two Mdr1 genes, denoted as Mdr1a and Mdr1b. It was shown that Mdr1b mRNA expression in lung parenchyma of outbred rats is very variable and this may also be the case in humans [30]. The possible effects of MDR1 polymorphisms was studied in tobacco-related lung cancer [31]. No clear association was found between the T/T genotype of the C3435T polymorphism and susceptibility to lung cancer in a group of 268 Caucasian men who were current smokers. No relation was found between SNP C3435T in MDR1 and survival in 62 docetaxel-cisplatin-treated NSCLC patients [32]. Immunosuppressive agents such as cyclosporin A and tacrolimus (both calcineurin antagonists) are P-gp substrates. No relation was found of MDR1 G2677T and C3435T genotypes with tacrolimus blood levels in 83 lung transplant patients treated with tacrolimus [33]. Altogether, these data implicate that there is still no clear association between MDR1 polymorphisms and effects on outcome of treatment of lung cancer or lung transplant patients.
MDR1 in animal models
Scheffer et al. detected high P-gp levels in lungs of mice [4]. In rats, Mdr1a and Mdr1b mRNA expression were highest in the ileum [34]. The Mdr1a expression level in rat lung was 2% of the expression in ileum and expression of Mdr1b was 47% of that in ileum. In mice orally treated with dexamethasone for 24 hours, Mdr1b mRNA expression in lungs was decreased, from which the authors deduce that dexamethasone treatment of lung tumours may reverse MDR [35]. To study the in vivo distribution of P-gp, nude rats were injected with a P-gp overexpressing SCLC cell line (GLC4/Pgp) and with a P-gp negative cell line (GLC4) [36]. P-gp function was visualised with radiolabeled P-gp substrate [11C]verapamil by positron emission tomography (PET) with or without P-gp modulator cyclosporin A. The accumulation of [11C]verapamil was significantly increased by cyclosporin A in brains and GLC4/Pgp tumours in these rats. In all other investigated organs including lungs, the accumulation after cyclosporin A treatment was unaltered. In intact rabbit lung, vascular P-gp kinetics were measured in vivo using the lipophilic amine dye rhodamine 6G (R6G) by measuring R6G in the perfusate during circulation [37]. Inhibition of P-gp function with verapamil or GF120918 resulted in higher accumulation of R6G in lung. It was proposed that the opposite would happen when epithelial P-gp was inhibited because R6G would then be retained in the airspace. We propose another possibility that inhibition of epithelial P-gp will also result in higher R6G accumulation. In that case, R6G transport to the lumen is inhibited and as a compensation mechanism it may be transported back to the interstitial side where it either may be retained in the tissue or subsequently be transported into the circulation. This model could be useful in testing a large variety of pulmonary therapeutic agents, such as corticosteroids and sympathicomimetics that may be substrates for transporters in the lung or modulate their activity. Similar studies were also carried out in perfused rat liver to assess the effect of P-gp modulators on the hepatobiliary system, supporting the usefulness of this approach [38].
Mdr1a/1b (-/-) mice lack both genes encoding for P-gp and these mice seem physiologically normal. The penetration of [3H]digoxin (a P-gp substrate) was higher in brain, ovary and adrenal gland. In the lung tissue of these (-/-) mice [39], the level of [3H]digoxin was rather low compared to other organs. The level was 2.6 times higher in (-/-) mice than in (+/+) mice, but this was not significant. The pharmacokinetics of the central nervous system (CNS) drug amitryptiline (a P-gp substrate) and its metabolites were examined in Mdr1a/1b (-/-) mice by high performance liquid chromatography (HPLC) in several organs. Higher concentrations of these metabolites were measured in the brain of these mice, but not in other examined organs including the lung [40]. Thus, P-gp likely plays a more active role in exporting CNS drugs out of the brain than out of the lung. Another possibility is that other transporters in the lung were capable to efflux these drugs as a compensation mechanism.
MDR1 in vitro
Several lung cell culture models have been described for determining P-gp expression and functionality. The immortalised human bronchial epithelial cell line 16HBE14o- resembles primary epithelium and was reported to be suitable for drug metabolism studies [41]. In 16HBE14o- cells, P-gp was expressed at the apical side. Its functional activity was measured with P-gp substrate rhodamine 123 and its transport was inhibited by verapamil [42]. Two lung cell lines, Calu-3 and A549 cells, were compared for P-gp expression and functionality [43]. The bronchiolar adenocarcinoma cell line Calu-3 is suitable for drug transport studies because these cells form tight junctions in culture, which is not the case for the type II alveolar carcinoma cell line A549. P-gp expression was higher in A549 cells than in Calu-3 cells. However, efflux of rhodamine 123 was higher in Calu-3 cells. This may be explained by additional MRP1 activity in these cells because rhodamine 123 is also an MRP1 substrate [44]. In primary rat alveolar type II cells, Mdr1b mRNA levels increased in a time dependent manner in cultures at day 1, 2 and 3 compared to freshly isolated cells. Mdr1b mRNA was present at low levels and increased after oxygen radical induction with paraquat [45]. P-gp expression was below the detection limit in these cells at time of isolation. In freshly isolated primary human bronchial epithelial cells, P-gp was present and increased after 24 hours paraquat exposure. Rhodamine 123 efflux could be measured in these cells, confirming functional activity of P-gp [45]. These results demonstrate that P-gp is upregulated during stress, both from radical production and from ex vivo culturing or differentiation.
MDR3/P-gp
The MDR3/P-gp (ABCB4) gene maps closely to MDR1/P-gp on chromosome 7q21.12 and has high homology with MDR1 although its function is very different. It is involved in phosphatidyl choline transport from the liver into the bile. Its RNA and protein have not been detected in human and mouse lung and trachea [4,6]. and it is therefore not expected that MDR3 is present in the lungs.
MRP1
MRP1 localisation and function
The MRP1 (ABCC1) gene is located on chromosome 16p13.12. Cole et al. discovered in 1992 a non-Pgp mediated MDR mechanism in the human lung cancer cell line H69AR [46]. MRP1 was overexpressed in these cells and it was found that glutathione-, glucuronide-, and sulfate-conjugated organic anions are substrates for MRP1 [47,48]. MRP1 confers resistance to several chemotherapeutic agents including vincristine, daunorubicin and methotrexate [49-51]. Physiological substrates for MRP1 are e.g. leukotriene C4 (LTC4) and glutathione disulfide (GSSG) [52]. Interestingly, these substrates play an important role in lung physiology with respect to inflammation and oxidative stress. MRP1 is highly expressed mainly at the basolateral side of human bronchial epithelial cells (Figure 1 and 2) [53]. Ciliated and basal cells have been collected from brushes of main or lobar bronchi. Basal cells stained strongly on the entire circumference of the plasma membrane. Ciliated epithelial cells and mucous cells stained positive at the basolateral membrane but not at the apical membrane. No intracytoplasmic staining was observed. However, strong apical cytoplasmic staining has been detected below the cilia in respiratory columnar epithelial cells in paraffin sections with antibody MRPr1 [4,13,54]. The discrepancy between these findings may be due to different fixation procedures. Basal cells of seromucinous glands of the lungs also stain positive for MRP1 with a higher intensity in the serous area than mucinous cells. Alveolar macrophages are MRP1 positive with variable staining between individual lung samples (Figure 2) [4]. Kool et al. investigated MRP1, 2, 3, 4, and 5 with an RNase protection assay in total lung RNA and found high MRP1, absence of MRP2, low MRP3 and MRP4 and moderate MRP5 gene transcripts [55]. Total RNA was obtained from lung tissue collected during surgery or autopsy. Therefore, the original cell types can not be distinguished in these samples.
Possibly, the high MRP1 expression at the basolateral side of lung epithelium may assist in the clearance of toxins coming from the luminal or interstitial side (back) into the interstitial fluid [56]. Function of MRP1 expression in the lung epithelium, glands and alveolar macrophages may include extrusion of toxic intracellular substances, antioxidant defence or production of LTC4 as an inflammatory response. Given the fact that MRP1 expression is higher in the lung compared to other solid organs, decreased or increased functional MRP1 expression may have a high impact on development and/or progression of lung diseases and protection against air pollution and inhaled toxic substances such as present in cigarette smoke [10].
MRP1 in tumours
Using mRNA in situ hybridisation Thomas et al. [57] detected high MRP1 expression in normal epithelium whereas the major component of the tumour epithelium showed a negative hybridisation signal. However, MRP1 transcript expression at the invasive front of the lung tumours was consistently stronger and particularly strong in areas with lymphatic or blood vessels. In addition, endothelial cells and lymphocytic infiltrates stained strongly positive for MRP1. These results may indicate a role of MRP1 in invasion or in mitotic activity. In a recent study, all 102 NSCLC tumours were MRP1 positive. In addition, the level of expression was 3-fold higher in DNA-aneuploid cells compared to normal bronchial and carcinomatous DNA-diploid cells [58]. This was associated with more frequent gain of chromosome 16 where the MRP1 gene is located. Overexpression of MRP1 may therefore be an important factor of intrinsic resistance to chemotherapy in NSCLC. High MRP1 expression can also result in increased apoptosis as shown by experiments with verapamil in MRP1 transfected baby hamster kidney-21 (BHK-21) cells [59]. Verapamil acts as an apoptogen in these cells when compared to MRP1 negative cells or mutant MRP1 transfected control cells. This was accompanied by depletion of intracellular GSH due to transport of GSH by MRP1. Indeed, addition of extracellular GSH prevented cell death. This mechanism may be valuable for treatment of MRP1-positive tumours.
MRP1 in non-malignant diseases
Cigarette smoking is the principle risk factor for the development of COPD. Our preliminary results indicate that MRP1 expression is diminished in bronchial epithelium of COPD patients (ex-smokers) and that lower expression is related to worse lung function [10]. In addition, bronchial MRP1 expression was higher during smoking than after one year of smoking cessation (Van der Deen et al., submitted). The expression of MRP1 measured with RT-PCR was similar in lungs of smokers and a combined group of ex- and non-smokers, suggesting that current smoking does not affect MRP1 gene expression [53]. This was semi-quantitatively analysed and the power of this study was rather low (smokers n = 13 and ex- or non-smokers n = 8). Cigarette smoke extract is a complex mixture of many substances. One of these is nitrosamine 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK), a carcinogenic nitrosamine. NNK is converted intracellularly in NNAL-O-glucuronide which is transported by MRP1 in the presence of glutathione (GSH) [60]. Thus, functional activity of MRP1 in the lung may play an important role in the antioxidant defence against toxic compounds generated by cigarette smoke.
MRP1 polymorphisms
Polymorphisms of the MRP1 gene or in regulatory genes may influence function of MRP1 in the lung. MRP1 has been screened for genetic variations and several mutations have been found in the MRP1 gene in the human population [61,62]. Ethnic differences for MRP1 expression were observed between Caucasian and Japanese subjects but the clinical consequences have to be determined to date [63,64]. In preclinical models, a single mutation in the MRP1 gene can result in loss of transport of some, but not all, MRP1 substrates [65] which implies that potentially minor differences in the MRP1 gene may result in aberrant functional capacities of the MRP1 protein. A low frequency (<1%) naturally occurring mutation in MRP1 was related to functional differences in organic anion transport and drug resistance [66]. Therefore, some individuals could be more susceptible to a set of xenobiotics than others.
MRP1 in animal models
Scheffer et al. detected, besides P-gp, high levels of Mrp1 in lungs of mice [4]. Mrp1 (-/-) mice develop normally [67]. Viability, fertility, and a range of histological, hematological, and serum-chemical parameters were similar in Mrp (+/+) and Mrp (-/-) mice. However, Mrp1 (-/-) mice are hypersensitive to exposure to several drugs e.g. etoposide, resulting in loss of body weight and death [56,67,68]. In Mrp1 deficient mice there were no gross abnormalities in lungs and other tissues after treatment with the chemotherapeutic drug etoposide-phosphate, but abnormal mucous production was seen around the mouth. Further examination of tissues showed that these mice suffered from oropharyngeal toxicity [56]. In Mrp1 (-/-) mice, GSH levels were elevated in e.g. lung, kidney, heart, testes and skeletal muscle. In organs that express little MRP1, such as the liver and small intestine, GSH levels were unchanged [67,69]. The increase in GSH was not related to increased levels of gamma-glutamylcysteine synthetase (γ-GCS), the rate-limiting enzyme for production of the tripeptide GSH. This suggests that MRP1 plays a role in GSH metabolism because MRP1 expression determines cellular GSH levels which is γ-GCS independent. In liver tissue, Mrp2 and Mrp5 mRNA levels were increased in Mrp1(-/-) mice compared to wild-type mice, probably due to a compensation mechanism. Mdr1a and Mrp3 levels were unchanged in these animals. Unfortunately, lung tissue was not examined [69]. Mrp1 (-/-) mice showed impaired inflammatory responses accompanied by a decreased LTC4 excretion by leukotriene producing cells [67,68]. The outgrowth of Mycobacterium tuberculosis was enhanced in Mrp1 (-/-) mice, but there was no difference in survival compared to wild-type mice [70]. Strikingly, Schultz et al. [71] observed that the survival of Mrp1 (-/-) mice inoculated with Streptococcus pneumoniae was better than of wild-type mice. This was accompanied by a lower LTC4 concentration but a higher LTB4 level in bronchoalveolar lavage fluid (BALF). Treatment with an LTB4 antagonist abolished the positive effect on survival rate. Altering MRP1 function may therefore be a target for further studies on treatment of pneumonia. Interestingly, LTB4 levels are also elevated in sputum and exhaled breath condensate of COPD patients [72,73]. This is thought to result from a higher amount of neutrophils present, yet also lower functional MRP1 expression may play a role, according to the observations in Mrp1 (-/-) mice [10,71]. The pulmonary and hepatic carcinogen aflatoxin B1 (a mycotoxin) and its GSH conjugate are MRP1 substrates in vitro [74]. MRP1 may therefore play an important role in detoxification of aflatoxin B1 in the lung. No difference in occurrence of lung tumours (and liver tumours) 12 months after an 8 week exposure to aflatoxin B1 was observed in Mrp1 (-/-) and wild-type mice [75]. This observation may be explained by differences in exposure time compared to humans (who may be chronically exposed for many years) or by redundancy of other ABC transporters in knock-out mice to export aflatoxin B1 and its conjugates.
NRF2 (nuclear factor-E2 p45-related factor), a transcription factor for many genes that play a role in antioxidant defence and detoxification processes, was recently identified as a transcription factor for MRP1 [76]. Interestingly, the onset of cigarette smoke-induced emphysema was earlier and the extent of emphysema was more severe in Nrf2 (-/-) mice as compared to wild-type mice [77]. The number of inflammatory cells, mainly consisting of macrophages, in bronchoalveolar lavage fluid and lung tissue was higher in Nrf2 (-/-) mice, accompanied by more extensive apoptosis of endothelial cells and alveolar type II cells. Forty-five Nrf2-dependent protective genes were induced by smoke exposure in wild-type mice compared to knock-out mice as measured with microarray analysis. Regretfully, Mrp1 was not present on this gene expression array. Similar smoke exposure experiments with Mrp1 (-/-) mice would be interesting to investigate if these mice are also more susceptible to develop emphysema. Three polymorphisms were identified in the promotor region of NRF2 in healthy individuals (n = 81) and in patients with COPD (n = 87) and sytemic lupus erythematosis(n = 51) (all Japanese individuals) [78]. There was no relation between these polymorphisms and risk of these diseases, but the power of this study was rather low.
Mdr1a/1b(-/-)/Mrp1(-/-) triple knockout (TKO) mice are indistinguishable from wild-type mice under normal physiological conditions [79,80]. Intravenous injection of [3H]etoposide in TKO mice resulted after 4 hours in a higher accumulation in brown adipose tissue, colon, urogenital tract, salivery gland and heart when compared to Mdr1a/1b (-/-) mice. The accumulation of [3H]etoposide in lung did not increase [79]. Accumulation of [3H]vincristine was 26-fold higher in lungs of weaning TKO mice compared to adult TKO mice, 4 hours after a single dose intraperitoneal injection [81]. In TKO compared to wild-type mice, [3H]vincristine was 82.7 and 5.8-fold increased in respectively weaning and adult mice after 4 hours in this experimental setting. Eight hours post-injection, these ratios were 33 and 51.1. Thus, accumulation of vincristine is both time and age-related. In another study, toxicity of the chemotherapeutic drugs vincristine and etoposide was elevated in TKO mice (respectively 128-fold and 3–5 fold) compared to wild-type mice. Lung tissue was not further examined [80].
MRP1 in vitro
MRP1 overexpression was first described in the human lung cancer cell line H69AR [46]. The human SCLC cell line GLC4/ADR, which displays multiple copies of the MRP1 gene, is often used to study MRP1 function [47]. The lung cell line Calu-3 (from bronchiolar adenocarcinoma/glandular origin) has been reported to express MRP1 basolaterally and has functional MRP1 activity. It may serve as a good model for in vitro analysis of transport activities [44]. Besides P-gp, MRP1 protein was present in freshly obtained primary human bronchial epithelial cells. Efflux of MRP1 substrate carboxy-dichlorofluorescein could be measured in these cells and was inhibited by MRP modulator MK571 [45]. In contrast to upregulation of P-gp, epithelial cells did not respond with MRP1 upregulation to paraquat and MRP1 levels were stable in time until 12 weeks, but highly increased after 18 weeks culturing. An interesting observation was made by Bandi et al. [82] who incubated the airway epithelial cell line Calu-1 with budesonide, an anti-asthma drug. MRP1 expression decreased after 7 and 14 days incubation with budesonide (10 μM) but not at day 1 and 4. MRP1 function was also diminished after 14 days incubation with increasing amounts of budesonide (1, 10 and 100 μM). It was therefore proposed that budesonide could be used as a chemosensitiser in lung tumours. The immortalised bronchial epithelial cell line 16HBE14o- expresses high MRP1 protein levels and activity [83]. MRP1 activity, but not P-gp, was blocked by incubation with cigarette smoke extract. Thus, the expression of MRP1 in human lungs may contribute to defence mechanisms of toxic inhaled substances present in cigarette smoke to protect against pulmonary diseases such as lung cancer or COPD. It was shown that benzo [a]pyrene (B [a]P), a polycyclic aromatic hydrocarbon that is a.o. present in cigarette smoke and car exhaust fuels, increased the efflux of monochlorobimane GSH conjugate (mBCl-SG, which is an MRP substrate) out of primary rabbit alveolar type II cells [84]. This was explained by an increase in GSH levels. MK571 reduced the efflux of mBCL-SG, suggesting that the transport was MRP-mediated. Surprisingly, the transport direction of this substrate was higher to the apical than to the basolateral compartment. This may imply that MRP1 (or other MRPs) is apically located in alveolar type II cells in contrast to the basolateral location that is usually reported for MRP1, or that another transporter is responsible for this unexpected result. MRP1-5 expression was studied in primary human lung cell cultures and in A549 cells [85]. MRP1 and MRP3 expression were membrane-associated whereas MRP2, 4, and 5 were located in intracellular structures in primary cells (both bronchial and peripheral epithelial cells). In A549 cells, all transporters were expressed in the cellular membrane. The fluorescent microscopic pictures, however, did not precisely show the exact cellular localisation and in several cases, the Golgi apparatus also seems to stain beside the cellular membrane.
MRP2
MRP2 localisation and function
MRP2 (ABCC2) is located on chromosome 10q24.2 and is also named the canalicular multispecific organic anion transporter (cMOAT) in the liver [86]. MRP2 and MRP1 share very similar substrate specificities. High affinity endogenous substrates for MRP2 include amphiphilic anions, such as LTC4 [87,88]. and bilirubin glucuronosides [89,90]. Data on the presence of MRP2 protein in the lung are conflicting. Only one out of three antibodies stained positive at the apical side in bronchial epithelium (Figure 1) [4,90]. If MRP2 is expressed apically, it could be involved in transport of noxious compounds into the lumen of the lung.
MRP2 in animal models
There are several rat strains that have a mutation in Mrp2/cMOAT and the phenotypes resemble that of the human Dubin-Johnson syndrome, a disease in which MRP2 is mutated. Examples are the GY/TR- rat [91] and the Eisai hyperbilirubinemic rat (EHBR) [92]. Subjects with the Dubin-Johnson syndrome and also Mrp2 deficient rats display jaundice and have impaired organic anion transport. Recently, it was shown that deficient Mrp2 function in EHBR rats was compensated by upregulation of Mrp3 in liver and kidney [93]. Mrp2 (-/-) mice have not been described. There are no reports in the literature of pulmonary malfunction in rats or humans with defects in the MRP2 gene.
MRP3
The MRP3 (ABCC3) gene is the closest MRP1 homologue [94] and is located on chromosome 17q21.33. It is involved in resistance against the anti-cancer drugs etoposide, teniposide and at higher concentrations also to methotrexate [95]. MRP3 is located in basolateral plasma membranes of liver, adrenals, pancreas, kidney, gut and gallbladder. The physiological function of MRP3 is still unknown but its cellular localisation and the information on substrates implies a role for MRP3 in transport of organic anions from the liver into the blood, especially when secretion into bile is being blocked [96]. MRP3 protein and RNA was not detected in bronchial epithelium [4,96]. However, data in literature are conflicting and preliminary unpublished results indicate that Mrp3 is present in mouse lung. As already mentioned in the discussion on MRP1, primary epithelial cells (of bronchial and peripheral origin) and A549 cells stain positive for MRP3 (Figure 1) [85]. MRP3 mRNA was measured by quantitative RT-PCR in normal lung and lung tumours. MRP3 transcript levels were related to exposure to vincristine, etoposide and platinum drugs in lung tumours [97,98].
MRP4 and MRP5
MRP4 (ABCC4) and MRP5 (ABCC5) genes map to 13q32.1 and 3q27.1 respectively. MRP4 protein is highly expressed in the kidney and prostate and MRP4 mRNA is present in the lung [55,99]. MRP5 mRNA is ubiquitously expressed, mainly in brain and skeletal muscle and also in the lung [6,55]. Primary epithelial cells (of bronchial and peripheral origin) and A549 cells stain positive for MRP4 and 5 (Figure 1) [85]. The substrate specificities of these two transporters differ from the other transporters of the ABCC family, i.e. they transport a variety of nucleoside analogues. The physiological role of both proteins is still unknown but they can serve as an efflux pump of the nucleosides cAMP and cGMP at low affinity, likely in a GSH independent manner [100-103]. Interestingly, AMP levels in lungs of asthma and COPD patients are elevated and therefore, MRP4 and MRP5 activity may be of clinical relevance in these diseases [104]. MRP4 can actively efflux prostaglandins E1 and E2 [105]. It was also suggested that MRP4 plays a role in the transport of conjugated steroids and bile acids [106], whereas MRP5 was reported to transport hyaluran out of cells [107]. In a recent study, a panel of 60 human cancer cell lines (the NCI-60) were screened with real-time RT-PCR for 48 human ABC transporters. Among these were several lung cancer cell lines. An association was found between MRP4 and MRP5 gene expression and resistance against platinum drugs in lung cancer [108-110].
MRP6
MRP6 (ABCC6) maps to chromosome 16p13.12 and is mainly expressed in liver and kidney. MRP6 was neither detected in the lung nor in lung derived tumour cell lines SW1573 and GLC4 by immunohistochemistry [111]. However, MRP6 mRNA was moderately present in human lung extracts [6]. MRP6 is mutated in the hereditary connective tissue disorder pseudoxanthoma elasticum which affects skin, retina and blood vessels (for review, see [112]). Still, its physiological role in this disease and its substrate specificity is unclear. Pulmonary abnormalities are rare in pseudoxanthoma elasticum. However, in some patients calcification and elastic tissue damage in the lung have been described [113]. This may be due to an altered MRP6 function. Further studies are required to assess whether dysfunction of MRP6 also plays a role in development of other lung diseases such as emphysema. With in situ hybridisation and RNase protection assay in C57BL/6 mice, Mrp6 mRNA could be detected in tracheal and bronchial epithelium [114].
MRP7, MRP8 and MRP9
To date, the functions of MRP7 (ABCC10, gene 6p21.1), MRP8 (ABCC11, gene 16q12.1) and MRP9 (ABCC12, gene 16q12.1) are largely unknown. Of these recently discovered ABC transporters, only MRP7 mRNA is highly expressed in total lung and trachea RNA extracts [6]. MRP7 function resembles P-gp function in the resistance against taxanes [115]. MRP8 resembles MRP4 function more than MRP5, and the physiological role of MRP8 may involve transport of conjugated steroids, cyclic nucleotides and bile acids [116].
CFTR
CFTR localisation and function
CFTR (ABCC7) is located on chromosome 7q31.31. CFTR is the only member of the ABC superfamily which is not an active transporter. It functions as a chloride channel and in normal human airway tissue CFTR is highly expressed at the luminal side in serous cells of the submucosal glands. In addition, it is restricted to the apical membrane domain of well-differentiated epithelial cells such as ciliated cells, and probably also non-ciliated Clara cells and alveolar type I cells (Figure 1) [117-120]. CFTR is also expressed in normal nasal respiratory mucosa [13].
CFTR in cystic fibrosis
Mutations in the CFTR gene can cause cystic fibrosis [7] and are associated with abnormal Cl- and Na+ ion transport in several tissues including the lungs, pancreas, gastrointestinal tract, liver, sweat glands and male reproductive organs. Although the normal expression of CFTR in the lung is lower compared to tissues such as the intestine and pancreas, its function in the lung is of major importance. The most frequent mutation in this gene is the delta F508 mutation which leads to cystic fibrosis. Defective CFTR function can cause viscous secretions in the lungs which leads to chronic inflammation with acute exacerbations by impaired mucociliary clearance. There are major risks for colonisation with Pseudonomas aeruginosa which leads to pneumonia and respiratory insufficiency [121]. Besides abnormal CFTR localisation and expression in cystic fibrosis, also in non-cystic fibrosis airway tissue CFTR can be abnormally expressed in remodelled or dedifferentiated epithelium [122,123]. whereas in delta F508 CFTR epithelial cells there may be a normally processed CFTR [120,124]. The regulation and transport function of CFTR are dependent of the state of differentiation and polarisation of epithelial cell cultures [125,126]. Dedifferentiation with hyperplasia or metaplasia was associated with an intracellular localisation or absence of CFTR protein [122]. Hurbain et al. [127] analysed MRP1-5 transcript levels in nasal respiratory cells from cystic fibrosis patients with homozygous delta F508 mutation. Surprisingly, low MRP1 levels were associated with more severe disease and in addition, MRP1 levels were related to cAMP-independent chloride transport suggesting that MRP1 regulates another chloride channel in the apical membrane. CFTR function is cAMP regulated [121], and therefore, MRP4 or MRP5 activity may play a role in regulation of CFTR, since these proteins transport cAMP. Another study showed that CFTR function is blocked by two MRP substrates (taurolithocholate-3-sulphate and beta-estradiol) [128]. This implies that these two substances are also substrates for CFTR, or that CFTR and MRP proteins possess similar anion binding sites. Bebok et al. showed that nitric oxide (NO) and reactive oxygen nitrogen species (RONS) decrease wild-type CFTR protein levels in airway epithelial cell monolayers [129]. Natural sources of NO and RONS are activated in alveolar and interstitial macrophages [130,131], neutrophils [132], alveolar type II cells [133,134]. and airway epithelial cells [135]. In this view, CFTR function may be important in pulmonary infections and in the effect of oxidative stress generated by e.g. cigarette smoke.
CFTR in animal models
Cftr knock-out mice show reduced viability in contrast to most other ABC transporter knock-outs which are viable and fertile. Trezise et al. developed Cftr knock-out mice that have a severe cystic fibrosis phenotype accompanied by a lack of Cftr-related chloride conductance in e.g. tracheal epithelium [136,137]. They found that the P-gp mRNA level was four-fold increased in intestines of neonatal and 3- to 4-week-old Cftr knock-out mice compared to littermates of the same age. However, in 10 weeks-old mice P-gp levels were three-fold decreased. Apparently, a reduction or loss of Cftr function influences P-gp expression.
CFTR in vitro
The Calu-3 cell line, which has properties from serous cells of the pulmonary submucosal glands, is often used to study function and expression of CFTR [129,138]. It was demonstrated in Calu-3 cells that CFTR mRNA expression is downregulated after ouabain incubation whereas P-gp expression was upregulated [139]. The bronchial epithelial cell line 16HBE14o- is also invaluable in CFTR research [129,140]. In wild-type and mutant CFTR expressing Sf9 insect cells, it was demonstrated that besides chloride transport, another function of CFTR is transport of GSH in a nucleotide-dependent manner [141]. This observation suggests that CFTR plays a role in the control of oxidative stress. Indeed, several studies in patients, mice and cell lines have shown that GSH levels are lower in case of defective CFTR function.
BCRP
BCRP localisation and function
BCRP (ABCG2) is the breast cancer resistance protein (BCRP), located on chromosome 4q22. BCRP is a half-transporter that probably acts as a homo- or heterodimer [142] and is involved in resistance against toxins and several chemotherapeutic agents (e.g. mitoxantrone and topoisomerase 1 inhibitors) [143,144]. Protein levels of BCRP in the lung are lower than P-gp and MRP1 but distinct in the epithelial cell layer and in seromucinous glands (Figure 1) [4]. BCRP was absent in alveolar macrophages, suggesting that BCRP does not play a major role in innate inflammatory responses in the lung. Small endothelial capillaries also stain positive for BCRP, thus BCRP may protect the lungs against noxious compounds that enter systemically.
BCRP in tumours
In a study of untreated solid lung tumours, i.e. squamous cell carcinoma (n = 5), adenocarcinoma (n = 2) and SCLC (n = 3), most cases expressed moderate or strong BCRP [145]. In addition, higher BCRP expression in blood vessels was observed than in vessels of the surrounding tumour, indicating that BCRP may play a role in tumour angiogenesis. In 72 cases of advanced NSCLC, BCRP expression, but not P-gp, MRP1, MRP2 and MRP3, predicted poor clinical outcome [146].
BCRP in animal models
Bcrp knock-out mice do not display any abnormalities compared to wild-type mice under normal conditions, except for the colour of their bile which is red instead of yellow on certain diets [147]. However, these mice appear to be extremely sensitive to the phototoxin pheophorbide A, showing that BCRP is important to protect against toxic food components. Abnormalities of lungs in Bcrp (-/-) mice have not been reported thus far. Recently, Bcrp1 (Bcrp in the mouse is also called Bcrp1) localisation in the lung of 4- to 8-week old C57Bl/6J mice was studied to identify lung stem cells. These cells possess high activity of efflux of Hoechst dye that is transported by Bcrp1 [148]. In peripheral blood and bone marrow, cells with side population activity represent BCRP positive (haematopoietic) stem cells and this might also be the case for side population cells in the lungs. Indeed, Bcrp1 (-/-) mice did not display side population cells as shown by Hoechst staining of lung digests. In paraffin-embedded tissue sections, Bcrp1 was restricted to smooth muscle cells (of both arteries and airways) and a subpopulation of unidentified round cells in the alveolar space but not detectable in endothelial or epithelial cells. BAL cells of mice were Bcrp1 positive, but did not display side population (SP) activity [148]. Clara cells, a nonciliated bronchiolar epithelial cell type, were also phenotypically described to possess stem cell characteristics [149]. Further investigation is required to identify the stem cell pool in the lung in relation to expression of ABC transporters. Repairing and replacing lost lung tissue is a research area of promising therapeutic possibilities [150].
ABCA1
The ABCA1 gene is mapped to chromosome 9q31.1. It is the causative gene in the development of Tangier disease, a disorder of cholesterol transport [8]. The ABCA1 protein controls transport of cholesterol and phospholipids to apolipoprotein 1 (apoA-1) in alveolar type II cells (Figure 1). Using the siRNA technique for ABCA1, it was shown that ABCA1 is involved in basolateral transport of surfactant that is activated by oxysterol [151]. Abca1 knock-out mice show increased concentrations of cholesterol precursors in lung, plasma, intestine and faeces [152]. It was demonstrated that there were major morphologic abnormalities in lungs of these mice, increasing with age. In 30% of 18 months old mice, lung parenchyma was affected. Lesions were characterised by foamy type II pneumocytes with aberrant lamellar bodies, intraalveolar macrophages and cholesterol clefts [153]. In addition, ABCA7, a close homologue of ABCA1 was also found to be highly expressed in mouse lung tissue by Western blot analysis [154].
ABCA3
The ABCA3 gene is located on chromosome 16p13.3. The function of this ABC transporter has not been studied in detail but ABCA3 protein is present in lamellar bodies in human lung alveolar type II cells (Figure 1). Langmann et al. performed quantitative RT-PCR in 20 human tissues and found ABCA3 to be expressed restrictively in the lung [6]. In a recent study in cell lines, an association was found between lung cancer and ABCA3 (and also ABCA2) gene expression by means of quantitative RT-PCR [108].
The high expression of ABCA3 in alveolar type II cells suggests that ABCA3 may play a role in surfactant regulation [155]. Surfactant is important to lower the surface tension in the air-liquid interphase in alveoli. Indeed, in patients with surfactant deficiency and with severe neonatal lung disease it was demonstrated that the ABCA3 gene was frequently mutated [9]. ABCA3 mutations were found in 16 out of 21 patients, but polymorphisms (SNPs) were not found. One mutation was not fatal but was associated with a chronic lung disorder in a 6-year old patient. Other mutations turned out to be fatal and these patients died shortly after birth. Electron micrographs of tissue of patients with surfactant deficiency showed abnormal dense and small lamellar bodies. It was suggested that ABCA3 plays a role in phospholipid metabolism, since it closely resembles ABCA1 and ABCA4 that are known to transport phospholipids.
Conclusion
Little is known about the function of ABC transporters that are expressed in the lung although their overall expression is very high compared to many other organs [6]. This review shows that ABC transporters in the lung are not only relevant for relatively rare diseases such as cystic fibrosis, Tangier disease and surfactant deficiency. Preliminary data indicate that MRP1 expression is lower in COPD patients than in healthy controls. Mutations and polymorphisms in ABC transporters may have important clinical consequences for development of lung diseases. However, overlap in substrate specificities may be compensatory in cases of malfunction of one (or more) transporter(s). Given the complexity of lung architecture, research on detailed cellular processes is difficult but challenging. Several ABC transporter deficient animal models have been developed that are of great value to study the role of these proteins. To date, exposure to cigarette smoke has never been tested in ABC transporter deficient animal models and would potentially give interesting information about the role of ABC transporters in protection against inhaled toxic substances such as present in tobacco smoke. Cell line models have been used to study transport processes and pulmonary drug metabolism. The delivery of pulmonary drugs to the site of action is probably highly dependent on the presence and activity of many ABC transporters in several cell types in the lung. The first barrier after inhalation is the pulmonary epithelium and transporters in the pulmonary endothelium may be critical for the delivery of intravenously or orally administered drugs. Insight in the function of ABC transporters in the lung may open new ways to facilitate treatment of lung diseases.
Abbreviations
ABC = ATP-binding cassette
AMP = adenosine monophosphate
BALF = bronchoalveolar lavage fluid
B [a]P = benzo [a]pyrene
BCRP = breast cancer resistance protein
cAMP = cyclic adenosine monophosphate
CFTR = cystic fibrosis transmembrane conductance regulator
cGMP = cyclic guanosine monophosphate
cMOAT = canalicular multispecific organic anion transporter
CNS = central nervous sytem
COPD = chronic obstructive pulmonary disease
EHBR = Eisai hyperbilirubinemic rat
γ-GCS = gamma-glutamylcysteine synthetase
GSH = glutathione, reduced form
GSSG = glutathione disulfide; oxidized glutathione
GST = glutathione S-transferase
HPLC = high liquid performance chromatography
LTC4 = leuktotriene C4
LTB4 = leukotriene B4
MDR = multidrug resistance
MRP = multidrug resistance-associated protein
NSCLC = non-small cell lung cancer
NNAL = nitrosamine 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol
NNK = nitrosamine 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone
NO = nitric oxide
NRF2 = nuclear factor-E2 p45-related factor
P-gp = P-glycoprotein
R6G = rhodamine 6G
RONS = reactive oxygen nitrogen species
RT-PCR = reverse transcriptase-polymerase chain reaction
SCLC = small cell lung cancer
SNP = single nucleotide polymorphism
TKO = triple knock-out mice
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MD mainly drafted the manuscript, EV, WT, RS, HT and DP helped to draft the manuscript. All authors read and approved the final manuscript.
Acknowledgements
This work was supported by a grant from the Netherlands Asthma Foundation (NAF97.35) and the "Stichting Astma Bestrijding" (SAB).
==== Refs
Crapo JD Barry BE Gehr P Bachofen M Weibel ER Cell number and cell characteristics of the normal human lung Am Rev Respir Dis 1982 125 740 745 7091881
Dean M Hamon Y Chimini G The human ATP-binding cassette (ABC) transporter superfamily J Lipid Res 2001 42 1007 1017 11441126
Muller M Human ABC-transporters
Scheffer GL Pijnenborg AC Smit EF Muller M Postma DS Timens W van der Valk P de Vries EG Scheper RJ Multidrug resistance related molecules in human and murine lung J Clin Pathol 2002 55 332 339 11986335
Hamilton KO Yazdanian MA Audus KL Contribution of efflux pump activity to the delivery of pulmonary therapeutics Curr Drug Metab 2002 3 1 12 11876574 10.2174/1389200023338170
Langmann T Mauerer R Zahn A Moehle C Probst M Stremmel W Schmitz G Real-time reverse transcription-PCR expression profiling of the complete human ATP-binding cassette transporter superfamily in various tissues Clin Chem 2003 49 230 238 12560344 10.1373/49.2.230
Riordan JR Rommens JM Kerem B Alon N Rozmahel R Grzelczak Z Zielenski J Lok S Plavsic N Chou JL Identification of the cystic fibrosis gene: cloning and characterization of complementary DNA Science 1989 245 1066 1073 2475911
Brooks-Wilson A Marcil M Clee SM Zhang LH Roomp K van Dam M Yu L Brewer C Collins JA Molhuizen HO Loubser O Ouelette BF Fichter K Ashbourne-Excoffon KJ Sensen CW Scherer S Mott S Denis M Martindale D Frohlich J Morgan K Koop B Pimstone S Kastelein JJ Hayden MR Mutations in ABC1 in Tangier disease and familial high-density lipoprotein deficiency Nat Genet 1999 22 336 345 10431236 10.1038/11905
Shulenin S Nogee LM Annilo T Wert SE Whitsett JA Dean M ABCA3 gene mutations in newborns with fatal surfactant deficiency N Engl J Med 2004 350 1296 1303 15044640 10.1056/NEJMoa032178
Van der Deen M Marks H Muller M Pijnenborg AC Postma DS Scheffer G Scheper RJ Smit EF Timens W de Vries EG Diminished expression of Multidrug Resistance-associated Protein 1 (MRP1) in bronchial epithelium of COPD patients [abstract] Am J Respir Crit Care Med 2003 167 A397
Gottesman MM Fojo T Bates SE Multidrug resistance in cancer: role of ATP-dependent transporters Nat Rev Cancer 2002 2 48 58 11902585 10.1038/nrc706
Lechapt-Zalcman E Hurbain I Lacave R Commo F Urban T Antoine M Milleron B Bernaudin JF MDR1-Pgp 170 expression in human bronchus Eur Respir J 1997 10 1837 1843 9272928 10.1183/09031936.97.10081837
Wioland MA Fleury-Feith J Corlieu P Commo F Monceaux G Lacau-St-Guily J Bernaudin JF CFTR, MDR1, and MRP1 immunolocalization in normal human nasal respiratory mucosa J Histochem Cytochem 2000 48 1215 1222 10950878
Campbell L Abulrob AN Kandalaft LE Plummer S Hollins AJ Gibbs A Gumbleton M Constitutive expression of p-glycoprotein in normal lung alveolar epithelium and functionality in primary alveolar epithelial cultures J Pharmacol Exp Ther 2003 304 441 452 12490621 10.1124/jpet.102.042994
Cordon-Cardo C O'Brien JP Boccia J Casals D Bertino JR Melamed MR Expression of the multidrug resistance gene product (P-glycoprotein) in human normal and tumor tissues J Histochem Cytochem 1990 38 1277 1287 1974900
Van der Valk P van Kalken CK Ketelaars H Broxterman HJ Scheffer G Kuiper CM Tsuruo T Lankelma J Meijer CJ Pinedo HM Distribution of multi-drug resistance-associated P-glycoprotein in normal and neoplastic human tissues. Analysis with 3 monoclonal antibodies recognizing different epitopes of the P-glycoprotein molecule Ann Oncol 1990 1 56 64 1706611
Puddu P Fais S Luciani F Gherardi G Dupuis ML Romagnoli G Ramoni C Cianfriglia M Gessani S Interferon-gamma up-regulates expression and activity of P-glycoprotein in human peripheral blood monocyte-derived macrophages Lab Invest 1999 79 1299 1309 10532592
Valverde MA Diaz M Sepulveda FV Gill DR Hyde SC Higgins CF Volume-regulated chloride channels associated with the human multidrug-resistance P-glycoprotein Nature 1992 355 830 833 1371598 10.1038/355830a0
Gill DR Hyde SC Higgins CF Valverde MA Mintenig GM Sepulveda FV Separation of drug transport and chloride channel functions of the human multidrug resistance P-glycoprotein Cell 1992 71 23 32 1382860 10.1016/0092-8674(92)90263-C
Higgins CF P-glycoprotein and cell volume-activated chloride channels J Bioenerg Biomembr 1995 27 63 70 7543087
Kunzelmann K Slotki IN Klein P Koslowsky T Ausiello DA Greger R Cabantchik ZI Effects of P-glycoprotein expression on cyclic AMP and volume-activated ion fluxes and conductances in HT-29 colon adenocarcinoma cells J Cell Physiol 1994 161 393 406 7962123 10.1002/jcp.1041610302
De Greef C Sehrer J Viana F van Acker K Eggermont J Mertens L Raeymaekers L Droogmans G Nilius B Volume-activated chloride currents are not correlated with P-glycoprotein expression Biochem J 1995 307 713 718 7741701
Waters CM Krejcie TC Avram MJ Facilitated uptake of fentanyl, but not alfentanil, by human pulmonary endothelial cells Anesthesiology 2000 93 825 831 10969317 10.1097/00000542-200009000-00033
Wood L Palmer M Hewitt J Urtasun R Bruera E Rapp E Thaell JF Results of a phase III, double-blind, placebo-controlled trial of megestrol acetate modulation of P-glycoprotein-mediated drug resistance in the first-line management of small-cell lung carcinoma Br J Cancer 1998 77 627 631 9484821
Beer TW Rowlands DC Crocker J Detection of the multidrug resistance marker P-glycoprotein by immunohistochemistry in malignant lung tumours Thorax 1996 51 526 529 8711682
Volm M Mattern J Samsel B Relationship of inherent resistance to doxorubicin, proliferative activity and expression of P-glycoprotein 170, and glutathione S-transferase-pi in human lung tumors Cancer 1992 70 764 769 1353700
Volm M Mattern J Samsel B Overexpression of P-glycoprotein and glutathione S-transferase-pi in resistant non-small cell lung carcinomas of smokers Br J Cancer 1991 64 700 704 1680367
Hoffmeyer S Burk O von Richter O Arnold HP Brockmoller J Johne A Cascorbi I Gerloff T Roots I Eichelbaum M Brinkmann U Functional polymorphisms of the human multidrug-resistance gene: multiple sequence variations and correlation of one allele with P-glycoprotein expression and activity in vivo Proc Natl Acad Sci U S A 2000 97 3473 3478 10716719 10.1073/pnas.050585397
Eichelbaum M Fromm MF Schwab M Clinical aspects of the MDR1 (ABCB1) gene polymorphism Ther Drug Monit 2004 26 180 185 15228162 10.1097/00007691-200404000-00017
Johannesson M Nordqvist AC Bogdanovic N Hjelte L Schalling M Polymorphic expression of multidrug resistance mRNA in lung parenchyma of nonpregnant and pregnant rats: a comparison to cystic fibrosis mRNA expression Biochem Biophys Res Commun 1997 239 606 611 9344878 10.1006/bbrc.1997.7519
Sinues B Fanlo A Bernal ML Mayayo E Bello S Rubio E Isla D MDR-1 C3435T genetic polymorphism and tobacco-related lung cancer Oncology 2003 64 183 185 12566917 10.1159/000067770
Isla D Sarries C Rosell R Alonso G Domine M Taron M Lopez-Vivanco G Camps C Botia M Nunez L Sanchez-Ronco M Sanchez JJ Lopez-Brea M Barneto I Paredes A Medina B Artal A Lianes P Single nucleotide polymorphisms and outcome in docetaxel-cisplatin-treated advanced non-small-cell lung cancer Ann Oncol 2004 15 1194 1203 15277258 10.1093/annonc/mdh319
Zheng H Zeevi A Schuetz E Lamba J McCurry K Griffith BP Webber S Ristich J Dauber J Iacono A Grgurich W Zaldonis D McDade K Zhang J Burckart GJ Tacrolimus dosing in adult lung transplant patients is related to cytochrome P4503A5 gene polymorphism J Clin Pharmacol 2004 44 135 140 14747421 10.1177/0091270003262108
Brady JM Cherrington NJ Hartley DP Buist SC Li N Klaassen CD Tissue distribution and chemical induction of multiple drug resistance genes in rats Drug Metab Dispos 2002 30 838 844 12065443 10.1124/dmd.30.7.838
Seree E Villard PH Hever A Guigal N Puyoou F Charvet B Point-Scomma H Lechevalier E Lacarelle B Barra Y Modulation of MDR1 and CYP3A expression by dexamethasone: evidence for an inverse regulation in adrenals Biochem Biophys Res Commun 1998 252 392 395 9826540 10.1006/bbrc.1998.9662
Hendrikse NH de Vries EG Eriks-Fluks L van der Graaf WT Hospers GA Willemsen AT Vaalburg W Franssen EJ A new in vivo method to study P-glycoprotein transport in tumors and the blood-brain barrier Cancer Res 1999 59 2411 2416 10344751
Roerig DL Audi SH Ahlf SB Kinetic characterization of p-glycoprotein-mediated efflux of rhodamine 6g in the intact rabbit lung Drug Metab Dispos 2004 32 953 958 15319336 10.1124/dmd.104.000042
Booth CL Brouwer KR Brouwer KL Effect of multidrug resistance modulators on the hepatobiliary disposition of doxorubicin in the isolated perfused rat liver Cancer Res 1998 58 3641 3648 9721873
Schinkel AH Mayer U Wagenaar E Mol CA van Deemter L Smit JJ van der Valk MA Voordouw AC Spits H van Tellingen O Zijlmans JM Fibbe WE Borst P Normal viability and altered pharmacokinetics in mice lacking mdr1-type (drug-transporting) P-glycoproteins Proc Natl Acad Sci U S A 1997 94 4028 4033 9108099 10.1073/pnas.94.8.4028
Grauer MT Uhr M P-glycoprotein reduces the ability of amitriptyline metabolites to cross the blood brain barrier in mice after a 10-day administration of amitriptyline J Psychopharmacol 2004 18 66 74 15107187 10.1177/0269881104042831
Forbes II Human airway epithelial cell lines for in vitro drug transport and metabolism studies Pharm Sci Technol Today 2000 3 18 27 10637597 10.1016/S1461-5347(99)00231-X
Ehrhardt C Kneuer C Laue M Schaefer UF Kim KJ Lehr CM 16HBE14o- human bronchial epithelial cell layers express P-glycoprotein, lung resistance-related protein, and caveolin-1 Pharm Res 2003 20 545 551 12739760 10.1023/A:1023230328687
Hamilton KO Yazdanian MA Audus KL Modulation of P-glycoprotein activity in Calu-3 cells using steroids and beta-ligands Int J Pharm 2001 228 171 179 11576779 10.1016/S0378-5173(01)00836-5
Hamilton KO Topp E Makagiansar I Siahaan T Yazdanian M Audus KL Multidrug resistance-associated protein-1 functional activity in Calu-3 cells J Pharmacol Exp Ther 2001 298 1199 1205 11504821
Lehmann T Kohler C Weidauer E Taege C Foth H Expression of MRP1 and related transporters in human lung cells in culture Toxicology 2001 167 59 72 11557130 10.1016/S0300-483X(01)00458-9
Cole SP Bhardwaj G Gerlach JH Mackie JE Grant CE Almquist KC Stewart AJ Kurz EU Duncan AM Deeley RG Overexpression of a transporter gene in a multidrug-resistant human lung cancer cell line Science 1992 258 1650 1654 1360704
Muller M Meijer C Zaman GJ Borst P Scheper RJ Mulder NH de Vries EG Jansen PL Overexpression of the gene encoding the multidrug resistance-associated protein results in increased ATP-dependent glutathione S-conjugate transport Proc Natl Acad Sci U S A 1994 91 13033 13037 7809167
Hipfner DR Deeley RG Cole SP Structural, mechanistic and clinical aspects of MRP1 Biochim Biophys Acta 1999 1461 359 376 10581367
Loe DW Deeley RG Cole SP Characterization of vincristine transport by the M(r) 190,000 multidrug resistance protein (MRP): evidence for cotransport with reduced glutathione Cancer Res 1998 58 5130 5136 9823323
Renes J de Vries EG Nienhuis EF Jansen PL Muller M ATP- and glutathione-dependent transport of chemotherapeutic drugs by the multidrug resistance protein MRP1 Br J Pharmacol 1999 126 681 688 10188979 10.1038/sj.bjp.0702360
Hooijberg JH Broxterman HJ Kool M Assaraf YG Peters GJ Noordhuis P Scheper RJ Borst P Pinedo HM Jansen G Antifolate resistance mediated by the multidrug resistance proteins MRP1 and MRP2 Cancer Res 1999 59 2532 2535 10363967
Leier I Jedlitschky G Buchholz U Keppler D Characterization of the ATP-dependent leukotriene C4 export carrier in mastocytoma cells Eur J Biochem 1994 220 599 606 8125120 10.1111/j.1432-1033.1994.tb18661.x
Brechot JM Hurbain I Fajac A Daty N Bernaudin JF Different pattern of MRP localization in ciliated and basal cells from human bronchial epithelium J Histochem Cytochem 1998 46 513 517 9524197
Flens MJ Zaman GJ van der Valk P Izquierdo MA Schroeijers AB Scheffer GL van der Groep P de Haas M Meijer CJ Scheper RJ Tissue distribution of the multidrug resistance protein Am J Pathol 1996 148 1237 1247 8644864
Kool M de Haas M Scheffer GL Scheper RJ van Eijk MJ Juijn JA Baas F Borst P Analysis of expression of cMOAT (MRP2), MRP3, MRP4, and MRP5, homologues of the multidrug resistance-associated protein gene (MRP1), in human cancer cell lines Cancer Res 1997 57 3537 3547 9270026
Wijnholds J Scheffer GL van der Valk M van der Valk P Beijnen JH Scheper RJ Borst P Multidrug resistance protein 1 protects the oropharyngeal mucosal layer and the testicular tubules against drug-induced damage J Exp Med 1998 188 797 808 9730882 10.1084/jem.188.5.797
Thomas GA Barrand MA Stewart S Rabbitts PH Williams ED Twentyman PR Expression of the multidrug resistance-associated protein (MRP) gene in human lung tumours and normal tissue as determined by in situ hybridisation Eur J Cancer 1994 30A 1705 1709 7833148 10.1016/0959-8049(94)00290-L
Doubre H Cesari D Mairovitz A Benac C Chantot-Bastaraud S Dagnon K Antoine M Danel C Bernaudin JF Fleury-Feith J Multidrug resistance-associated protein (MRP1) is overexpressed in DNA aneuploid carcinomatous cells in non-small cell lung cancer (NSCLC) Int J Cancer 2005 113 568 574 15472893 10.1002/ijc.20617
Trompier D Chang XB Barattin R du Moulinet d'Hardemare A Di Pietro A Baubichon-Cortay H Verapamil and its derivative trigger apoptosis through glutathione extrusion by multidrug resistance protein MRP1 Cancer Res 2004 64 4950 4956 15256468
Leslie EM Ito K Upadhyaya P Hecht SS Deeley RG Cole SP Transport of the beta -O-glucuronide conjugate of the tobacco-specific carcinogen 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol (NNAL) by the multidrug resistance protein 1 (MRP1). Requirement for glutathione or a non-sulfur-containing analog J Biol Chem 2001 276 27846 27854 11375986 10.1074/jbc.M102453200
Kerb R Hoffmeyer S Brinkmann U ABC drug transporters: hereditary polymorphisms and pharmacological impact in MDR1, MRP1 and MRP2 Pharmacogenomics 2001 2 51 64 11258197 10.1517/14622416.2.1.51
Saito S Iida A Sekine A Miura Y Ogawa C Kawauchi S Higuchi S Nakamura Y Identification of 779 genetic variations in eight genes encoding members of the ATP-binding cassette, subfamily C (ABCC/MRP/CFTR J Hum Genet 2002 47 147 171 12166651 10.1007/s100380200018
Oselin K Mrozikiewicz PM Gaikovitch E Pahkla R Roots I Frequency of MRP1 genetic polymorphisms and their functional significance in Caucasians: detection of a novel mutation G816A in the human MRP1 gene Eur J Clin Pharmacol 2003 59 347 350 12856092 10.1007/s00228-003-0625-z
Ito S Ieiri I Tanabe M Suzuki A Higuchi S Otsubo K Polymorphism of the ABC transporter genes, MDR1, MRP1 and MRP2/cMOAT, in healthy Japanese subjects Pharmacogenetics 2001 11 175 184 11266082 10.1097/00008571-200103000-00008
Ito K Olsen SL Qiu W Deeley RG Cole SP Mutation of a single conserved tryptophan in multidrug resistance protein 1 (MRP1/ABCC1) results in loss of drug resistance and selective loss of organic anion transport J Biol Chem 2001 276 15616 15624 11278867 10.1074/jbc.M011246200
Conrad S Kauffmann HM Ito K Leslie EM Deeley RG Schrenk D Cole SP A naturally occurring mutation in MRP1 results in a selective decrease in organic anion transport and in increased doxorubicin resistance Pharmacogenetics 2002 12 321 330 12042670 10.1097/00008571-200206000-00008
Lorico A Rappa G Finch RA Yang D Flavell RA Sartorelli AC Disruption of the murine MRP (multidrug resistance protein) gene leads to increased sensitivity to etoposide (VP-16) and increased levels of glutathione Cancer Res 1997 57 5238 5242 9393741
Wijnholds J Evers R van Leusden MR Mol CA Zaman GJ Mayer U Beijnen JH van der Valk M Krimpenfort P Borst P Increased sensitivity to anticancer drugs and decreased inflammatory response in mice lacking the multidrug resistance-associated protein Nat Med 1997 3 1275 1279 9359705 10.1038/nm1197-1275
Bain LJ Feldman RA Altered expression of sulfotransferases, glucuronosyltransferases and mrp transporters in FVB/mrp1-/- mice Xenobiotica 2003 33 1173 1183 14742140 10.1080/00498250310001609138
Verbon A Leemans JC Weijer S Florquin S van der PT Mice lacking the multidrug resistance protein 1 have a transiently impaired immune response during tuberculosis Clin Exp Immunol 2002 130 32 36 12296850 10.1046/j.1365-2249.2002.01952.x
Schultz MJ Wijnholds J Peppelenbosch MP Vervoordeldonk MJ Speelman P van Deventer SJ Borst P van der Poll T Mice lacking the multidrug resistance protein 1 are resistant to Streptococcus pneumoniae-induced pneumonia J Immunol 2001 166 4059 4064 11238654
Crooks SW Bayley DL Hill SL Stockley RA Bronchial inflammation in acute bacterial exacerbations of chronic bronchitis: the role of leukotriene B4 Eur Respir J 2000 15 274 280 10706491 10.1034/j.1399-3003.2000.15b09.x
Biernacki WA Kharitonov SA Barnes PJ Increased leukotriene B4 and 8-isoprostane in exhaled breath condensate of patients with exacerbations of COPD Thorax 2003 58 294 298 12668789 10.1136/thorax.58.4.294
Loe DW Stewart RK Massey TE Deeley RG Cole SP ATP-dependent transport of aflatoxin B1 and its glutathione conjugates by the product of the multidrug resistance protein (MRP) gene Mol Pharmacol 1997 51 1034 1041 9187270
Lorico A Nesland J Emilsen E Fodstad O Rappa G Role of the multidrug resistance protein 1 gene in the carcinogenicity of aflatoxin B1: investigations using mrp1-null mice Toxicology 2002 171 201 205 11836025 10.1016/S0300-483X(01)00584-4
Hayashi A Suzuki H Itoh K Yamamoto M Sugiyama Y Transcription factor Nrf2 is required for the constitutive and inducible expression of multidrug resistance-associated protein 1 in mouse embryo fibroblasts Biochem Biophys Res Commun 2003 310 824 829 14550278 10.1016/j.bbrc.2003.09.086
Rangasamy T Cho CY Thimmulappa RK Zhen L Srisuma SS Kensler TW Yamamoto M Petrache I Tuder RM Biswal S Genetic ablation of Nrf2 enhances susceptibility to cigarette smoke-induced emphysema in mice J Clin Invest 2004 114 1248 1259 15520857 10.1172/JCI200421146
Yamamoto T Yoh K Kobayashi A Ishii Y Kure S Koyama A Sakamoto T Sekizawa K Motohashi H Yamamoto M Identification of polymorphisms in the promoter region of the human NRF2 gene Biochem Biophys Res Commun 2004 321 72 79 15358217 10.1016/j.bbrc.2004.06.112
Wijnholds J deLange EC Scheffer GL van den Berg DJ Mol CA van der Valk M Schinkel AH Scheper RJ Breimer DD Borst P Multidrug resistance protein 1 protects the choroid plexus epithelium and contributes to the blood-cerebrospinal fluid barrier J Clin Invest 2000 105 279 285 10675353
Johnson DR Finch RA Lin ZP Zeiss CJ Sartorelli AC The pharmacological phenotype of combined multidrug-resistance mdr1a/1b- and mrp1-deficient mice Cancer Res 2001 61 1469 1476 11245453
Muramatsu T Johnson DR Finch RA Johnson LK Leffert JJ Lin ZP Pizzorno G Sartorelli AC Age-related differences in vincristine toxicity and biodistribution in wild-type and transporter-deficient mice Oncol Res 2004 14 331 343 15301424
Bandi N Kompella UB Budesonide reduces multidrug resistance-associated protein 1 expression in an airway epithelial cell line (Calu-1) Eur J Pharmacol 2002 437 9 17 11864633 10.1016/S0014-2999(02)01267-0
Van der Deen M Timmer-Bosscha H Timens W Postma DS de Vries EG Effect of cigarette smoke extract on MRP1 function in bronchial epithelial cells [abstract] Proc Amer Assoc Cancer Res 2004 45 A508
Nagayoshi K Nemoto T Yokoyama S Yamashita F Hashida M Effect of polycyclic aromatic hydrocarbons on generation and efflux of glutathione conjugates in primary cultured alveolar epithelial cells Drug Metab Pharmacokinet 2004 19 407 412 15681894 10.2133/dmpk.19.407
Torky AR Stehfest E Viehweger K Taege C Foth H Immuno-histochemical detection of MRPs in human lung cells in culture Toxicology 2005 207 437 450 15664271 10.1016/j.tox.2004.10.014
Buchler M Konig J Brom M Kartenbeck J Spring H Horie T Keppler D cDNA cloning of the hepatocyte canalicular isoform of the multidrug resistance protein, cMrp, reveals a novel conjugate export pump deficient in hyperbilirubinemic mutant rats J Biol Chem 1996 271 15091 15098 8662992 10.1074/jbc.271.25.15091
Cui Y Konig J Buchholz JK Spring H Leier I Keppler D Drug resistance and ATP-dependent conjugate transport mediated by the apical multidrug resistance protein, MRP2, permanently expressed in human and canine cells Mol Pharmacol 1999 55 929 937 10220572
Leier I Jedlitschky G Buchholz U Cole SP Deeley RG Keppler D The MRP gene encodes an ATP-dependent export pump for leukotriene C4 and structurally related conjugates J Biol Chem 1994 269 27807 27810 7961706
Kamisako T Leier I Cui Y Konig J Buchholz U Hummel-Eisenbeiss J Keppler D Transport of monoglucuronosyl and bisglucuronosyl bilirubin by recombinant human and rat multidrug resistance protein 2 Hepatology 1999 30 485 490 10421658 10.1002/hep.510300220
Holland IB Cole SP Kuchler K Higgins CF ABC proteins, from bacteria to men 2003 London: Elsevier Sci Ltd p427
Paulusma CC Bosma PJ Zaman GJ Bakker CT Otter M Scheffer GL Scheper RJ Borst P Oude Elferink RP Congenital jaundice in rats with a mutation in a multidrug resistance-associated protein gene Science 1996 271 1126 1128 8599091
Hosokawa S Tagaya O Mikami T Nozaki Y Kawaguchi A Yamatsu K Shamoto M A new rat mutant with chronic conjugated hyperbilirubinemia and renal glomerular lesions Lab Anim Sci 1992 42 27 34 1316504
Kuroda M Kobayashi Y Tanaka Y Itani T Mifuji R Araki J Kaito M Adachi Y Increased hepatic and renal expressions of multidrug resistance-associated protein 3 in Eisai hyperbilirubinuria rats J Gastroenterol Hepatol 2004 19 146 153 14731123 10.1111/j.1440-1746.2004.03275.x
Kiuchi Y Suzuki H Hirohashi T Tyson CA Sugiyama Y cDNA cloning and inducible expression of human multidrug resistance associated protein 3 (MRP3) FEBS Lett 1998 433 149 152 9738950 10.1016/S0014-5793(98)00899-0
Kool M van der Linden M de Haas M Scheffer GL de Vree JM Smith AJ Jansen G Peters GJ Ponne N Scheper RJ Elferink RP Baas F Borst P MRP3, an organic anion transporter able to transport anti-cancer drugs Proc Natl Acad Sci U S A 1999 96 6914 6919 10359813 10.1073/pnas.96.12.6914
Scheffer GL Kool M de Haas M de Vree JM Pijnenborg AC Bosman DK Elferink RP van der Valk P Borst P Scheper RJ Tissue distribution and induction of human multidrug resistant protein 3 Lab Invest 2002 82 193 201 11850532
Oguri T Isobe T Fujitaka K Ishikawa N Kohno N Association between expression of the MRP3 gene and exposure to platinum drugs in lung cancer Int J Cancer 2001 93 584 589 11477564 10.1002/ijc.1369
Young LC Campling BG Voskoglou-Nomikos T Cole SP Deeley RG Gerlach JH Expression of multidrug resistance protein-related genes in lung cancer: correlation with drug response Clin Cancer Res 1999 5 673 680 10100721
Borst P Evers R Kool M Wijnholds J A family of drug transporters: the multidrug resistance-associated proteins J Natl Cancer Inst 2000 92 1295 1302 10944550 10.1093/jnci/92.16.1295
Wielinga PR van der Heijden I Reid G Beijnen JH Wijnholds J Borst P Characterization of the MRP4- and MRP5-mediated transport of cyclic nucleotides from intact cells J Biol Chem 2003 278 17664 17671 12637526 10.1074/jbc.M212723200
Jedlitschky G Burchell B Keppler D The multidrug resistance protein 5 functions as an ATP-dependent export pump for cyclic nucleotides J Biol Chem 2000 275 30069 30074 10893247 10.1074/jbc.M005463200
Chen ZS Lee K Kruh GD Transport of cyclic nucleotides and estradiol 17-beta-D-glucuronide by multidrug resistance protein 4. Resistance to 6-mercaptopurine and 6-thioguanine J Biol Chem 2001 276 33747 33754 11447229 10.1074/jbc.M104833200
Van Aubel RA Smeets PH Peters JG Bindels RJ Russel FG The MRP4/ABCC4 gene encodes a novel apical organic anion transporter in human kidney proximal tubules: putative efflux pump for urinary cAMP and cGMP J Am Soc Nephrol 2002 13 595 603 11856762
Van den Berge M Polosa R Kerstjens HA Postma DS The role of endogenous and exogenous AMP in asthma and chronic obstructive pulmonary disease J Allergy Clin Immunol 2004 114 737 746 15480309 10.1016/j.jaci.2004.05.071
Reid G Wielinga P Zelcer N van der Heijden I Kuil A de Haas M Wijnholds J Borst P The human multidrug resistance protein MRP4 functions as a prostaglandin efflux transporter and is inhibited by nonsteroidal antiinflammatory drugs Proc Natl Acad Sci USA 2003
Zelcer N Reid G Wielinga P Kuil A van der Heijden I Schuetz JD Borst P Steroid and bile acid conjugates are substrates of human multidrug-resistance protein (MRP) 4 (ATP-binding cassette C4) Biochem J 2003 371 361 367 12523936 10.1042/BJ20021886
Prehm P Schumacher U Inhibition of hyaluronan export from human fibroblasts by inhibitors of multidrug resistance transporters Biochem Pharmacol 2004 68 1401 1410 15345330 10.1016/j.bcp.2004.06.017
Szakacs G Annereau JP Lababidi S Shankavaram U Arciello A Bussey KJ Reinhold W Guo Y Kruh GD Reimers M Weinstein JN Gottesman MM Predicting drug sensitivity and resistance; Profiling ABC transporter genes in cancer cells Cancer Cell 2004 6 129 137 15324696 10.1016/j.ccr.2004.06.026
Savaraj N Wu C Wangpaichitr M Kuo MT Lampidis T Robles C Furst AJ Feun L Overexpression of mutated MRP4 in cisplatin resistant small cell lung cancer cell line: collateral sensitivity to azidothymidine Int J Oncol 2003 23 173 179 12792791
Suzuki T Nishio K Tanabe S The MRP family and anticancer drug metabolism Curr Drug Metab 2001 2 367 377 11766988 10.2174/1389200013338289
Scheffer GL Hu X Pijnenborg AC Wijnholds J Bergen AA Scheper RJ MRP6 (ABCC6) detection in normal human tissues and tumors Lab Invest 2002 82 515 518 11950908
Hu X Plomp AS van Soest S Wijnholds J de Jong PT Bergen AA Pseudoxanthoma elasticum: a clinical, histopathological, and molecular update Surv Ophthalmol 2003 48 424 438 12850230 10.1016/S0039-6257(03)00053-5
Jackson A Loh CL Pulmonary calcification and elastic tissue damage in pseudoxanthoma elasticum Histopathology 1980 4 607 611 7439889
Beck K Hayashi K Nishiguchi B Le Saux O Hayashi M Boyd CD The distribution of Abcc6 in normal mouse tissues suggests multiple functions for this ABC transporter J Histochem Cytochem 2003 51 887 902 12810839
Hopper-Borge E Chen ZS Shchaveleva I Belinsky MG Kruh GD Analysis of the drug resistance profile of multidrug resistance protein 7 (ABCC10): resistance to docetaxel Cancer Res 2004 64 4927 4930 15256465
Chen ZS Guo Y Belinsky MG Kotova E Kruh GD Transport of bile acids, sulfated steroids, estradiol 17-beta-D-glucuronide, and leukotriene C4 by human multidrug resistance protein 8 (ABCC11) Mol Pharmacol 2005 67 545 557 15537867 10.1124/mol.104.007138
Denning GM Ostedgaard LS Welsh MJ Abnormal localization of cystic fibrosis transmembrane conductance regulator in primary cultures of cystic fibrosis airway epithelia J Cell Biol 1992 118 551 559 1379244 10.1083/jcb.118.3.551
Engelhardt JF Zepeda M Cohn JA Yankaskas JR Wilson JM Expression of the cystic fibrosis gene in adult human lung J Clin Invest 1994 93 737 749 7509347
Puchelle E Gaillard D Ploton D Hinnrasky J Fuchey C Boutterin MC Jacquot J Dreyer D Pavirani A Dalemans W Differential localization of the cystic fibrosis transmembrane conductance regulator in normal and cystic fibrosis airway epithelium Am J Respir Cell Mol Biol 1992 7 485 491 1384582
Kalin N Claass A Sommer M Puchelle E Tummler B DeltaF508 CFTR protein expression in tissues from patients with cystic fibrosis J Clin Invest 1999 103 1379 1389 10330420
Pilewski JM Frizzell RA Role of CFTR in airway disease Physiol Rev 1999 79 S215 S255 9922383
Brezillon S Dupuit F Hinnrasky J Marchand V Kalin N Tummler B Puchelle E Decreased expression of the CFTR protein in remodeled human nasal epithelium from non-cystic fibrosis patients Lab Invest 1995 72 191 200 7531792
Brezillon S Hamm H Heilmann M Schafers HJ Hinnrasky J Wagner TO Puchelle E Tummler B Decreased expression of the cystic fibrosis transmembrane conductance regulator protein in remodeled airway epithelium from lung transplanted patients Hum Pathol 1997 28 944 952 9269831 10.1016/S0046-8177(97)90010-1
Penque D Mendes F Beck S Farinha C Pacheco P Nogueira P Lavinha J Malho R Amaral MD Cystic fibrosis F508del patients have apically localized CFTR in a reduced number of airway cells Lab Invest 2000 80 857 868 10879737
Morris AP Cunningham SA Benos DJ Frizzell RA Cellular differentiation is required for cAMP but not Ca(2+)-dependent Cl- secretion in colonic epithelial cells expressing high levels of cystic fibrosis transmembrane conductance regulator J Biol Chem 1992 267 5575 5583 1372005
Hollande E Fanjul M Chemin-Thomas C Devaux C Demolombe S Van Rietschoten J Guy-Crotte O Figarella C Targeting of CFTR protein is linked to the polarization of human pancreatic duct cells in culture Eur J Cell Biol 1998 76 220 227 9716269
Hurbain I Sermet-Gaudelus I Vallee B Feuillet MN Lenoir G Bernaudin JF Edelman A Fajac A Evaluation of MRP1-5 gene expression in cystic fibrosis patients homozygous for the delta F508 mutation Pediatr Res 2003 54 627 634 12930913 10.1203/01.PDR.0000090926.16166.3F
Linsdell P Hanrahan JW Substrates of multidrug resistance-associated proteins block the cystic fibrosis transmembrane conductance regulator chloride channel Br J Pharmacol 1999 126 1471 1477 10217542 10.1038/sj.bjp.0702458
Bebok Z Varga K Hicks JK Venglarik CJ Kovacs T Chen L Hardiman KM Collawn JF Sorscher EJ Matalon S Reactive oxygen nitrogen species decrease cystic fibrosis transmembrane conductance regulator expression and cAMP-mediated Cl- secretion in airway epithelia J Biol Chem 2002 277 43041 43049 12194970 10.1074/jbc.M203154200
Hickman-Davis J Gibbs-Erwin J Lindsey JR Matalon S Surfactant protein A mediates mycoplasmacidal activity of alveolar macrophages by production of peroxynitrite Proc Natl Acad Sci U S A 1999 96 4953 4958 10220400 10.1073/pnas.96.9.4953
Hickman-Davis JM O'Reilly P Davis IC Peti-Peterdi J Davis G Young KR Devlin RB Matalon S Killing of Klebsiella pneumoniae by human alveolar macrophages Am J Physiol Lung Cell Mol Physiol 2002 282 L944 L956 11943658
Eiserich JP Hristova M Cross CE Jones AD Freeman BA Halliwell B van der Vliet A Formation of nitric oxide-derived inflammatory oxidants by myeloperoxidase in neutrophils Nature 1998 391 393 397 9450756 10.1038/34923
Punjabi CJ Laskin JD Pendino KJ Goller NL Durham SK Laskin DL Production of nitric oxide by rat type II pneumocytes: increased expression of inducible nitric oxide synthase following inhalation of a pulmonary irritant Am J Respir Cell Mol Biol 1994 11 165 172 7519435
Weinberger B Heck DE Laskin DL Laskin JD Nitric oxide in the lung: therapeutic and cellular mechanisms of action Pharmacol Ther 1999 84 401 411 10665837 10.1016/S0163-7258(99)00044-3
Meng QH Polak JM Edgar AJ Chacon MR Evans TJ Gruenert DC Bishop AE Neutrophils enhance expression of inducible nitric oxide synthase in human normal but not cystic fibrosis bronchial epithelial cells J Pathol 2000 190 126 132 10657009 10.1002/(SICI)1096-9896(200002)190:2<126::AID-PATH500>3.0.CO;2-#
Trezise AE Ratcliff R Hawkins TE Evans MJ Freeman TC Romano PR Higgins CF Colledge WH Co-ordinate regulation of the cystic fibrosis and multidrug resistance genes in cystic fibrosis knockout mice Hum Mol Genet 1997 6 527 537 9097955 10.1093/hmg/6.4.527
Ratcliff R Evans MJ Cuthbert AW MacVinish LJ Foster D Anderson JR Colledge WH Production of a severe cystic fibrosis mutation in mice by gene targeting Nat Genet 1993 4 35 41 7685652 10.1038/ng0593-35
Shen BQ Finkbeiner WE Wine JJ Mrsny RJ Widdicombe JH Calu-3: a human airway epithelial cell line that shows cAMP-dependent Cl- secretion Am J Physiol 1994 266 L493 L501 7515578
Baudouin-Legros M Brouillard F Tondelier D Hinzpeter A Edelman A Effect of ouabain on CFTR gene expression in human Calu-3 cells Am J Physiol Cell Physiol 2003 284 C620 C626 12556359
Cozens AL Yezzi MJ Kunzelmann K Ohrui T Chin L Eng K Finkbeiner WE Widdicombe JH Gruenert DC CFTR expression and chloride secretion in polarized immortal human bronchial epithelial cells Am J Respir Cell Mol Biol 1994 10 38 47 7507342
Kogan I Ramjeesingh M Li C Kidd JF Wang Y Leslie EM Cole SP Bear CE CFTR directly mediates nucleotide-regulated glutathione flux EMBO J 2003 22 1981 1989 12727866 10.1093/emboj/cdg194
Croop JM Tiller GE Fletcher JA Lux ML Raab E Goldenson D Son D Arciniegas S Wu RL Isolation and characterization of a mammalian homolog of the Drosophila white gene Gene 1997 185 77 85 9034316 10.1016/S0378-1119(96)00633-6
Doyle LA Yang W Abruzzo LV Krogmann T Gao Y Rishi AK Ross DD A multidrug resistance transporter from human MCF-7 breast cancer cells Proc Natl Acad Sci U S A 1998 95 15665 15670 9861027 10.1073/pnas.95.26.15665
Komatani H Kotani H Hara Y Nakagawa R Matsumoto M Arakawa H Nishimura S Identification of breast cancer resistant protein/mitoxantrone resistance/placenta-specific, ATP-binding cassette transporter as a transporter of NB-506 and J-10 topoisomerase I inhibitors with an indolocarbazole structure Cancer Res 7088 61 2827 2832 11306452
Diestra JE Scheffer GL Catala II Maliepaard M Schellens JH Scheper RJ Germa-Lluch JR Izquierdo MA Frequent expression of the multi-drug resistance-associated protein BCRP/MXR/ABCP/ABCG2 in human tumours detected by the BXP-21 monoclonal antibody in paraffin-embedded material J Pathol 2002 198 213 219 12237881 10.1002/path.1203
Yoh K Ishii G Yokose T Minegishi Y Tsuta K Goto K Nishiwaki Y Kodama T Suga M Ochiai A Breast cancer resistance protein impacts clinical outcome in platinum-based chemotherapy for advanced non-small cell lung cancer Clin Cancer Res 2004 10 1691 1697 15014021
Jonker JW Buitelaar M Wagenaar E van der Valk MA Scheffer GL Scheper RJ Plosch T Kuipers F Elferink RP Rosing H Beijnen JH Schinkel AH The breast cancer resistance protein protects against a major chlorophyll-derived dietary phototoxin and protoporphyria Proc Natl Acad Sci U S A 2002 99 15649 15654 12429862 10.1073/pnas.202607599
Summer R Kotton DN Sun X Ma B Fitzsimmons K Fine A Side population cells and Bcrp1 expression in lung Am J Physiol Lung Cell Mol Physiol 2003 285 L97 104 12626330
Giangreco A Shen H Reynolds SD Stripp BR Molecular Phenotype of Airway Side Population Cells Am J Physiol Lung Cell Mol Physiol 2003
Neuringer IP Randell SH Stem cells and repair of lung injuries Respir Res 2004 5 6 15285789 10.1186/1465-9921-5-6
Agassandian M Mathur SN Zhou J Field FJ Mallampalli RK Oxysterols trigger ABCA1-mediated basolateral surfactant efflux Am J Respir Cell Mol Biol 2004 31 227 233 15039140 10.1165/rcmb.2004-0038OC
Drobnik W Lindenthal B Lieser B Ritter M Christiansen WT Liebisch G Giesa U Igel M Borsukova H Buchler C Fung-Leung WP Von Bergmann K Schmitz G ATP-binding cassette transporter A1 (ABCA1) affects total body sterol metabolism Gastroenterology 2001 120 1203 1211 11266384 10.1053/gast.2001.23250
McNeish J Aiello RJ Guyot D Turi T Gabel C Aldinger C Hoppe KL Roach ML Royer LJ de Wet J Broccardo C Chimini G Francone OL High density lipoprotein deficiency and foam cell accumulation in mice with targeted disruption of ATP-binding cassette transporter-1 Proc Natl Acad Sci U S A 2000 97 4245 4250 10760292 10.1073/pnas.97.8.4245
Wang N Lan D Gerbod-Giannone M Linsel-Nitschke P Jehle AW Chen W Martinez LO Tall AR ATP-binding cassette transporter A7 (ABCA7) binds apolipoprotein A-I and mediates cellular phospholipid but not cholesterol efflux J Biol Chem 2003 278 42906 42912 12917409 10.1074/jbc.M307831200
Yamano G Funahashi H Kawanami O Zhao LX Ban N Uchida Y Morohoshi T Ogawa J Shioda S Inagaki N ABCA3 is a lamellar body membrane protein in human lung alveolar type II cells FEBS Lett 2001 508 221 225 11718719 10.1016/S0014-5793(01)03056-3
|
15967026
|
PMC1200430
|
CC BY
|
2021-01-04 16:23:25
|
no
|
Respir Res. 2005 Jun 20; 6(1):59
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-59
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-591596702610.1186/1465-9921-6-59ReviewATP-binding cassette (ABC) transporters in normal and pathological lung van der Deen Margaretha [email protected] Vries Elisabeth GE [email protected] Wim [email protected] Rik J [email protected] Hetty [email protected] Dirkje S [email protected] University Medical Center Groningen, Department of Internal Medicine, Medical Oncology, Hanzeplein 1, 9713 GZ, Groningen, The Netherlands2 Department of Pulmonary Medicine, Hanzeplein 1, 9713 GZ, Groningen, The Netherlands3 Department of Pathology and Laboratory Medicine, Hanzeplein 1, 9713 GZ, Groningen, The Netherlands4 Free University, Department of Pathology, Boelelaan 1117, 1081 HV, Amsterdam, The Netherlands2005 20 6 2005 6 1 59 59 4 4 2005 20 6 2005 Copyright © 2005 van der Deen et al; licensee BioMed Central Ltd.2005van der Deen et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ATP-binding cassette (ABC) transporters are a family of transmembrane proteins that can transport a wide variety of substrates across biological membranes in an energy-dependent manner. Many ABC transporters such as P-glycoprotein (P-gp), multidrug resistance-associated protein 1 (MRP1) and breast cancer resistance protein (BCRP) are highly expressed in bronchial epithelium. This review aims to give new insights in the possible functions of ABC molecules in the lung in view of their expression in different cell types. Furthermore, their role in protection against noxious compounds, e.g. air pollutants and cigarette smoke components, will be discussed as well as the (mal)function in normal and pathological lung. Several pulmonary drugs are substrates for ABC transporters and therefore, the delivery of these drugs to the site of action may be highly dependent on the presence and activity of many ABC transporters in several cell types. Three ABC transporters are known to play an important role in lung functioning. Mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) gene can cause cystic fibrosis, and mutations in ABCA1 and ABCA3 are responsible for respectively Tangier disease and fatal surfactant deficiency. The role of altered function of ABC transporters in highly prevalent pulmonary diseases such as asthma or chronic obstructive pulmonary disease (COPD) have hardly been investigated so far. We especially focused on polymorphisms, knock-out mice models and in vitro results of pulmonary research. Insight in the function of ABC transporters in the lung may open new ways to facilitate treatment of lung diseases.
==== Body
Introduction
The prime role of the airways (trachea, bronchi, bronchioles and terminal bronchioles) is to conduct air into and out of the lung and to form a first line of defence against undesired constituents of inhaled air. The airways are continuously exposed to pathogens, irritants, pollutants and agents that produce oxidative stress and therefore, the composition of the respiratory tract surface is very important. The upper airways contain specialised cell types such as ciliated cells and mucous secreting goblet cells. The lower conducting airways (respiratory bronchioles, alveolar ducts and alveolar sacs) participate in gas exchange by diffusion. The alveolar epithelial surface comprises essentially two cell types, the alveolar epithelial type I cell and the cuboidal alveolar epithelial type II cell. Type I cells flatten out and in this way constitute approximately 95% of the total alveolar surface, whereas type II cells are more numerous and produce surfactant [1].
ABC (ATP-binding cassette) transporters are a family of transmembrane proteins that can transport a wide variety of substrates across biological membranes in an energy-dependent manner. They are phylogenetically classified in seven distinct subfamilies of transporters (ABCA to ABCG), which are again divided into subgroups. To date, 48 ABC transporters have been detected in the human body [2,3]. Overexpression of ABC transporters such as P-glycoprotein (P-gp) and the multidrug resistance-associated protein 1 (MRP1) were initially detected in tumour cell lines. Their overexpression is associated with increased efflux of chemotherapeutic drugs such as anthracyclines, epipodophyllotoxins and vinca-alkaloids, and this can result in so-called multidrug resistance (MDR). Many MDR proteins can act as drug efflux pumps, resulting in decreased intracellular concentrations of toxic compounds at the site of action. MRP1 and P-gp expression in the lung have especially been studied in the context of small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC). Currently we know that ABC transporters are present in virtually every cell of all species and play central roles in physiology. The prominent expression of P-gp and MRP1 in the human lung [4] suggests that these transporters may be pivotal in the protection against endogenous or exogenous toxic compounds entering the lung.The delivery of pulmonary drugs to reach the site of action may also depend on the presence and activity of many ABC transporters [5]. Langmann et al. developed quantitative real-time RT-PCR expression profiling of 47 ABC transporters in 20 different human tissues. Tissues with a barrier function such as lung and trachea were identified to have high transcriptional activity for many ABC transporters [6]. There are already clear proofs of important functions of ABC transporters in the lung. The ABC transporter most widely investigated is the cystic fibrosis transmembrane conductance regulator (CFTR), because mutations in this gene are responsible for the development of cystic fibrosis [7]. Mutations in ABCA1 are causative for Tangier disease [8]. In newborns it was shown that mutations in ABCA3 cause aberrant production of surfactant which can be lethal [9]. The role of altered function of ABC transporters in highly prevalent pulmonary diseases such as asthma or chronic obstructive pulmonary disease (COPD) have hardly been investigated so far [10].
This review aims to give new insights in the possible functions of ABC molecules in the lung in view of their expression in different cell types. Furthermore, their roles in protection against noxious compounds will be discussed as well as the (mal)function in normal and pathological lung. We especially focused on the members of the ABCC subfamily of transporters (to which the MRPs and CFTR belong), the ABCB subfamily (a.o. MDR1/Pgp and MDR3/Pgp) and the ABCA subfamily (a.o. ABCA1 and ABCA3), because these are a selection of the best characterised human transporters and because they have been investigated in human lungs. Results of in vitro studies in pulmonary research are being reviewed and an update will be given about what is known about pulmonary functional changes in ABC transporter knock-out mice models. In addition, the (potential) effects of polymorphisms in ABC transporters on lung functioning are discussed.
MDR1/P-gp
MDR1 localisation and function
The MDR1 (ABCB1) gene is located on chromosome 7q21.12 and encodes for P-gp. P-gp plays a role in cell defence against environmental attacks such as generated by xenobiotics. It transports many hydrophobic substrates and anti-cancer drugs including etoposide, doxorubicin and vinblastine (for review, see [11]) and is mainly apically expressed in organs involved in excretion such as liver and intestine. In the lung, P-gp is expressed at the apical side of ciliated epithelial cells or ciliated collecting ducts, and on apical and lateral surfaces of serous cells of bronchial glands but not in mucus-secreting goblet cells (Figure 1 and 2) (Table 1) [12]. Epithelial cells of the trachea and major bronchi stain strongly while staining of the smaller bronchi is patchy or absent. In addition, P-gp is present in the lateral membranes of normal nasal respiratory mucosa [13]. In human and rat type I alveolar epithelium, P-gp is located at the lumenal side whereas freshly isolated type II cells lack P-gp [14]. In another study, pneumocytes did not stain for P-gp [15]. Some antibodies visualised P-gp in endothelial cells of blood vessels (Figure 1) [12,15,16]. In addition, alveolar and peripheral blood monocyte-derived macrophages stained positive but variably for P-gp [4,16,17]. The observed apical epithelial expression may signify that P-gp plays a role in transport of compounds from the interstitium into the lumen. However, the precise function of P-gp in the lung is as yet unknown. Interestingly, P-gp was found to play a role in the regulation of cell volume activated chloride channels and to possess channel activities [18,19]. However, Cl- and K+ conductances were not affected by the level of P-gp expression and the physiological role of P-gp in volume-activated chloride currents is still unclear [20-22]. Many lipophilic amines, such as fentanyl, highly accumulate in the lung. It was suggested that P-gp plays a role in the disposition of these pulmonary amine drugs [23]. Cellular uptake in lung microvascular endothelial cells of fentanyl was inhibited by the P-gp substrate verapamil, but not by the P-gp blocking antibody UIC2. The authors suggested that this results from an inhibition of fentanyl uptake. This is however unlikely, since P-gp is primarily responsible for removal of drugs out of cells.
Figure 1 Expression of ATP-binding cassette proteins in several cell types of human lung. ABC, ATP-binding cassette; BM, basement membrane; BCRP, breast cancer resistance protein; CFTR, cystic fibrosis transmembrane conductance regulator; MRP, multidrug resistance-associated protein; P-gp, P-glycoprotein. *Conflicting results exist in literature.
Figure 2 Immunohistochemical staining of ATP-binding cassette proteins in human lung. A. apical expression of P-gp in bronchial epithelium (COPD patient; frozen section, antibody C219), B. basolateral expression of MRP1 in bronchial epithelium (COPD patient; antibody MRPr1), C. MRP1 expression in bronchoalveolar lavage cells (healthy individual; antibody MRPr1). Lu, lumen. Scale bar = 25 μM.
Table 1 Summary of features of ABC transporters in human lung.
ABC gene Gene location Functional role or substrates Linked to disease Protein expression in lung cells
(+ localisation$) References#
MDR1/P-gp (ABCB1) 7q21.12 Drug resistance, Hydrophobic organic cations Unknown Bronchial epithelium (ap), mucinous glands (ap), alveolar type I cells* (ap), endothelium* (ap), alv. macroph. [4,12,14-16]
MDR3/P-gp (ABCB4) 7q21.12 Phosphatidyl choline Unknown Absent [4]
MRP1 (ABCC1) 16p13.12 Drug resistance, Organic anions (e.g. GSH conjugates, LTC4) COPD? Bronchial epithelium (ba/la), goblet cells (ba), peripheral epithelial cells* (ba/la), seromucinous glands (ba/la), alv. macroph. [4,53,85]
MRP2 (ABCC2) 10q24.2 Drug resistance, Organic anions Dubin-Johnson syndrome Bronchial epithelium* (ap), primary bronchial and peripheral epithelial cells* (in) [4,85,90]
MRP3 (ABCC3) 17q21.33 Drug resistance, Organic anions Unknown Primary bronchial and peripheral epithelial cells* (ba/la) [85]
MRP4 (ABCC4) 13q32.1 Nucleoside analogues, Prostaglandin E1, E2 Unknown Primary bronchial and peripheral epithelial cells* (in) [85]
MRP5 (ABCC5) 3q27.1 Nucleoside analogues, Hyaluron Unknown Primary bronchial and peripheral epithelial cells* (in) [85]
MRP6 (ABCC6) 16p13.12 Unknown Pseudoxanthoma elasticum Unknown
MRP7 (ABCC10) 6p21.1 Drug resistance Unknown Unknown
MRP8 (ABCC11) 16q12.1 Conjugated steroids, Nucleoside analogues, bile acids Unknown Unknown
MRP9 (ABCC12) 16q12.1 Unknown Unknown Unknown
CFTR (ABCC7) 7q31.31 Chloride ion channel Cystic fibrosis Bronchial epithelium (ap), seromucinous glands (ap), Clara cells*, Alv. type I cells* [117-120]
BCRP (ABCG2) 4q22 Drug resistance, Protection food toxins Unknown Bronchial epithelium (ba/la), endothelium, seromucinous glands [4]
ABCA1 9q31.1 Cholesterol and phospholipids Tangier disease Alv. type II cells [151]
ABCA3 16p13.3 Surfactant secretion Surfactant deficiency Alv. type II cells [155]
$Cellular localisation: ap, apical; ba, basal; la, lateral; in, intracellular; alv., alveolar; macroph., macrophages; #References are mentioned that demonstrate protein localisation histologically; *Conflicting results exist in literature.
P-gp mRNA expression was not increased in smokers (n = 11) compared to ex-and non-smokers (n = 7) [12]. Whether P-gp expression levels may play a defensive role towards tobacco-derived agents remains to be investigated.
MDR1 in tumours
High P-gp expression can imply chemotherapeutic resistance due to increased chemotherapeutic drug efflux. In cancer therapy, many attempts have been made to reverse MDR mechanisms. However, in a randomised double-blind trial in 130 SCLC patients no positive effects were seen with the P-gp modulator megestrol acetate in addition to chemotherapeutic drugs, suggesting that levels of P-gp expression in lung tumours were not relevant or that modulation of P-gp activity was not complete in this treatment [24]. Some studies show higher P-gp expression at the invasion front of lung tumours and it was suggested that P-gp expression augments invasion properties of tumour cells [25]. Only two out of 22 NSCLC samples (both adenocarcinomas) stained positive with three P-gp antibodies [15] and no P-gp was detected on pulmonary carcinoids. Other studies revealed a relation between P-gp and glutathione S-transferase-pi (GST-pi) expression in NSCLC that were exposed in vitro to doxorubicin [26], suggesting that these two factors play a role in doxorubicin resistance. There was also a correlation between current smoking and doxorubicin resistance of NSCLC. Forty-two out of 72 NSCLC smokers expressed P-gp, whereas only two out of 22 tumours of non-smokers were P-gp positive [27].
MDR1 polymorphisms
MDR1 polymorphisms were first described by Hoffmeyer et al. [28] who found a correlation between lower intestinal expression of P-gp and a polymorphism in exon 26. Many single nucleotide polymorphisms (SNPs) have been recognised in the MDR1 gene (see reference [29] for recent review about clinical aspects). The impact of these polymorphisms on lung diseases is still speculative. It was proposed that polymorphisms in the MDR1 gene may have clinical consequences in patients with cystic fibrosis, since MDR1 plays a role in CFTR regulation. Rodents contain two Mdr1 genes, denoted as Mdr1a and Mdr1b. It was shown that Mdr1b mRNA expression in lung parenchyma of outbred rats is very variable and this may also be the case in humans [30]. The possible effects of MDR1 polymorphisms was studied in tobacco-related lung cancer [31]. No clear association was found between the T/T genotype of the C3435T polymorphism and susceptibility to lung cancer in a group of 268 Caucasian men who were current smokers. No relation was found between SNP C3435T in MDR1 and survival in 62 docetaxel-cisplatin-treated NSCLC patients [32]. Immunosuppressive agents such as cyclosporin A and tacrolimus (both calcineurin antagonists) are P-gp substrates. No relation was found of MDR1 G2677T and C3435T genotypes with tacrolimus blood levels in 83 lung transplant patients treated with tacrolimus [33]. Altogether, these data implicate that there is still no clear association between MDR1 polymorphisms and effects on outcome of treatment of lung cancer or lung transplant patients.
MDR1 in animal models
Scheffer et al. detected high P-gp levels in lungs of mice [4]. In rats, Mdr1a and Mdr1b mRNA expression were highest in the ileum [34]. The Mdr1a expression level in rat lung was 2% of the expression in ileum and expression of Mdr1b was 47% of that in ileum. In mice orally treated with dexamethasone for 24 hours, Mdr1b mRNA expression in lungs was decreased, from which the authors deduce that dexamethasone treatment of lung tumours may reverse MDR [35]. To study the in vivo distribution of P-gp, nude rats were injected with a P-gp overexpressing SCLC cell line (GLC4/Pgp) and with a P-gp negative cell line (GLC4) [36]. P-gp function was visualised with radiolabeled P-gp substrate [11C]verapamil by positron emission tomography (PET) with or without P-gp modulator cyclosporin A. The accumulation of [11C]verapamil was significantly increased by cyclosporin A in brains and GLC4/Pgp tumours in these rats. In all other investigated organs including lungs, the accumulation after cyclosporin A treatment was unaltered. In intact rabbit lung, vascular P-gp kinetics were measured in vivo using the lipophilic amine dye rhodamine 6G (R6G) by measuring R6G in the perfusate during circulation [37]. Inhibition of P-gp function with verapamil or GF120918 resulted in higher accumulation of R6G in lung. It was proposed that the opposite would happen when epithelial P-gp was inhibited because R6G would then be retained in the airspace. We propose another possibility that inhibition of epithelial P-gp will also result in higher R6G accumulation. In that case, R6G transport to the lumen is inhibited and as a compensation mechanism it may be transported back to the interstitial side where it either may be retained in the tissue or subsequently be transported into the circulation. This model could be useful in testing a large variety of pulmonary therapeutic agents, such as corticosteroids and sympathicomimetics that may be substrates for transporters in the lung or modulate their activity. Similar studies were also carried out in perfused rat liver to assess the effect of P-gp modulators on the hepatobiliary system, supporting the usefulness of this approach [38].
Mdr1a/1b (-/-) mice lack both genes encoding for P-gp and these mice seem physiologically normal. The penetration of [3H]digoxin (a P-gp substrate) was higher in brain, ovary and adrenal gland. In the lung tissue of these (-/-) mice [39], the level of [3H]digoxin was rather low compared to other organs. The level was 2.6 times higher in (-/-) mice than in (+/+) mice, but this was not significant. The pharmacokinetics of the central nervous system (CNS) drug amitryptiline (a P-gp substrate) and its metabolites were examined in Mdr1a/1b (-/-) mice by high performance liquid chromatography (HPLC) in several organs. Higher concentrations of these metabolites were measured in the brain of these mice, but not in other examined organs including the lung [40]. Thus, P-gp likely plays a more active role in exporting CNS drugs out of the brain than out of the lung. Another possibility is that other transporters in the lung were capable to efflux these drugs as a compensation mechanism.
MDR1 in vitro
Several lung cell culture models have been described for determining P-gp expression and functionality. The immortalised human bronchial epithelial cell line 16HBE14o- resembles primary epithelium and was reported to be suitable for drug metabolism studies [41]. In 16HBE14o- cells, P-gp was expressed at the apical side. Its functional activity was measured with P-gp substrate rhodamine 123 and its transport was inhibited by verapamil [42]. Two lung cell lines, Calu-3 and A549 cells, were compared for P-gp expression and functionality [43]. The bronchiolar adenocarcinoma cell line Calu-3 is suitable for drug transport studies because these cells form tight junctions in culture, which is not the case for the type II alveolar carcinoma cell line A549. P-gp expression was higher in A549 cells than in Calu-3 cells. However, efflux of rhodamine 123 was higher in Calu-3 cells. This may be explained by additional MRP1 activity in these cells because rhodamine 123 is also an MRP1 substrate [44]. In primary rat alveolar type II cells, Mdr1b mRNA levels increased in a time dependent manner in cultures at day 1, 2 and 3 compared to freshly isolated cells. Mdr1b mRNA was present at low levels and increased after oxygen radical induction with paraquat [45]. P-gp expression was below the detection limit in these cells at time of isolation. In freshly isolated primary human bronchial epithelial cells, P-gp was present and increased after 24 hours paraquat exposure. Rhodamine 123 efflux could be measured in these cells, confirming functional activity of P-gp [45]. These results demonstrate that P-gp is upregulated during stress, both from radical production and from ex vivo culturing or differentiation.
MDR3/P-gp
The MDR3/P-gp (ABCB4) gene maps closely to MDR1/P-gp on chromosome 7q21.12 and has high homology with MDR1 although its function is very different. It is involved in phosphatidyl choline transport from the liver into the bile. Its RNA and protein have not been detected in human and mouse lung and trachea [4,6]. and it is therefore not expected that MDR3 is present in the lungs.
MRP1
MRP1 localisation and function
The MRP1 (ABCC1) gene is located on chromosome 16p13.12. Cole et al. discovered in 1992 a non-Pgp mediated MDR mechanism in the human lung cancer cell line H69AR [46]. MRP1 was overexpressed in these cells and it was found that glutathione-, glucuronide-, and sulfate-conjugated organic anions are substrates for MRP1 [47,48]. MRP1 confers resistance to several chemotherapeutic agents including vincristine, daunorubicin and methotrexate [49-51]. Physiological substrates for MRP1 are e.g. leukotriene C4 (LTC4) and glutathione disulfide (GSSG) [52]. Interestingly, these substrates play an important role in lung physiology with respect to inflammation and oxidative stress. MRP1 is highly expressed mainly at the basolateral side of human bronchial epithelial cells (Figure 1 and 2) [53]. Ciliated and basal cells have been collected from brushes of main or lobar bronchi. Basal cells stained strongly on the entire circumference of the plasma membrane. Ciliated epithelial cells and mucous cells stained positive at the basolateral membrane but not at the apical membrane. No intracytoplasmic staining was observed. However, strong apical cytoplasmic staining has been detected below the cilia in respiratory columnar epithelial cells in paraffin sections with antibody MRPr1 [4,13,54]. The discrepancy between these findings may be due to different fixation procedures. Basal cells of seromucinous glands of the lungs also stain positive for MRP1 with a higher intensity in the serous area than mucinous cells. Alveolar macrophages are MRP1 positive with variable staining between individual lung samples (Figure 2) [4]. Kool et al. investigated MRP1, 2, 3, 4, and 5 with an RNase protection assay in total lung RNA and found high MRP1, absence of MRP2, low MRP3 and MRP4 and moderate MRP5 gene transcripts [55]. Total RNA was obtained from lung tissue collected during surgery or autopsy. Therefore, the original cell types can not be distinguished in these samples.
Possibly, the high MRP1 expression at the basolateral side of lung epithelium may assist in the clearance of toxins coming from the luminal or interstitial side (back) into the interstitial fluid [56]. Function of MRP1 expression in the lung epithelium, glands and alveolar macrophages may include extrusion of toxic intracellular substances, antioxidant defence or production of LTC4 as an inflammatory response. Given the fact that MRP1 expression is higher in the lung compared to other solid organs, decreased or increased functional MRP1 expression may have a high impact on development and/or progression of lung diseases and protection against air pollution and inhaled toxic substances such as present in cigarette smoke [10].
MRP1 in tumours
Using mRNA in situ hybridisation Thomas et al. [57] detected high MRP1 expression in normal epithelium whereas the major component of the tumour epithelium showed a negative hybridisation signal. However, MRP1 transcript expression at the invasive front of the lung tumours was consistently stronger and particularly strong in areas with lymphatic or blood vessels. In addition, endothelial cells and lymphocytic infiltrates stained strongly positive for MRP1. These results may indicate a role of MRP1 in invasion or in mitotic activity. In a recent study, all 102 NSCLC tumours were MRP1 positive. In addition, the level of expression was 3-fold higher in DNA-aneuploid cells compared to normal bronchial and carcinomatous DNA-diploid cells [58]. This was associated with more frequent gain of chromosome 16 where the MRP1 gene is located. Overexpression of MRP1 may therefore be an important factor of intrinsic resistance to chemotherapy in NSCLC. High MRP1 expression can also result in increased apoptosis as shown by experiments with verapamil in MRP1 transfected baby hamster kidney-21 (BHK-21) cells [59]. Verapamil acts as an apoptogen in these cells when compared to MRP1 negative cells or mutant MRP1 transfected control cells. This was accompanied by depletion of intracellular GSH due to transport of GSH by MRP1. Indeed, addition of extracellular GSH prevented cell death. This mechanism may be valuable for treatment of MRP1-positive tumours.
MRP1 in non-malignant diseases
Cigarette smoking is the principle risk factor for the development of COPD. Our preliminary results indicate that MRP1 expression is diminished in bronchial epithelium of COPD patients (ex-smokers) and that lower expression is related to worse lung function [10]. In addition, bronchial MRP1 expression was higher during smoking than after one year of smoking cessation (Van der Deen et al., submitted). The expression of MRP1 measured with RT-PCR was similar in lungs of smokers and a combined group of ex- and non-smokers, suggesting that current smoking does not affect MRP1 gene expression [53]. This was semi-quantitatively analysed and the power of this study was rather low (smokers n = 13 and ex- or non-smokers n = 8). Cigarette smoke extract is a complex mixture of many substances. One of these is nitrosamine 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK), a carcinogenic nitrosamine. NNK is converted intracellularly in NNAL-O-glucuronide which is transported by MRP1 in the presence of glutathione (GSH) [60]. Thus, functional activity of MRP1 in the lung may play an important role in the antioxidant defence against toxic compounds generated by cigarette smoke.
MRP1 polymorphisms
Polymorphisms of the MRP1 gene or in regulatory genes may influence function of MRP1 in the lung. MRP1 has been screened for genetic variations and several mutations have been found in the MRP1 gene in the human population [61,62]. Ethnic differences for MRP1 expression were observed between Caucasian and Japanese subjects but the clinical consequences have to be determined to date [63,64]. In preclinical models, a single mutation in the MRP1 gene can result in loss of transport of some, but not all, MRP1 substrates [65] which implies that potentially minor differences in the MRP1 gene may result in aberrant functional capacities of the MRP1 protein. A low frequency (<1%) naturally occurring mutation in MRP1 was related to functional differences in organic anion transport and drug resistance [66]. Therefore, some individuals could be more susceptible to a set of xenobiotics than others.
MRP1 in animal models
Scheffer et al. detected, besides P-gp, high levels of Mrp1 in lungs of mice [4]. Mrp1 (-/-) mice develop normally [67]. Viability, fertility, and a range of histological, hematological, and serum-chemical parameters were similar in Mrp (+/+) and Mrp (-/-) mice. However, Mrp1 (-/-) mice are hypersensitive to exposure to several drugs e.g. etoposide, resulting in loss of body weight and death [56,67,68]. In Mrp1 deficient mice there were no gross abnormalities in lungs and other tissues after treatment with the chemotherapeutic drug etoposide-phosphate, but abnormal mucous production was seen around the mouth. Further examination of tissues showed that these mice suffered from oropharyngeal toxicity [56]. In Mrp1 (-/-) mice, GSH levels were elevated in e.g. lung, kidney, heart, testes and skeletal muscle. In organs that express little MRP1, such as the liver and small intestine, GSH levels were unchanged [67,69]. The increase in GSH was not related to increased levels of gamma-glutamylcysteine synthetase (γ-GCS), the rate-limiting enzyme for production of the tripeptide GSH. This suggests that MRP1 plays a role in GSH metabolism because MRP1 expression determines cellular GSH levels which is γ-GCS independent. In liver tissue, Mrp2 and Mrp5 mRNA levels were increased in Mrp1(-/-) mice compared to wild-type mice, probably due to a compensation mechanism. Mdr1a and Mrp3 levels were unchanged in these animals. Unfortunately, lung tissue was not examined [69]. Mrp1 (-/-) mice showed impaired inflammatory responses accompanied by a decreased LTC4 excretion by leukotriene producing cells [67,68]. The outgrowth of Mycobacterium tuberculosis was enhanced in Mrp1 (-/-) mice, but there was no difference in survival compared to wild-type mice [70]. Strikingly, Schultz et al. [71] observed that the survival of Mrp1 (-/-) mice inoculated with Streptococcus pneumoniae was better than of wild-type mice. This was accompanied by a lower LTC4 concentration but a higher LTB4 level in bronchoalveolar lavage fluid (BALF). Treatment with an LTB4 antagonist abolished the positive effect on survival rate. Altering MRP1 function may therefore be a target for further studies on treatment of pneumonia. Interestingly, LTB4 levels are also elevated in sputum and exhaled breath condensate of COPD patients [72,73]. This is thought to result from a higher amount of neutrophils present, yet also lower functional MRP1 expression may play a role, according to the observations in Mrp1 (-/-) mice [10,71]. The pulmonary and hepatic carcinogen aflatoxin B1 (a mycotoxin) and its GSH conjugate are MRP1 substrates in vitro [74]. MRP1 may therefore play an important role in detoxification of aflatoxin B1 in the lung. No difference in occurrence of lung tumours (and liver tumours) 12 months after an 8 week exposure to aflatoxin B1 was observed in Mrp1 (-/-) and wild-type mice [75]. This observation may be explained by differences in exposure time compared to humans (who may be chronically exposed for many years) or by redundancy of other ABC transporters in knock-out mice to export aflatoxin B1 and its conjugates.
NRF2 (nuclear factor-E2 p45-related factor), a transcription factor for many genes that play a role in antioxidant defence and detoxification processes, was recently identified as a transcription factor for MRP1 [76]. Interestingly, the onset of cigarette smoke-induced emphysema was earlier and the extent of emphysema was more severe in Nrf2 (-/-) mice as compared to wild-type mice [77]. The number of inflammatory cells, mainly consisting of macrophages, in bronchoalveolar lavage fluid and lung tissue was higher in Nrf2 (-/-) mice, accompanied by more extensive apoptosis of endothelial cells and alveolar type II cells. Forty-five Nrf2-dependent protective genes were induced by smoke exposure in wild-type mice compared to knock-out mice as measured with microarray analysis. Regretfully, Mrp1 was not present on this gene expression array. Similar smoke exposure experiments with Mrp1 (-/-) mice would be interesting to investigate if these mice are also more susceptible to develop emphysema. Three polymorphisms were identified in the promotor region of NRF2 in healthy individuals (n = 81) and in patients with COPD (n = 87) and sytemic lupus erythematosis(n = 51) (all Japanese individuals) [78]. There was no relation between these polymorphisms and risk of these diseases, but the power of this study was rather low.
Mdr1a/1b(-/-)/Mrp1(-/-) triple knockout (TKO) mice are indistinguishable from wild-type mice under normal physiological conditions [79,80]. Intravenous injection of [3H]etoposide in TKO mice resulted after 4 hours in a higher accumulation in brown adipose tissue, colon, urogenital tract, salivery gland and heart when compared to Mdr1a/1b (-/-) mice. The accumulation of [3H]etoposide in lung did not increase [79]. Accumulation of [3H]vincristine was 26-fold higher in lungs of weaning TKO mice compared to adult TKO mice, 4 hours after a single dose intraperitoneal injection [81]. In TKO compared to wild-type mice, [3H]vincristine was 82.7 and 5.8-fold increased in respectively weaning and adult mice after 4 hours in this experimental setting. Eight hours post-injection, these ratios were 33 and 51.1. Thus, accumulation of vincristine is both time and age-related. In another study, toxicity of the chemotherapeutic drugs vincristine and etoposide was elevated in TKO mice (respectively 128-fold and 3–5 fold) compared to wild-type mice. Lung tissue was not further examined [80].
MRP1 in vitro
MRP1 overexpression was first described in the human lung cancer cell line H69AR [46]. The human SCLC cell line GLC4/ADR, which displays multiple copies of the MRP1 gene, is often used to study MRP1 function [47]. The lung cell line Calu-3 (from bronchiolar adenocarcinoma/glandular origin) has been reported to express MRP1 basolaterally and has functional MRP1 activity. It may serve as a good model for in vitro analysis of transport activities [44]. Besides P-gp, MRP1 protein was present in freshly obtained primary human bronchial epithelial cells. Efflux of MRP1 substrate carboxy-dichlorofluorescein could be measured in these cells and was inhibited by MRP modulator MK571 [45]. In contrast to upregulation of P-gp, epithelial cells did not respond with MRP1 upregulation to paraquat and MRP1 levels were stable in time until 12 weeks, but highly increased after 18 weeks culturing. An interesting observation was made by Bandi et al. [82] who incubated the airway epithelial cell line Calu-1 with budesonide, an anti-asthma drug. MRP1 expression decreased after 7 and 14 days incubation with budesonide (10 μM) but not at day 1 and 4. MRP1 function was also diminished after 14 days incubation with increasing amounts of budesonide (1, 10 and 100 μM). It was therefore proposed that budesonide could be used as a chemosensitiser in lung tumours. The immortalised bronchial epithelial cell line 16HBE14o- expresses high MRP1 protein levels and activity [83]. MRP1 activity, but not P-gp, was blocked by incubation with cigarette smoke extract. Thus, the expression of MRP1 in human lungs may contribute to defence mechanisms of toxic inhaled substances present in cigarette smoke to protect against pulmonary diseases such as lung cancer or COPD. It was shown that benzo [a]pyrene (B [a]P), a polycyclic aromatic hydrocarbon that is a.o. present in cigarette smoke and car exhaust fuels, increased the efflux of monochlorobimane GSH conjugate (mBCl-SG, which is an MRP substrate) out of primary rabbit alveolar type II cells [84]. This was explained by an increase in GSH levels. MK571 reduced the efflux of mBCL-SG, suggesting that the transport was MRP-mediated. Surprisingly, the transport direction of this substrate was higher to the apical than to the basolateral compartment. This may imply that MRP1 (or other MRPs) is apically located in alveolar type II cells in contrast to the basolateral location that is usually reported for MRP1, or that another transporter is responsible for this unexpected result. MRP1-5 expression was studied in primary human lung cell cultures and in A549 cells [85]. MRP1 and MRP3 expression were membrane-associated whereas MRP2, 4, and 5 were located in intracellular structures in primary cells (both bronchial and peripheral epithelial cells). In A549 cells, all transporters were expressed in the cellular membrane. The fluorescent microscopic pictures, however, did not precisely show the exact cellular localisation and in several cases, the Golgi apparatus also seems to stain beside the cellular membrane.
MRP2
MRP2 localisation and function
MRP2 (ABCC2) is located on chromosome 10q24.2 and is also named the canalicular multispecific organic anion transporter (cMOAT) in the liver [86]. MRP2 and MRP1 share very similar substrate specificities. High affinity endogenous substrates for MRP2 include amphiphilic anions, such as LTC4 [87,88]. and bilirubin glucuronosides [89,90]. Data on the presence of MRP2 protein in the lung are conflicting. Only one out of three antibodies stained positive at the apical side in bronchial epithelium (Figure 1) [4,90]. If MRP2 is expressed apically, it could be involved in transport of noxious compounds into the lumen of the lung.
MRP2 in animal models
There are several rat strains that have a mutation in Mrp2/cMOAT and the phenotypes resemble that of the human Dubin-Johnson syndrome, a disease in which MRP2 is mutated. Examples are the GY/TR- rat [91] and the Eisai hyperbilirubinemic rat (EHBR) [92]. Subjects with the Dubin-Johnson syndrome and also Mrp2 deficient rats display jaundice and have impaired organic anion transport. Recently, it was shown that deficient Mrp2 function in EHBR rats was compensated by upregulation of Mrp3 in liver and kidney [93]. Mrp2 (-/-) mice have not been described. There are no reports in the literature of pulmonary malfunction in rats or humans with defects in the MRP2 gene.
MRP3
The MRP3 (ABCC3) gene is the closest MRP1 homologue [94] and is located on chromosome 17q21.33. It is involved in resistance against the anti-cancer drugs etoposide, teniposide and at higher concentrations also to methotrexate [95]. MRP3 is located in basolateral plasma membranes of liver, adrenals, pancreas, kidney, gut and gallbladder. The physiological function of MRP3 is still unknown but its cellular localisation and the information on substrates implies a role for MRP3 in transport of organic anions from the liver into the blood, especially when secretion into bile is being blocked [96]. MRP3 protein and RNA was not detected in bronchial epithelium [4,96]. However, data in literature are conflicting and preliminary unpublished results indicate that Mrp3 is present in mouse lung. As already mentioned in the discussion on MRP1, primary epithelial cells (of bronchial and peripheral origin) and A549 cells stain positive for MRP3 (Figure 1) [85]. MRP3 mRNA was measured by quantitative RT-PCR in normal lung and lung tumours. MRP3 transcript levels were related to exposure to vincristine, etoposide and platinum drugs in lung tumours [97,98].
MRP4 and MRP5
MRP4 (ABCC4) and MRP5 (ABCC5) genes map to 13q32.1 and 3q27.1 respectively. MRP4 protein is highly expressed in the kidney and prostate and MRP4 mRNA is present in the lung [55,99]. MRP5 mRNA is ubiquitously expressed, mainly in brain and skeletal muscle and also in the lung [6,55]. Primary epithelial cells (of bronchial and peripheral origin) and A549 cells stain positive for MRP4 and 5 (Figure 1) [85]. The substrate specificities of these two transporters differ from the other transporters of the ABCC family, i.e. they transport a variety of nucleoside analogues. The physiological role of both proteins is still unknown but they can serve as an efflux pump of the nucleosides cAMP and cGMP at low affinity, likely in a GSH independent manner [100-103]. Interestingly, AMP levels in lungs of asthma and COPD patients are elevated and therefore, MRP4 and MRP5 activity may be of clinical relevance in these diseases [104]. MRP4 can actively efflux prostaglandins E1 and E2 [105]. It was also suggested that MRP4 plays a role in the transport of conjugated steroids and bile acids [106], whereas MRP5 was reported to transport hyaluran out of cells [107]. In a recent study, a panel of 60 human cancer cell lines (the NCI-60) were screened with real-time RT-PCR for 48 human ABC transporters. Among these were several lung cancer cell lines. An association was found between MRP4 and MRP5 gene expression and resistance against platinum drugs in lung cancer [108-110].
MRP6
MRP6 (ABCC6) maps to chromosome 16p13.12 and is mainly expressed in liver and kidney. MRP6 was neither detected in the lung nor in lung derived tumour cell lines SW1573 and GLC4 by immunohistochemistry [111]. However, MRP6 mRNA was moderately present in human lung extracts [6]. MRP6 is mutated in the hereditary connective tissue disorder pseudoxanthoma elasticum which affects skin, retina and blood vessels (for review, see [112]). Still, its physiological role in this disease and its substrate specificity is unclear. Pulmonary abnormalities are rare in pseudoxanthoma elasticum. However, in some patients calcification and elastic tissue damage in the lung have been described [113]. This may be due to an altered MRP6 function. Further studies are required to assess whether dysfunction of MRP6 also plays a role in development of other lung diseases such as emphysema. With in situ hybridisation and RNase protection assay in C57BL/6 mice, Mrp6 mRNA could be detected in tracheal and bronchial epithelium [114].
MRP7, MRP8 and MRP9
To date, the functions of MRP7 (ABCC10, gene 6p21.1), MRP8 (ABCC11, gene 16q12.1) and MRP9 (ABCC12, gene 16q12.1) are largely unknown. Of these recently discovered ABC transporters, only MRP7 mRNA is highly expressed in total lung and trachea RNA extracts [6]. MRP7 function resembles P-gp function in the resistance against taxanes [115]. MRP8 resembles MRP4 function more than MRP5, and the physiological role of MRP8 may involve transport of conjugated steroids, cyclic nucleotides and bile acids [116].
CFTR
CFTR localisation and function
CFTR (ABCC7) is located on chromosome 7q31.31. CFTR is the only member of the ABC superfamily which is not an active transporter. It functions as a chloride channel and in normal human airway tissue CFTR is highly expressed at the luminal side in serous cells of the submucosal glands. In addition, it is restricted to the apical membrane domain of well-differentiated epithelial cells such as ciliated cells, and probably also non-ciliated Clara cells and alveolar type I cells (Figure 1) [117-120]. CFTR is also expressed in normal nasal respiratory mucosa [13].
CFTR in cystic fibrosis
Mutations in the CFTR gene can cause cystic fibrosis [7] and are associated with abnormal Cl- and Na+ ion transport in several tissues including the lungs, pancreas, gastrointestinal tract, liver, sweat glands and male reproductive organs. Although the normal expression of CFTR in the lung is lower compared to tissues such as the intestine and pancreas, its function in the lung is of major importance. The most frequent mutation in this gene is the delta F508 mutation which leads to cystic fibrosis. Defective CFTR function can cause viscous secretions in the lungs which leads to chronic inflammation with acute exacerbations by impaired mucociliary clearance. There are major risks for colonisation with Pseudonomas aeruginosa which leads to pneumonia and respiratory insufficiency [121]. Besides abnormal CFTR localisation and expression in cystic fibrosis, also in non-cystic fibrosis airway tissue CFTR can be abnormally expressed in remodelled or dedifferentiated epithelium [122,123]. whereas in delta F508 CFTR epithelial cells there may be a normally processed CFTR [120,124]. The regulation and transport function of CFTR are dependent of the state of differentiation and polarisation of epithelial cell cultures [125,126]. Dedifferentiation with hyperplasia or metaplasia was associated with an intracellular localisation or absence of CFTR protein [122]. Hurbain et al. [127] analysed MRP1-5 transcript levels in nasal respiratory cells from cystic fibrosis patients with homozygous delta F508 mutation. Surprisingly, low MRP1 levels were associated with more severe disease and in addition, MRP1 levels were related to cAMP-independent chloride transport suggesting that MRP1 regulates another chloride channel in the apical membrane. CFTR function is cAMP regulated [121], and therefore, MRP4 or MRP5 activity may play a role in regulation of CFTR, since these proteins transport cAMP. Another study showed that CFTR function is blocked by two MRP substrates (taurolithocholate-3-sulphate and beta-estradiol) [128]. This implies that these two substances are also substrates for CFTR, or that CFTR and MRP proteins possess similar anion binding sites. Bebok et al. showed that nitric oxide (NO) and reactive oxygen nitrogen species (RONS) decrease wild-type CFTR protein levels in airway epithelial cell monolayers [129]. Natural sources of NO and RONS are activated in alveolar and interstitial macrophages [130,131], neutrophils [132], alveolar type II cells [133,134]. and airway epithelial cells [135]. In this view, CFTR function may be important in pulmonary infections and in the effect of oxidative stress generated by e.g. cigarette smoke.
CFTR in animal models
Cftr knock-out mice show reduced viability in contrast to most other ABC transporter knock-outs which are viable and fertile. Trezise et al. developed Cftr knock-out mice that have a severe cystic fibrosis phenotype accompanied by a lack of Cftr-related chloride conductance in e.g. tracheal epithelium [136,137]. They found that the P-gp mRNA level was four-fold increased in intestines of neonatal and 3- to 4-week-old Cftr knock-out mice compared to littermates of the same age. However, in 10 weeks-old mice P-gp levels were three-fold decreased. Apparently, a reduction or loss of Cftr function influences P-gp expression.
CFTR in vitro
The Calu-3 cell line, which has properties from serous cells of the pulmonary submucosal glands, is often used to study function and expression of CFTR [129,138]. It was demonstrated in Calu-3 cells that CFTR mRNA expression is downregulated after ouabain incubation whereas P-gp expression was upregulated [139]. The bronchial epithelial cell line 16HBE14o- is also invaluable in CFTR research [129,140]. In wild-type and mutant CFTR expressing Sf9 insect cells, it was demonstrated that besides chloride transport, another function of CFTR is transport of GSH in a nucleotide-dependent manner [141]. This observation suggests that CFTR plays a role in the control of oxidative stress. Indeed, several studies in patients, mice and cell lines have shown that GSH levels are lower in case of defective CFTR function.
BCRP
BCRP localisation and function
BCRP (ABCG2) is the breast cancer resistance protein (BCRP), located on chromosome 4q22. BCRP is a half-transporter that probably acts as a homo- or heterodimer [142] and is involved in resistance against toxins and several chemotherapeutic agents (e.g. mitoxantrone and topoisomerase 1 inhibitors) [143,144]. Protein levels of BCRP in the lung are lower than P-gp and MRP1 but distinct in the epithelial cell layer and in seromucinous glands (Figure 1) [4]. BCRP was absent in alveolar macrophages, suggesting that BCRP does not play a major role in innate inflammatory responses in the lung. Small endothelial capillaries also stain positive for BCRP, thus BCRP may protect the lungs against noxious compounds that enter systemically.
BCRP in tumours
In a study of untreated solid lung tumours, i.e. squamous cell carcinoma (n = 5), adenocarcinoma (n = 2) and SCLC (n = 3), most cases expressed moderate or strong BCRP [145]. In addition, higher BCRP expression in blood vessels was observed than in vessels of the surrounding tumour, indicating that BCRP may play a role in tumour angiogenesis. In 72 cases of advanced NSCLC, BCRP expression, but not P-gp, MRP1, MRP2 and MRP3, predicted poor clinical outcome [146].
BCRP in animal models
Bcrp knock-out mice do not display any abnormalities compared to wild-type mice under normal conditions, except for the colour of their bile which is red instead of yellow on certain diets [147]. However, these mice appear to be extremely sensitive to the phototoxin pheophorbide A, showing that BCRP is important to protect against toxic food components. Abnormalities of lungs in Bcrp (-/-) mice have not been reported thus far. Recently, Bcrp1 (Bcrp in the mouse is also called Bcrp1) localisation in the lung of 4- to 8-week old C57Bl/6J mice was studied to identify lung stem cells. These cells possess high activity of efflux of Hoechst dye that is transported by Bcrp1 [148]. In peripheral blood and bone marrow, cells with side population activity represent BCRP positive (haematopoietic) stem cells and this might also be the case for side population cells in the lungs. Indeed, Bcrp1 (-/-) mice did not display side population cells as shown by Hoechst staining of lung digests. In paraffin-embedded tissue sections, Bcrp1 was restricted to smooth muscle cells (of both arteries and airways) and a subpopulation of unidentified round cells in the alveolar space but not detectable in endothelial or epithelial cells. BAL cells of mice were Bcrp1 positive, but did not display side population (SP) activity [148]. Clara cells, a nonciliated bronchiolar epithelial cell type, were also phenotypically described to possess stem cell characteristics [149]. Further investigation is required to identify the stem cell pool in the lung in relation to expression of ABC transporters. Repairing and replacing lost lung tissue is a research area of promising therapeutic possibilities [150].
ABCA1
The ABCA1 gene is mapped to chromosome 9q31.1. It is the causative gene in the development of Tangier disease, a disorder of cholesterol transport [8]. The ABCA1 protein controls transport of cholesterol and phospholipids to apolipoprotein 1 (apoA-1) in alveolar type II cells (Figure 1). Using the siRNA technique for ABCA1, it was shown that ABCA1 is involved in basolateral transport of surfactant that is activated by oxysterol [151]. Abca1 knock-out mice show increased concentrations of cholesterol precursors in lung, plasma, intestine and faeces [152]. It was demonstrated that there were major morphologic abnormalities in lungs of these mice, increasing with age. In 30% of 18 months old mice, lung parenchyma was affected. Lesions were characterised by foamy type II pneumocytes with aberrant lamellar bodies, intraalveolar macrophages and cholesterol clefts [153]. In addition, ABCA7, a close homologue of ABCA1 was also found to be highly expressed in mouse lung tissue by Western blot analysis [154].
ABCA3
The ABCA3 gene is located on chromosome 16p13.3. The function of this ABC transporter has not been studied in detail but ABCA3 protein is present in lamellar bodies in human lung alveolar type II cells (Figure 1). Langmann et al. performed quantitative RT-PCR in 20 human tissues and found ABCA3 to be expressed restrictively in the lung [6]. In a recent study in cell lines, an association was found between lung cancer and ABCA3 (and also ABCA2) gene expression by means of quantitative RT-PCR [108].
The high expression of ABCA3 in alveolar type II cells suggests that ABCA3 may play a role in surfactant regulation [155]. Surfactant is important to lower the surface tension in the air-liquid interphase in alveoli. Indeed, in patients with surfactant deficiency and with severe neonatal lung disease it was demonstrated that the ABCA3 gene was frequently mutated [9]. ABCA3 mutations were found in 16 out of 21 patients, but polymorphisms (SNPs) were not found. One mutation was not fatal but was associated with a chronic lung disorder in a 6-year old patient. Other mutations turned out to be fatal and these patients died shortly after birth. Electron micrographs of tissue of patients with surfactant deficiency showed abnormal dense and small lamellar bodies. It was suggested that ABCA3 plays a role in phospholipid metabolism, since it closely resembles ABCA1 and ABCA4 that are known to transport phospholipids.
Conclusion
Little is known about the function of ABC transporters that are expressed in the lung although their overall expression is very high compared to many other organs [6]. This review shows that ABC transporters in the lung are not only relevant for relatively rare diseases such as cystic fibrosis, Tangier disease and surfactant deficiency. Preliminary data indicate that MRP1 expression is lower in COPD patients than in healthy controls. Mutations and polymorphisms in ABC transporters may have important clinical consequences for development of lung diseases. However, overlap in substrate specificities may be compensatory in cases of malfunction of one (or more) transporter(s). Given the complexity of lung architecture, research on detailed cellular processes is difficult but challenging. Several ABC transporter deficient animal models have been developed that are of great value to study the role of these proteins. To date, exposure to cigarette smoke has never been tested in ABC transporter deficient animal models and would potentially give interesting information about the role of ABC transporters in protection against inhaled toxic substances such as present in tobacco smoke. Cell line models have been used to study transport processes and pulmonary drug metabolism. The delivery of pulmonary drugs to the site of action is probably highly dependent on the presence and activity of many ABC transporters in several cell types in the lung. The first barrier after inhalation is the pulmonary epithelium and transporters in the pulmonary endothelium may be critical for the delivery of intravenously or orally administered drugs. Insight in the function of ABC transporters in the lung may open new ways to facilitate treatment of lung diseases.
Abbreviations
ABC = ATP-binding cassette
AMP = adenosine monophosphate
BALF = bronchoalveolar lavage fluid
B [a]P = benzo [a]pyrene
BCRP = breast cancer resistance protein
cAMP = cyclic adenosine monophosphate
CFTR = cystic fibrosis transmembrane conductance regulator
cGMP = cyclic guanosine monophosphate
cMOAT = canalicular multispecific organic anion transporter
CNS = central nervous sytem
COPD = chronic obstructive pulmonary disease
EHBR = Eisai hyperbilirubinemic rat
γ-GCS = gamma-glutamylcysteine synthetase
GSH = glutathione, reduced form
GSSG = glutathione disulfide; oxidized glutathione
GST = glutathione S-transferase
HPLC = high liquid performance chromatography
LTC4 = leuktotriene C4
LTB4 = leukotriene B4
MDR = multidrug resistance
MRP = multidrug resistance-associated protein
NSCLC = non-small cell lung cancer
NNAL = nitrosamine 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol
NNK = nitrosamine 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone
NO = nitric oxide
NRF2 = nuclear factor-E2 p45-related factor
P-gp = P-glycoprotein
R6G = rhodamine 6G
RONS = reactive oxygen nitrogen species
RT-PCR = reverse transcriptase-polymerase chain reaction
SCLC = small cell lung cancer
SNP = single nucleotide polymorphism
TKO = triple knock-out mice
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MD mainly drafted the manuscript, EV, WT, RS, HT and DP helped to draft the manuscript. All authors read and approved the final manuscript.
Acknowledgements
This work was supported by a grant from the Netherlands Asthma Foundation (NAF97.35) and the "Stichting Astma Bestrijding" (SAB).
==== Refs
Crapo JD Barry BE Gehr P Bachofen M Weibel ER Cell number and cell characteristics of the normal human lung Am Rev Respir Dis 1982 125 740 745 7091881
Dean M Hamon Y Chimini G The human ATP-binding cassette (ABC) transporter superfamily J Lipid Res 2001 42 1007 1017 11441126
Muller M Human ABC-transporters
Scheffer GL Pijnenborg AC Smit EF Muller M Postma DS Timens W van der Valk P de Vries EG Scheper RJ Multidrug resistance related molecules in human and murine lung J Clin Pathol 2002 55 332 339 11986335
Hamilton KO Yazdanian MA Audus KL Contribution of efflux pump activity to the delivery of pulmonary therapeutics Curr Drug Metab 2002 3 1 12 11876574 10.2174/1389200023338170
Langmann T Mauerer R Zahn A Moehle C Probst M Stremmel W Schmitz G Real-time reverse transcription-PCR expression profiling of the complete human ATP-binding cassette transporter superfamily in various tissues Clin Chem 2003 49 230 238 12560344 10.1373/49.2.230
Riordan JR Rommens JM Kerem B Alon N Rozmahel R Grzelczak Z Zielenski J Lok S Plavsic N Chou JL Identification of the cystic fibrosis gene: cloning and characterization of complementary DNA Science 1989 245 1066 1073 2475911
Brooks-Wilson A Marcil M Clee SM Zhang LH Roomp K van Dam M Yu L Brewer C Collins JA Molhuizen HO Loubser O Ouelette BF Fichter K Ashbourne-Excoffon KJ Sensen CW Scherer S Mott S Denis M Martindale D Frohlich J Morgan K Koop B Pimstone S Kastelein JJ Hayden MR Mutations in ABC1 in Tangier disease and familial high-density lipoprotein deficiency Nat Genet 1999 22 336 345 10431236 10.1038/11905
Shulenin S Nogee LM Annilo T Wert SE Whitsett JA Dean M ABCA3 gene mutations in newborns with fatal surfactant deficiency N Engl J Med 2004 350 1296 1303 15044640 10.1056/NEJMoa032178
Van der Deen M Marks H Muller M Pijnenborg AC Postma DS Scheffer G Scheper RJ Smit EF Timens W de Vries EG Diminished expression of Multidrug Resistance-associated Protein 1 (MRP1) in bronchial epithelium of COPD patients [abstract] Am J Respir Crit Care Med 2003 167 A397
Gottesman MM Fojo T Bates SE Multidrug resistance in cancer: role of ATP-dependent transporters Nat Rev Cancer 2002 2 48 58 11902585 10.1038/nrc706
Lechapt-Zalcman E Hurbain I Lacave R Commo F Urban T Antoine M Milleron B Bernaudin JF MDR1-Pgp 170 expression in human bronchus Eur Respir J 1997 10 1837 1843 9272928 10.1183/09031936.97.10081837
Wioland MA Fleury-Feith J Corlieu P Commo F Monceaux G Lacau-St-Guily J Bernaudin JF CFTR, MDR1, and MRP1 immunolocalization in normal human nasal respiratory mucosa J Histochem Cytochem 2000 48 1215 1222 10950878
Campbell L Abulrob AN Kandalaft LE Plummer S Hollins AJ Gibbs A Gumbleton M Constitutive expression of p-glycoprotein in normal lung alveolar epithelium and functionality in primary alveolar epithelial cultures J Pharmacol Exp Ther 2003 304 441 452 12490621 10.1124/jpet.102.042994
Cordon-Cardo C O'Brien JP Boccia J Casals D Bertino JR Melamed MR Expression of the multidrug resistance gene product (P-glycoprotein) in human normal and tumor tissues J Histochem Cytochem 1990 38 1277 1287 1974900
Van der Valk P van Kalken CK Ketelaars H Broxterman HJ Scheffer G Kuiper CM Tsuruo T Lankelma J Meijer CJ Pinedo HM Distribution of multi-drug resistance-associated P-glycoprotein in normal and neoplastic human tissues. Analysis with 3 monoclonal antibodies recognizing different epitopes of the P-glycoprotein molecule Ann Oncol 1990 1 56 64 1706611
Puddu P Fais S Luciani F Gherardi G Dupuis ML Romagnoli G Ramoni C Cianfriglia M Gessani S Interferon-gamma up-regulates expression and activity of P-glycoprotein in human peripheral blood monocyte-derived macrophages Lab Invest 1999 79 1299 1309 10532592
Valverde MA Diaz M Sepulveda FV Gill DR Hyde SC Higgins CF Volume-regulated chloride channels associated with the human multidrug-resistance P-glycoprotein Nature 1992 355 830 833 1371598 10.1038/355830a0
Gill DR Hyde SC Higgins CF Valverde MA Mintenig GM Sepulveda FV Separation of drug transport and chloride channel functions of the human multidrug resistance P-glycoprotein Cell 1992 71 23 32 1382860 10.1016/0092-8674(92)90263-C
Higgins CF P-glycoprotein and cell volume-activated chloride channels J Bioenerg Biomembr 1995 27 63 70 7543087
Kunzelmann K Slotki IN Klein P Koslowsky T Ausiello DA Greger R Cabantchik ZI Effects of P-glycoprotein expression on cyclic AMP and volume-activated ion fluxes and conductances in HT-29 colon adenocarcinoma cells J Cell Physiol 1994 161 393 406 7962123 10.1002/jcp.1041610302
De Greef C Sehrer J Viana F van Acker K Eggermont J Mertens L Raeymaekers L Droogmans G Nilius B Volume-activated chloride currents are not correlated with P-glycoprotein expression Biochem J 1995 307 713 718 7741701
Waters CM Krejcie TC Avram MJ Facilitated uptake of fentanyl, but not alfentanil, by human pulmonary endothelial cells Anesthesiology 2000 93 825 831 10969317 10.1097/00000542-200009000-00033
Wood L Palmer M Hewitt J Urtasun R Bruera E Rapp E Thaell JF Results of a phase III, double-blind, placebo-controlled trial of megestrol acetate modulation of P-glycoprotein-mediated drug resistance in the first-line management of small-cell lung carcinoma Br J Cancer 1998 77 627 631 9484821
Beer TW Rowlands DC Crocker J Detection of the multidrug resistance marker P-glycoprotein by immunohistochemistry in malignant lung tumours Thorax 1996 51 526 529 8711682
Volm M Mattern J Samsel B Relationship of inherent resistance to doxorubicin, proliferative activity and expression of P-glycoprotein 170, and glutathione S-transferase-pi in human lung tumors Cancer 1992 70 764 769 1353700
Volm M Mattern J Samsel B Overexpression of P-glycoprotein and glutathione S-transferase-pi in resistant non-small cell lung carcinomas of smokers Br J Cancer 1991 64 700 704 1680367
Hoffmeyer S Burk O von Richter O Arnold HP Brockmoller J Johne A Cascorbi I Gerloff T Roots I Eichelbaum M Brinkmann U Functional polymorphisms of the human multidrug-resistance gene: multiple sequence variations and correlation of one allele with P-glycoprotein expression and activity in vivo Proc Natl Acad Sci U S A 2000 97 3473 3478 10716719 10.1073/pnas.050585397
Eichelbaum M Fromm MF Schwab M Clinical aspects of the MDR1 (ABCB1) gene polymorphism Ther Drug Monit 2004 26 180 185 15228162 10.1097/00007691-200404000-00017
Johannesson M Nordqvist AC Bogdanovic N Hjelte L Schalling M Polymorphic expression of multidrug resistance mRNA in lung parenchyma of nonpregnant and pregnant rats: a comparison to cystic fibrosis mRNA expression Biochem Biophys Res Commun 1997 239 606 611 9344878 10.1006/bbrc.1997.7519
Sinues B Fanlo A Bernal ML Mayayo E Bello S Rubio E Isla D MDR-1 C3435T genetic polymorphism and tobacco-related lung cancer Oncology 2003 64 183 185 12566917 10.1159/000067770
Isla D Sarries C Rosell R Alonso G Domine M Taron M Lopez-Vivanco G Camps C Botia M Nunez L Sanchez-Ronco M Sanchez JJ Lopez-Brea M Barneto I Paredes A Medina B Artal A Lianes P Single nucleotide polymorphisms and outcome in docetaxel-cisplatin-treated advanced non-small-cell lung cancer Ann Oncol 2004 15 1194 1203 15277258 10.1093/annonc/mdh319
Zheng H Zeevi A Schuetz E Lamba J McCurry K Griffith BP Webber S Ristich J Dauber J Iacono A Grgurich W Zaldonis D McDade K Zhang J Burckart GJ Tacrolimus dosing in adult lung transplant patients is related to cytochrome P4503A5 gene polymorphism J Clin Pharmacol 2004 44 135 140 14747421 10.1177/0091270003262108
Brady JM Cherrington NJ Hartley DP Buist SC Li N Klaassen CD Tissue distribution and chemical induction of multiple drug resistance genes in rats Drug Metab Dispos 2002 30 838 844 12065443 10.1124/dmd.30.7.838
Seree E Villard PH Hever A Guigal N Puyoou F Charvet B Point-Scomma H Lechevalier E Lacarelle B Barra Y Modulation of MDR1 and CYP3A expression by dexamethasone: evidence for an inverse regulation in adrenals Biochem Biophys Res Commun 1998 252 392 395 9826540 10.1006/bbrc.1998.9662
Hendrikse NH de Vries EG Eriks-Fluks L van der Graaf WT Hospers GA Willemsen AT Vaalburg W Franssen EJ A new in vivo method to study P-glycoprotein transport in tumors and the blood-brain barrier Cancer Res 1999 59 2411 2416 10344751
Roerig DL Audi SH Ahlf SB Kinetic characterization of p-glycoprotein-mediated efflux of rhodamine 6g in the intact rabbit lung Drug Metab Dispos 2004 32 953 958 15319336 10.1124/dmd.104.000042
Booth CL Brouwer KR Brouwer KL Effect of multidrug resistance modulators on the hepatobiliary disposition of doxorubicin in the isolated perfused rat liver Cancer Res 1998 58 3641 3648 9721873
Schinkel AH Mayer U Wagenaar E Mol CA van Deemter L Smit JJ van der Valk MA Voordouw AC Spits H van Tellingen O Zijlmans JM Fibbe WE Borst P Normal viability and altered pharmacokinetics in mice lacking mdr1-type (drug-transporting) P-glycoproteins Proc Natl Acad Sci U S A 1997 94 4028 4033 9108099 10.1073/pnas.94.8.4028
Grauer MT Uhr M P-glycoprotein reduces the ability of amitriptyline metabolites to cross the blood brain barrier in mice after a 10-day administration of amitriptyline J Psychopharmacol 2004 18 66 74 15107187 10.1177/0269881104042831
Forbes II Human airway epithelial cell lines for in vitro drug transport and metabolism studies Pharm Sci Technol Today 2000 3 18 27 10637597 10.1016/S1461-5347(99)00231-X
Ehrhardt C Kneuer C Laue M Schaefer UF Kim KJ Lehr CM 16HBE14o- human bronchial epithelial cell layers express P-glycoprotein, lung resistance-related protein, and caveolin-1 Pharm Res 2003 20 545 551 12739760 10.1023/A:1023230328687
Hamilton KO Yazdanian MA Audus KL Modulation of P-glycoprotein activity in Calu-3 cells using steroids and beta-ligands Int J Pharm 2001 228 171 179 11576779 10.1016/S0378-5173(01)00836-5
Hamilton KO Topp E Makagiansar I Siahaan T Yazdanian M Audus KL Multidrug resistance-associated protein-1 functional activity in Calu-3 cells J Pharmacol Exp Ther 2001 298 1199 1205 11504821
Lehmann T Kohler C Weidauer E Taege C Foth H Expression of MRP1 and related transporters in human lung cells in culture Toxicology 2001 167 59 72 11557130 10.1016/S0300-483X(01)00458-9
Cole SP Bhardwaj G Gerlach JH Mackie JE Grant CE Almquist KC Stewart AJ Kurz EU Duncan AM Deeley RG Overexpression of a transporter gene in a multidrug-resistant human lung cancer cell line Science 1992 258 1650 1654 1360704
Muller M Meijer C Zaman GJ Borst P Scheper RJ Mulder NH de Vries EG Jansen PL Overexpression of the gene encoding the multidrug resistance-associated protein results in increased ATP-dependent glutathione S-conjugate transport Proc Natl Acad Sci U S A 1994 91 13033 13037 7809167
Hipfner DR Deeley RG Cole SP Structural, mechanistic and clinical aspects of MRP1 Biochim Biophys Acta 1999 1461 359 376 10581367
Loe DW Deeley RG Cole SP Characterization of vincristine transport by the M(r) 190,000 multidrug resistance protein (MRP): evidence for cotransport with reduced glutathione Cancer Res 1998 58 5130 5136 9823323
Renes J de Vries EG Nienhuis EF Jansen PL Muller M ATP- and glutathione-dependent transport of chemotherapeutic drugs by the multidrug resistance protein MRP1 Br J Pharmacol 1999 126 681 688 10188979 10.1038/sj.bjp.0702360
Hooijberg JH Broxterman HJ Kool M Assaraf YG Peters GJ Noordhuis P Scheper RJ Borst P Pinedo HM Jansen G Antifolate resistance mediated by the multidrug resistance proteins MRP1 and MRP2 Cancer Res 1999 59 2532 2535 10363967
Leier I Jedlitschky G Buchholz U Keppler D Characterization of the ATP-dependent leukotriene C4 export carrier in mastocytoma cells Eur J Biochem 1994 220 599 606 8125120 10.1111/j.1432-1033.1994.tb18661.x
Brechot JM Hurbain I Fajac A Daty N Bernaudin JF Different pattern of MRP localization in ciliated and basal cells from human bronchial epithelium J Histochem Cytochem 1998 46 513 517 9524197
Flens MJ Zaman GJ van der Valk P Izquierdo MA Schroeijers AB Scheffer GL van der Groep P de Haas M Meijer CJ Scheper RJ Tissue distribution of the multidrug resistance protein Am J Pathol 1996 148 1237 1247 8644864
Kool M de Haas M Scheffer GL Scheper RJ van Eijk MJ Juijn JA Baas F Borst P Analysis of expression of cMOAT (MRP2), MRP3, MRP4, and MRP5, homologues of the multidrug resistance-associated protein gene (MRP1), in human cancer cell lines Cancer Res 1997 57 3537 3547 9270026
Wijnholds J Scheffer GL van der Valk M van der Valk P Beijnen JH Scheper RJ Borst P Multidrug resistance protein 1 protects the oropharyngeal mucosal layer and the testicular tubules against drug-induced damage J Exp Med 1998 188 797 808 9730882 10.1084/jem.188.5.797
Thomas GA Barrand MA Stewart S Rabbitts PH Williams ED Twentyman PR Expression of the multidrug resistance-associated protein (MRP) gene in human lung tumours and normal tissue as determined by in situ hybridisation Eur J Cancer 1994 30A 1705 1709 7833148 10.1016/0959-8049(94)00290-L
Doubre H Cesari D Mairovitz A Benac C Chantot-Bastaraud S Dagnon K Antoine M Danel C Bernaudin JF Fleury-Feith J Multidrug resistance-associated protein (MRP1) is overexpressed in DNA aneuploid carcinomatous cells in non-small cell lung cancer (NSCLC) Int J Cancer 2005 113 568 574 15472893 10.1002/ijc.20617
Trompier D Chang XB Barattin R du Moulinet d'Hardemare A Di Pietro A Baubichon-Cortay H Verapamil and its derivative trigger apoptosis through glutathione extrusion by multidrug resistance protein MRP1 Cancer Res 2004 64 4950 4956 15256468
Leslie EM Ito K Upadhyaya P Hecht SS Deeley RG Cole SP Transport of the beta -O-glucuronide conjugate of the tobacco-specific carcinogen 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol (NNAL) by the multidrug resistance protein 1 (MRP1). Requirement for glutathione or a non-sulfur-containing analog J Biol Chem 2001 276 27846 27854 11375986 10.1074/jbc.M102453200
Kerb R Hoffmeyer S Brinkmann U ABC drug transporters: hereditary polymorphisms and pharmacological impact in MDR1, MRP1 and MRP2 Pharmacogenomics 2001 2 51 64 11258197 10.1517/14622416.2.1.51
Saito S Iida A Sekine A Miura Y Ogawa C Kawauchi S Higuchi S Nakamura Y Identification of 779 genetic variations in eight genes encoding members of the ATP-binding cassette, subfamily C (ABCC/MRP/CFTR J Hum Genet 2002 47 147 171 12166651 10.1007/s100380200018
Oselin K Mrozikiewicz PM Gaikovitch E Pahkla R Roots I Frequency of MRP1 genetic polymorphisms and their functional significance in Caucasians: detection of a novel mutation G816A in the human MRP1 gene Eur J Clin Pharmacol 2003 59 347 350 12856092 10.1007/s00228-003-0625-z
Ito S Ieiri I Tanabe M Suzuki A Higuchi S Otsubo K Polymorphism of the ABC transporter genes, MDR1, MRP1 and MRP2/cMOAT, in healthy Japanese subjects Pharmacogenetics 2001 11 175 184 11266082 10.1097/00008571-200103000-00008
Ito K Olsen SL Qiu W Deeley RG Cole SP Mutation of a single conserved tryptophan in multidrug resistance protein 1 (MRP1/ABCC1) results in loss of drug resistance and selective loss of organic anion transport J Biol Chem 2001 276 15616 15624 11278867 10.1074/jbc.M011246200
Conrad S Kauffmann HM Ito K Leslie EM Deeley RG Schrenk D Cole SP A naturally occurring mutation in MRP1 results in a selective decrease in organic anion transport and in increased doxorubicin resistance Pharmacogenetics 2002 12 321 330 12042670 10.1097/00008571-200206000-00008
Lorico A Rappa G Finch RA Yang D Flavell RA Sartorelli AC Disruption of the murine MRP (multidrug resistance protein) gene leads to increased sensitivity to etoposide (VP-16) and increased levels of glutathione Cancer Res 1997 57 5238 5242 9393741
Wijnholds J Evers R van Leusden MR Mol CA Zaman GJ Mayer U Beijnen JH van der Valk M Krimpenfort P Borst P Increased sensitivity to anticancer drugs and decreased inflammatory response in mice lacking the multidrug resistance-associated protein Nat Med 1997 3 1275 1279 9359705 10.1038/nm1197-1275
Bain LJ Feldman RA Altered expression of sulfotransferases, glucuronosyltransferases and mrp transporters in FVB/mrp1-/- mice Xenobiotica 2003 33 1173 1183 14742140 10.1080/00498250310001609138
Verbon A Leemans JC Weijer S Florquin S van der PT Mice lacking the multidrug resistance protein 1 have a transiently impaired immune response during tuberculosis Clin Exp Immunol 2002 130 32 36 12296850 10.1046/j.1365-2249.2002.01952.x
Schultz MJ Wijnholds J Peppelenbosch MP Vervoordeldonk MJ Speelman P van Deventer SJ Borst P van der Poll T Mice lacking the multidrug resistance protein 1 are resistant to Streptococcus pneumoniae-induced pneumonia J Immunol 2001 166 4059 4064 11238654
Crooks SW Bayley DL Hill SL Stockley RA Bronchial inflammation in acute bacterial exacerbations of chronic bronchitis: the role of leukotriene B4 Eur Respir J 2000 15 274 280 10706491 10.1034/j.1399-3003.2000.15b09.x
Biernacki WA Kharitonov SA Barnes PJ Increased leukotriene B4 and 8-isoprostane in exhaled breath condensate of patients with exacerbations of COPD Thorax 2003 58 294 298 12668789 10.1136/thorax.58.4.294
Loe DW Stewart RK Massey TE Deeley RG Cole SP ATP-dependent transport of aflatoxin B1 and its glutathione conjugates by the product of the multidrug resistance protein (MRP) gene Mol Pharmacol 1997 51 1034 1041 9187270
Lorico A Nesland J Emilsen E Fodstad O Rappa G Role of the multidrug resistance protein 1 gene in the carcinogenicity of aflatoxin B1: investigations using mrp1-null mice Toxicology 2002 171 201 205 11836025 10.1016/S0300-483X(01)00584-4
Hayashi A Suzuki H Itoh K Yamamoto M Sugiyama Y Transcription factor Nrf2 is required for the constitutive and inducible expression of multidrug resistance-associated protein 1 in mouse embryo fibroblasts Biochem Biophys Res Commun 2003 310 824 829 14550278 10.1016/j.bbrc.2003.09.086
Rangasamy T Cho CY Thimmulappa RK Zhen L Srisuma SS Kensler TW Yamamoto M Petrache I Tuder RM Biswal S Genetic ablation of Nrf2 enhances susceptibility to cigarette smoke-induced emphysema in mice J Clin Invest 2004 114 1248 1259 15520857 10.1172/JCI200421146
Yamamoto T Yoh K Kobayashi A Ishii Y Kure S Koyama A Sakamoto T Sekizawa K Motohashi H Yamamoto M Identification of polymorphisms in the promoter region of the human NRF2 gene Biochem Biophys Res Commun 2004 321 72 79 15358217 10.1016/j.bbrc.2004.06.112
Wijnholds J deLange EC Scheffer GL van den Berg DJ Mol CA van der Valk M Schinkel AH Scheper RJ Breimer DD Borst P Multidrug resistance protein 1 protects the choroid plexus epithelium and contributes to the blood-cerebrospinal fluid barrier J Clin Invest 2000 105 279 285 10675353
Johnson DR Finch RA Lin ZP Zeiss CJ Sartorelli AC The pharmacological phenotype of combined multidrug-resistance mdr1a/1b- and mrp1-deficient mice Cancer Res 2001 61 1469 1476 11245453
Muramatsu T Johnson DR Finch RA Johnson LK Leffert JJ Lin ZP Pizzorno G Sartorelli AC Age-related differences in vincristine toxicity and biodistribution in wild-type and transporter-deficient mice Oncol Res 2004 14 331 343 15301424
Bandi N Kompella UB Budesonide reduces multidrug resistance-associated protein 1 expression in an airway epithelial cell line (Calu-1) Eur J Pharmacol 2002 437 9 17 11864633 10.1016/S0014-2999(02)01267-0
Van der Deen M Timmer-Bosscha H Timens W Postma DS de Vries EG Effect of cigarette smoke extract on MRP1 function in bronchial epithelial cells [abstract] Proc Amer Assoc Cancer Res 2004 45 A508
Nagayoshi K Nemoto T Yokoyama S Yamashita F Hashida M Effect of polycyclic aromatic hydrocarbons on generation and efflux of glutathione conjugates in primary cultured alveolar epithelial cells Drug Metab Pharmacokinet 2004 19 407 412 15681894 10.2133/dmpk.19.407
Torky AR Stehfest E Viehweger K Taege C Foth H Immuno-histochemical detection of MRPs in human lung cells in culture Toxicology 2005 207 437 450 15664271 10.1016/j.tox.2004.10.014
Buchler M Konig J Brom M Kartenbeck J Spring H Horie T Keppler D cDNA cloning of the hepatocyte canalicular isoform of the multidrug resistance protein, cMrp, reveals a novel conjugate export pump deficient in hyperbilirubinemic mutant rats J Biol Chem 1996 271 15091 15098 8662992 10.1074/jbc.271.25.15091
Cui Y Konig J Buchholz JK Spring H Leier I Keppler D Drug resistance and ATP-dependent conjugate transport mediated by the apical multidrug resistance protein, MRP2, permanently expressed in human and canine cells Mol Pharmacol 1999 55 929 937 10220572
Leier I Jedlitschky G Buchholz U Cole SP Deeley RG Keppler D The MRP gene encodes an ATP-dependent export pump for leukotriene C4 and structurally related conjugates J Biol Chem 1994 269 27807 27810 7961706
Kamisako T Leier I Cui Y Konig J Buchholz U Hummel-Eisenbeiss J Keppler D Transport of monoglucuronosyl and bisglucuronosyl bilirubin by recombinant human and rat multidrug resistance protein 2 Hepatology 1999 30 485 490 10421658 10.1002/hep.510300220
Holland IB Cole SP Kuchler K Higgins CF ABC proteins, from bacteria to men 2003 London: Elsevier Sci Ltd p427
Paulusma CC Bosma PJ Zaman GJ Bakker CT Otter M Scheffer GL Scheper RJ Borst P Oude Elferink RP Congenital jaundice in rats with a mutation in a multidrug resistance-associated protein gene Science 1996 271 1126 1128 8599091
Hosokawa S Tagaya O Mikami T Nozaki Y Kawaguchi A Yamatsu K Shamoto M A new rat mutant with chronic conjugated hyperbilirubinemia and renal glomerular lesions Lab Anim Sci 1992 42 27 34 1316504
Kuroda M Kobayashi Y Tanaka Y Itani T Mifuji R Araki J Kaito M Adachi Y Increased hepatic and renal expressions of multidrug resistance-associated protein 3 in Eisai hyperbilirubinuria rats J Gastroenterol Hepatol 2004 19 146 153 14731123 10.1111/j.1440-1746.2004.03275.x
Kiuchi Y Suzuki H Hirohashi T Tyson CA Sugiyama Y cDNA cloning and inducible expression of human multidrug resistance associated protein 3 (MRP3) FEBS Lett 1998 433 149 152 9738950 10.1016/S0014-5793(98)00899-0
Kool M van der Linden M de Haas M Scheffer GL de Vree JM Smith AJ Jansen G Peters GJ Ponne N Scheper RJ Elferink RP Baas F Borst P MRP3, an organic anion transporter able to transport anti-cancer drugs Proc Natl Acad Sci U S A 1999 96 6914 6919 10359813 10.1073/pnas.96.12.6914
Scheffer GL Kool M de Haas M de Vree JM Pijnenborg AC Bosman DK Elferink RP van der Valk P Borst P Scheper RJ Tissue distribution and induction of human multidrug resistant protein 3 Lab Invest 2002 82 193 201 11850532
Oguri T Isobe T Fujitaka K Ishikawa N Kohno N Association between expression of the MRP3 gene and exposure to platinum drugs in lung cancer Int J Cancer 2001 93 584 589 11477564 10.1002/ijc.1369
Young LC Campling BG Voskoglou-Nomikos T Cole SP Deeley RG Gerlach JH Expression of multidrug resistance protein-related genes in lung cancer: correlation with drug response Clin Cancer Res 1999 5 673 680 10100721
Borst P Evers R Kool M Wijnholds J A family of drug transporters: the multidrug resistance-associated proteins J Natl Cancer Inst 2000 92 1295 1302 10944550 10.1093/jnci/92.16.1295
Wielinga PR van der Heijden I Reid G Beijnen JH Wijnholds J Borst P Characterization of the MRP4- and MRP5-mediated transport of cyclic nucleotides from intact cells J Biol Chem 2003 278 17664 17671 12637526 10.1074/jbc.M212723200
Jedlitschky G Burchell B Keppler D The multidrug resistance protein 5 functions as an ATP-dependent export pump for cyclic nucleotides J Biol Chem 2000 275 30069 30074 10893247 10.1074/jbc.M005463200
Chen ZS Lee K Kruh GD Transport of cyclic nucleotides and estradiol 17-beta-D-glucuronide by multidrug resistance protein 4. Resistance to 6-mercaptopurine and 6-thioguanine J Biol Chem 2001 276 33747 33754 11447229 10.1074/jbc.M104833200
Van Aubel RA Smeets PH Peters JG Bindels RJ Russel FG The MRP4/ABCC4 gene encodes a novel apical organic anion transporter in human kidney proximal tubules: putative efflux pump for urinary cAMP and cGMP J Am Soc Nephrol 2002 13 595 603 11856762
Van den Berge M Polosa R Kerstjens HA Postma DS The role of endogenous and exogenous AMP in asthma and chronic obstructive pulmonary disease J Allergy Clin Immunol 2004 114 737 746 15480309 10.1016/j.jaci.2004.05.071
Reid G Wielinga P Zelcer N van der Heijden I Kuil A de Haas M Wijnholds J Borst P The human multidrug resistance protein MRP4 functions as a prostaglandin efflux transporter and is inhibited by nonsteroidal antiinflammatory drugs Proc Natl Acad Sci USA 2003
Zelcer N Reid G Wielinga P Kuil A van der Heijden I Schuetz JD Borst P Steroid and bile acid conjugates are substrates of human multidrug-resistance protein (MRP) 4 (ATP-binding cassette C4) Biochem J 2003 371 361 367 12523936 10.1042/BJ20021886
Prehm P Schumacher U Inhibition of hyaluronan export from human fibroblasts by inhibitors of multidrug resistance transporters Biochem Pharmacol 2004 68 1401 1410 15345330 10.1016/j.bcp.2004.06.017
Szakacs G Annereau JP Lababidi S Shankavaram U Arciello A Bussey KJ Reinhold W Guo Y Kruh GD Reimers M Weinstein JN Gottesman MM Predicting drug sensitivity and resistance; Profiling ABC transporter genes in cancer cells Cancer Cell 2004 6 129 137 15324696 10.1016/j.ccr.2004.06.026
Savaraj N Wu C Wangpaichitr M Kuo MT Lampidis T Robles C Furst AJ Feun L Overexpression of mutated MRP4 in cisplatin resistant small cell lung cancer cell line: collateral sensitivity to azidothymidine Int J Oncol 2003 23 173 179 12792791
Suzuki T Nishio K Tanabe S The MRP family and anticancer drug metabolism Curr Drug Metab 2001 2 367 377 11766988 10.2174/1389200013338289
Scheffer GL Hu X Pijnenborg AC Wijnholds J Bergen AA Scheper RJ MRP6 (ABCC6) detection in normal human tissues and tumors Lab Invest 2002 82 515 518 11950908
Hu X Plomp AS van Soest S Wijnholds J de Jong PT Bergen AA Pseudoxanthoma elasticum: a clinical, histopathological, and molecular update Surv Ophthalmol 2003 48 424 438 12850230 10.1016/S0039-6257(03)00053-5
Jackson A Loh CL Pulmonary calcification and elastic tissue damage in pseudoxanthoma elasticum Histopathology 1980 4 607 611 7439889
Beck K Hayashi K Nishiguchi B Le Saux O Hayashi M Boyd CD The distribution of Abcc6 in normal mouse tissues suggests multiple functions for this ABC transporter J Histochem Cytochem 2003 51 887 902 12810839
Hopper-Borge E Chen ZS Shchaveleva I Belinsky MG Kruh GD Analysis of the drug resistance profile of multidrug resistance protein 7 (ABCC10): resistance to docetaxel Cancer Res 2004 64 4927 4930 15256465
Chen ZS Guo Y Belinsky MG Kotova E Kruh GD Transport of bile acids, sulfated steroids, estradiol 17-beta-D-glucuronide, and leukotriene C4 by human multidrug resistance protein 8 (ABCC11) Mol Pharmacol 2005 67 545 557 15537867 10.1124/mol.104.007138
Denning GM Ostedgaard LS Welsh MJ Abnormal localization of cystic fibrosis transmembrane conductance regulator in primary cultures of cystic fibrosis airway epithelia J Cell Biol 1992 118 551 559 1379244 10.1083/jcb.118.3.551
Engelhardt JF Zepeda M Cohn JA Yankaskas JR Wilson JM Expression of the cystic fibrosis gene in adult human lung J Clin Invest 1994 93 737 749 7509347
Puchelle E Gaillard D Ploton D Hinnrasky J Fuchey C Boutterin MC Jacquot J Dreyer D Pavirani A Dalemans W Differential localization of the cystic fibrosis transmembrane conductance regulator in normal and cystic fibrosis airway epithelium Am J Respir Cell Mol Biol 1992 7 485 491 1384582
Kalin N Claass A Sommer M Puchelle E Tummler B DeltaF508 CFTR protein expression in tissues from patients with cystic fibrosis J Clin Invest 1999 103 1379 1389 10330420
Pilewski JM Frizzell RA Role of CFTR in airway disease Physiol Rev 1999 79 S215 S255 9922383
Brezillon S Dupuit F Hinnrasky J Marchand V Kalin N Tummler B Puchelle E Decreased expression of the CFTR protein in remodeled human nasal epithelium from non-cystic fibrosis patients Lab Invest 1995 72 191 200 7531792
Brezillon S Hamm H Heilmann M Schafers HJ Hinnrasky J Wagner TO Puchelle E Tummler B Decreased expression of the cystic fibrosis transmembrane conductance regulator protein in remodeled airway epithelium from lung transplanted patients Hum Pathol 1997 28 944 952 9269831 10.1016/S0046-8177(97)90010-1
Penque D Mendes F Beck S Farinha C Pacheco P Nogueira P Lavinha J Malho R Amaral MD Cystic fibrosis F508del patients have apically localized CFTR in a reduced number of airway cells Lab Invest 2000 80 857 868 10879737
Morris AP Cunningham SA Benos DJ Frizzell RA Cellular differentiation is required for cAMP but not Ca(2+)-dependent Cl- secretion in colonic epithelial cells expressing high levels of cystic fibrosis transmembrane conductance regulator J Biol Chem 1992 267 5575 5583 1372005
Hollande E Fanjul M Chemin-Thomas C Devaux C Demolombe S Van Rietschoten J Guy-Crotte O Figarella C Targeting of CFTR protein is linked to the polarization of human pancreatic duct cells in culture Eur J Cell Biol 1998 76 220 227 9716269
Hurbain I Sermet-Gaudelus I Vallee B Feuillet MN Lenoir G Bernaudin JF Edelman A Fajac A Evaluation of MRP1-5 gene expression in cystic fibrosis patients homozygous for the delta F508 mutation Pediatr Res 2003 54 627 634 12930913 10.1203/01.PDR.0000090926.16166.3F
Linsdell P Hanrahan JW Substrates of multidrug resistance-associated proteins block the cystic fibrosis transmembrane conductance regulator chloride channel Br J Pharmacol 1999 126 1471 1477 10217542 10.1038/sj.bjp.0702458
Bebok Z Varga K Hicks JK Venglarik CJ Kovacs T Chen L Hardiman KM Collawn JF Sorscher EJ Matalon S Reactive oxygen nitrogen species decrease cystic fibrosis transmembrane conductance regulator expression and cAMP-mediated Cl- secretion in airway epithelia J Biol Chem 2002 277 43041 43049 12194970 10.1074/jbc.M203154200
Hickman-Davis J Gibbs-Erwin J Lindsey JR Matalon S Surfactant protein A mediates mycoplasmacidal activity of alveolar macrophages by production of peroxynitrite Proc Natl Acad Sci U S A 1999 96 4953 4958 10220400 10.1073/pnas.96.9.4953
Hickman-Davis JM O'Reilly P Davis IC Peti-Peterdi J Davis G Young KR Devlin RB Matalon S Killing of Klebsiella pneumoniae by human alveolar macrophages Am J Physiol Lung Cell Mol Physiol 2002 282 L944 L956 11943658
Eiserich JP Hristova M Cross CE Jones AD Freeman BA Halliwell B van der Vliet A Formation of nitric oxide-derived inflammatory oxidants by myeloperoxidase in neutrophils Nature 1998 391 393 397 9450756 10.1038/34923
Punjabi CJ Laskin JD Pendino KJ Goller NL Durham SK Laskin DL Production of nitric oxide by rat type II pneumocytes: increased expression of inducible nitric oxide synthase following inhalation of a pulmonary irritant Am J Respir Cell Mol Biol 1994 11 165 172 7519435
Weinberger B Heck DE Laskin DL Laskin JD Nitric oxide in the lung: therapeutic and cellular mechanisms of action Pharmacol Ther 1999 84 401 411 10665837 10.1016/S0163-7258(99)00044-3
Meng QH Polak JM Edgar AJ Chacon MR Evans TJ Gruenert DC Bishop AE Neutrophils enhance expression of inducible nitric oxide synthase in human normal but not cystic fibrosis bronchial epithelial cells J Pathol 2000 190 126 132 10657009 10.1002/(SICI)1096-9896(200002)190:2<126::AID-PATH500>3.0.CO;2-#
Trezise AE Ratcliff R Hawkins TE Evans MJ Freeman TC Romano PR Higgins CF Colledge WH Co-ordinate regulation of the cystic fibrosis and multidrug resistance genes in cystic fibrosis knockout mice Hum Mol Genet 1997 6 527 537 9097955 10.1093/hmg/6.4.527
Ratcliff R Evans MJ Cuthbert AW MacVinish LJ Foster D Anderson JR Colledge WH Production of a severe cystic fibrosis mutation in mice by gene targeting Nat Genet 1993 4 35 41 7685652 10.1038/ng0593-35
Shen BQ Finkbeiner WE Wine JJ Mrsny RJ Widdicombe JH Calu-3: a human airway epithelial cell line that shows cAMP-dependent Cl- secretion Am J Physiol 1994 266 L493 L501 7515578
Baudouin-Legros M Brouillard F Tondelier D Hinzpeter A Edelman A Effect of ouabain on CFTR gene expression in human Calu-3 cells Am J Physiol Cell Physiol 2003 284 C620 C626 12556359
Cozens AL Yezzi MJ Kunzelmann K Ohrui T Chin L Eng K Finkbeiner WE Widdicombe JH Gruenert DC CFTR expression and chloride secretion in polarized immortal human bronchial epithelial cells Am J Respir Cell Mol Biol 1994 10 38 47 7507342
Kogan I Ramjeesingh M Li C Kidd JF Wang Y Leslie EM Cole SP Bear CE CFTR directly mediates nucleotide-regulated glutathione flux EMBO J 2003 22 1981 1989 12727866 10.1093/emboj/cdg194
Croop JM Tiller GE Fletcher JA Lux ML Raab E Goldenson D Son D Arciniegas S Wu RL Isolation and characterization of a mammalian homolog of the Drosophila white gene Gene 1997 185 77 85 9034316 10.1016/S0378-1119(96)00633-6
Doyle LA Yang W Abruzzo LV Krogmann T Gao Y Rishi AK Ross DD A multidrug resistance transporter from human MCF-7 breast cancer cells Proc Natl Acad Sci U S A 1998 95 15665 15670 9861027 10.1073/pnas.95.26.15665
Komatani H Kotani H Hara Y Nakagawa R Matsumoto M Arakawa H Nishimura S Identification of breast cancer resistant protein/mitoxantrone resistance/placenta-specific, ATP-binding cassette transporter as a transporter of NB-506 and J-10 topoisomerase I inhibitors with an indolocarbazole structure Cancer Res 7088 61 2827 2832 11306452
Diestra JE Scheffer GL Catala II Maliepaard M Schellens JH Scheper RJ Germa-Lluch JR Izquierdo MA Frequent expression of the multi-drug resistance-associated protein BCRP/MXR/ABCP/ABCG2 in human tumours detected by the BXP-21 monoclonal antibody in paraffin-embedded material J Pathol 2002 198 213 219 12237881 10.1002/path.1203
Yoh K Ishii G Yokose T Minegishi Y Tsuta K Goto K Nishiwaki Y Kodama T Suga M Ochiai A Breast cancer resistance protein impacts clinical outcome in platinum-based chemotherapy for advanced non-small cell lung cancer Clin Cancer Res 2004 10 1691 1697 15014021
Jonker JW Buitelaar M Wagenaar E van der Valk MA Scheffer GL Scheper RJ Plosch T Kuipers F Elferink RP Rosing H Beijnen JH Schinkel AH The breast cancer resistance protein protects against a major chlorophyll-derived dietary phototoxin and protoporphyria Proc Natl Acad Sci U S A 2002 99 15649 15654 12429862 10.1073/pnas.202607599
Summer R Kotton DN Sun X Ma B Fitzsimmons K Fine A Side population cells and Bcrp1 expression in lung Am J Physiol Lung Cell Mol Physiol 2003 285 L97 104 12626330
Giangreco A Shen H Reynolds SD Stripp BR Molecular Phenotype of Airway Side Population Cells Am J Physiol Lung Cell Mol Physiol 2003
Neuringer IP Randell SH Stem cells and repair of lung injuries Respir Res 2004 5 6 15285789 10.1186/1465-9921-5-6
Agassandian M Mathur SN Zhou J Field FJ Mallampalli RK Oxysterols trigger ABCA1-mediated basolateral surfactant efflux Am J Respir Cell Mol Biol 2004 31 227 233 15039140 10.1165/rcmb.2004-0038OC
Drobnik W Lindenthal B Lieser B Ritter M Christiansen WT Liebisch G Giesa U Igel M Borsukova H Buchler C Fung-Leung WP Von Bergmann K Schmitz G ATP-binding cassette transporter A1 (ABCA1) affects total body sterol metabolism Gastroenterology 2001 120 1203 1211 11266384 10.1053/gast.2001.23250
McNeish J Aiello RJ Guyot D Turi T Gabel C Aldinger C Hoppe KL Roach ML Royer LJ de Wet J Broccardo C Chimini G Francone OL High density lipoprotein deficiency and foam cell accumulation in mice with targeted disruption of ATP-binding cassette transporter-1 Proc Natl Acad Sci U S A 2000 97 4245 4250 10760292 10.1073/pnas.97.8.4245
Wang N Lan D Gerbod-Giannone M Linsel-Nitschke P Jehle AW Chen W Martinez LO Tall AR ATP-binding cassette transporter A7 (ABCA7) binds apolipoprotein A-I and mediates cellular phospholipid but not cholesterol efflux J Biol Chem 2003 278 42906 42912 12917409 10.1074/jbc.M307831200
Yamano G Funahashi H Kawanami O Zhao LX Ban N Uchida Y Morohoshi T Ogawa J Shioda S Inagaki N ABCA3 is a lamellar body membrane protein in human lung alveolar type II cells FEBS Lett 2001 508 221 225 11718719 10.1016/S0014-5793(01)03056-3
|
0
|
PMC1200431
|
CC BY
|
2021-01-04 22:25:42
|
no
|
Biomed Eng Online. 2005 Jul 18; 4:43
|
latin-1
|
Biomed Eng Online
| 2,005 |
10.1186/1475-925X-4-43
|
oa_comm
|
==== Front
Aust New Zealand Health PolicyAustralia and New Zealand Health Policy1743-8462BioMed Central London 1743-8462-2-141598752010.1186/1743-8462-2-14CommentaryInterventions to facilitate health workforce restructure Duckett SJ [email protected] School of Public Health, La Trobe University, Melbourne Vic 3086, Australia2005 29 6 2005 2 14 14 10 5 2005 29 6 2005 Copyright © 2005 Duckett; licensee BioMed Central Ltd.2005Duckett; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
There are recognised shortages in most health professions in Australia. This is evidence that previous attempts at health workforce planning have failed. This paper argues that one reason for such failure is the lack of appropriate structures for health workforce planning. It also suggests that Australia needs to move beyond planning for particular professions and that health workforce planning needs to be based on identifying skill shortages as much as shortages in particular named professionals.
The paper proposes specific policy suggestions to facilitate workforce flexibility and health workforce planning in Australia.
==== Body
Background
Health workforce reform is clearly on the agenda of health policy makers in Australia. It has been the focus of discussion at the Council of Australian Governments which requested the Commonwealth to initiate further research in this area, operationalised by the Treasurer commissioning the current research study by the Productivity Commission . There are a number of immediate causative factors for this heightened policy attention, most notably contemporary perceived shortages of most categories of health professionals. Increasingly, health policy makers and health service managers are also recognising that the current structure of the health workforce is probably not suitable for 21st century healthcare delivery. [1]
Australia is not unique in facing workforce shortages [2,3], nor in recognising the inadequacy of current workforce structures [4]; the World Health Organisation is highlighting workforce issues internationally by making them the focus of its 2006 World Health Report . Although the headline problem is usually couched in terms of workforce supply, problems in flexibility of the workforce and workforce planning also confront policymakers.
The focus on workforce flexibility is in part a response to perceived overspecialisation of the health workforce. Specialisation, which in part was seen to be associated with higher quality, is now seen as possibly detracting from continuity of care and hence may have a deleterious impact on quality, especially in the context of the increased salience of chronic diseases in the health sector. Although all the benefits of specialisation should not be lost, the current assignment of roles for health professionals is perceived to be inefficient either because more staff are employed than would be required in an efficient organisation of roles, or staff at higher pay classifications being used to perform tasks which could be performed by staff at lower pay levels. The inflexibility of contemporary workforce structure also inhibits service delivery because of shortages of staff to perform key roles. Policy attention is therefore being directed towards strategies about workforce substitution [5] and to develop skills "escalators", that is to make it easy for existing health professionals to acquire additional skills to enable them to perform additional tasks.
Table 1 shows some of the task substitutions which could potentially take place in Australia. In some cases the substitution is already occurring and the potential is for expansion of this practice. In other cases, substitution will require:
Table 1 Examples of potential (or current) task substitutions
Task* Traditional professional Substitute professional/assistant
Anaesthesia Anaesthetist Nurse anaesthetist
Clerking of new hospital patients Hospital medical officer Nurse
Closure of wound Surgeon Nurse
Foot care Podiatrist Foot care assistant
Foot surgery Orthopaedic surgeon Podiatric surgeon
Laryngoscopy/Naso-endoscopy ENT surgeon Speech pathologist/Nurse
Maternity care Obstetrician Midwife or GP
Mobilisation assistance Physiotherapist Physiotherapy assistant
Patient management Medical practitioner Nurse practitioner
Plain X-ray Medical imaging technologist X-ray assistant
Refraction Optometrist Orthoptist
Reporting pathology Pathologist Scientist
Reporting X-rays Radiologist Medical imaging technologist
* Performance of the substituted tasks will generally require additional training and clear protocols, and will also depend on the complexity of the condition and the comorbities of the patient
• Identification and clarification of the precise range of tasks to be substituted;
• Protocols to identify the types of patients for whom the substitute professional or assistant is relevant;
• Clarification of the nature of supervision, and reporting and regulatory arrangements (if any);
• Negotiation of payment/salary arrangements.
Obviously new substitution arrangements need to be carefully planned and monitored, but over time as health agencies (and patients) become more confident and familiar with substitution, expanded roles and task substitution will become a recognised and routine part of service delivery.
The possible substitution examples outlined above mostly involve changing the scope of practice of existing professionals. Substitution can also occur through creation of new categories of professionals or assistants. [6] The more prevalent substitution becomes, the more there will be challenges to our contemporary conception of the definition and place of a "nurse" or "physiotherapist". This will not be an issue for members of health care teams who work closely or regularly with team members working in extended roles, but transient team members (such as agency staff or staff with only irregular or peripheral contact with the team) may not be fully aware of the team's skill mix and may make inappropriate referrals or consultative decisions. Consumers may also have different expectations of the treating team membership, and this too will need to be addressed.
The importance of addressing workforce flexibility and the associated issue of workforce substitution cannot be underestimated, particularly as predictions of future workforce requirements need to make some assumptions about the mix of tasks that will be performed in the future by the health professionals under review [7,8]. If the tasks undertaken by physiotherapists, for example, are expanded, then more physiotherapists will be required, but if tasks currently undertaken by physiotherapists are able to be delegated to other categories of the health workforce, then the number of physiotherapists required in the future will be reduced. For this reason, the term 'skills shortage' is preferred to 'workforce shortage' to describe the contemporary problem. The latter term focuses on particular professions, thus channelling policy attention into traditional professional structures, rather than recognising workforce flexibility and the potential for changed skill mix.
A second cluster of problems relates to health workforce planning. The legal aphorism, res ipsa loquitur, is relevant here. The existence of skills shortages damns current workforce planning efforts. Although there are technical problems with workforce and demand projections, a critical inhibiting factor is the lack of effective formal structural links between the health and education sectors. Figure 1 shows the current relationships.
Figure 1 Organisational relationships between health and education sectors.
A health agency, for example, relates most closely in organisational terms to the State Health Department. The State Health Department has an overview of the needs of the health agencies within a State and State Health Ministers may be politically exposed to shortages in particular health professions which lead to problems of service delivery.
State Health Departments have two sets of relationships which are of relevance here. One is to the Commonwealth Health Department structured through organisational arrangements such as the Australian Health Ministers' Conference and the Australian Health Ministers' Advisory Council. The other is to the parallel State Education Department. Relationships between state health and education departments are not always close and rarely involve structured joint planning arrangements. These somewhat looser relationships are indicated by dotted lines in the figure. The Commonwealth Health Department has links to the Commonwealth Education Department, which in turn has links to State Education Departments and to universities. At the bottom of the figure we note that health agencies have direct relationships with Faculties of Health Sciences within universities, for example, in terms of placement arrangements.
The mechanisms for a health agency or a State Health Department to influence the admission or curriculum decisions of universities are very indirect, typically progressing up and down the chain, mediated by the Commonwealth Departments. The longer the links in an implementation chain, the more the policies are likely to be attenuated or distorted. [9] The mechanisms for implementing health workforce decisions are very indirect and this could be predicted to be relatively ineffectual, which they are.
Proposals for reform
The incentives on health services and universities are very different. The nature of accountability of the two sectors does not automatically guarantee that the two sectors would see the problems of the health workforce in a similar way, nor that they would accord the same priority to different solutions.
These differences, coupled with the indirect mechanisms of influence highlighted in the figure, mean that the perceived problems of the health workforce may not be easily resolved. Faculties of Health Sciences within universities are not autonomous, and even if they identify with the problems faced by the health sector, they may not have sufficient internal power within universities to effect change. University management may not recognise or accept a need to change university priorities or respond to the perceived problems of the health sector.
There are similar problems of alignment in terms of facilitating workforce flexibility. The structure of Australia's Medicare benefits arrangements militate against proposals for medical practitioners to promote substitution strategies.
These problems suggest that new strategies are necessary to change incentives to facilitate workforce reform; such strategies can be initiated by individual players in the health workforce policy area which would facilitate improvements in flexibility and in workforce planning.
Workforce flexibility
In terms of workforce flexibility, one of the critical barriers to reform is the lack of financial incentives on the medical profession to engage in significant restructure of work roles. The Australian Medicare scheme places financial incentives on medical practitioners to provide services themselves because, in general, only services provided by the medical practitioner attract a rebate under the Medical Benefits Schedule. However, this is not universally true, and there are a number of items on the Schedule which do not require "personal provision" by the medical practitioner (see Section 12.1.1 of the Medicare Schedule, ). A typical item that does not require personal provision is a pathology test, where the test if performed by a medical scientist with the medical practitioner not necessarily being present, or even seeing the test result before it is issued.
One strategy to encourage flexibility in the workforce would be to increase the range of items which do not require personal provision by, for example, designating all procedural items in this category. In this way, for example, an anaesthetist would be able to bill for the work of a nurse anaesthetist using the anaesthetic items of the Schedule. Assuming salary costs for the substitute professional are lower than the medical specialist, this would then put a financial incentive on medical practitioners to utilise other health professionals for service delivery. It may also be appropriate to allow some consultation items to be billed without personal provision, eg. if provided by a nurse practitioner/advanced practice nurse. This change could be undertaken by the Commonwealth Department of Health unilaterally. A change to the 'personal provision' rules of the Schedule can be undertaken without affecting contemporary fee relativities. However, procedural items are often claimed to be over valued relative to non-procedural items, and a change in the personal provision rules could also involve some realignment of the Schedule to facilitate a greater recognition of the cognitively complex components of patient care which will remain the preserve of medically qualified practitioners.
Medical practitioners are rightly concerned at their place in the health care system. The political voices of the medical profession thus generally oppose structural changes which reduce medical autonomy or might increase competition and impact on medical incomes. A change to the Medicare Benefits Schedule of the kind proposed would ameliorate these concerns and might thus enable the medical professional to support such changes without any threat to their roles and income. It should be noted, however, that delegation from medical practitioners is not and should not be the only possible source of income to support new or expanded roles.
A parallel change could be to introduce powers of delegation within health professional registration Acts. A power of delegation would facilitate professionals delegating tasks by extending the reach of a health professional registration board to cover the work of any person to whom a professional registered with that board has delegated tasks (see, for example, . [10] Such a power of delegation would establish a regulatory framework for health professionals' delegating to other professionals or assistants, and would allow professionals to delegate tasks, knowing they were doing so within an accepted regulatory framework.
Thirdly, the Medical Benefits and Pharmaceutical Benefits arrangements (and associated state regulatory controls) could be changed to give other professionals access to test ordering or prescribing authority. This is already occurring through expanding nurse practitioner access to MBS and PBS arrangements, but it could also apply to podiatrists, optometrists, physiotherapists and other health professionals.
These flexibility arrangements should be accompanied by parallel educational reforms to facilitate upskilling and reskilling of health professionals. In this regard, increased graduate entry programs for health professionals, whereby graduates from other disciplines are able to undertake shortened courses to gain professional recognition, should be encouraged. Shortened courses for professionals to acquire some of the key skills beyond their normal range should also be developed (eg. nurses to be trained in foot care). Similarly, universities should be encouraged to provide modular and multiple pathways for developing health professionals. For example, there should be expanded arrangements for dental therapists to upgrade to become dentists. Similarly, there should be multiple pathways for training of psychiatric nurses, providing structured courses for people with an initial nursing background and for people with an initial psychology background.
Not all skill upgrades will need to involve university programs. Health agencies (and professional registration bodies) may cooperate to develop work-based programs to address skill gaps. Such programs should be conducted with a recognised qualification framework to ensure portability. The Vocational Education and Training sector may have a role here, possibly in collaboration with universities.
Changes to structures
The changes to facilitate flexibility outlined above can be undertaken unilaterally by States changing registration Board legislation or by the Commonwealth changing the MBS arrangements. However, in the long term there needs to be reform to funding and management structures to improve health workforce planning in Australia. The Nelson changes to higher education have improved the accountability of universities though closer monitoring of the course mix within universities. But this monitoring is still undertaken at a very broad level and cannot be expected to go into the detail of particular health professions. Further, previous experience suggests that, although the Department of Education will have an initial flush of enthusiasm for close monitoring, this enthusiasm abates and the relationships between the Commonwealth Department of Education and universities become more laissez faire over time. [10]
The situation in Australia for funding universities for health professional education contrasts with that in the United Kingdom where, for most health professions, there is a direct contractual relationship between the health sector and universities. For example, although degree-level education for nursing is funded through standard education sector funding for universities, diploma-level training, which represents the overwhelming majority of university-based nursing education in the United Kingdom, is funded in universities through a direct contract between the university and the National Health Service. This helps to ensure that universities are more accountable to the health sector to provide relevant and appropriate health professional education graduating adequate numbers of professionals. Such contracts are comprehensive, providing for an ongoing relationship between the health service and the university, and are not simply based on selection of the cheapest provider. It also provides a framework for experimentation and responsiveness in terms of preparing new types of health workers. The potential of these arrangements has not been fully realised, although there are pockets of innovation (see ). In the early years of NHS purchasing of professional education, the NHS tended to underestimate demand for health professionals from non-NHS agencies [7], highlighting the need for comprehensive skills planning.
There are lessons for Australia here, and more direct links between the health sector and universities could improve responsiveness of universities to emerging needs. The first stage of such a closer relationship could occur if the Commonwealth assigned responsibility for health professional education to the Department of Health and Ageing rather than the Department of Education, Science and Technology. Universities already face multiple sources of funding, and a shift of responsibility for health professional education to the Department of Health and Ageing would give that Department a direct involvement in setting priorities for the future health workforce and funding universities accordingly. In contrast to the Department of Education, Science and Training, the Department of Health and Ageing is much more likely to have an ongoing and continuing interest in ensuring adequate numbers of health professionals and the competencies attained by new graduates. Similarly, many reports commissioned by the Department of Health lament the adequacy of the curriculum of universities in a range of areas, but there have been few levers over universities to effect relevant changes. A shift of responsibility would reduce the number of links in the chain between health agencies and universities in terms of responsiveness and skills planning. Such a change in funding source could be undertaken unilaterally by the Commonwealth Government.
State health authorities could also exercise influence over universities. In many States, the State health authority provides subsidies to universities either directly or indirectly for professorial appointments. Many clinical Chairs in medicine, nursing and other disciplines, are funded by the State health authority (or health agencies). Similarly, health professional education would not be viable were it not for the access to State-funded health agencies for clinical education. Controlling this access could thus give State health authorities some levers over universities. Given these levers, State health authorities could take a much more direct role in negotiating with universities about health professional education than they have hitherto. This change could be undertaken unilaterally by any State, regardless of any changes in Commonwealth responsibilities.
Finally, it would be better if the actions of the Commonwealth and the States were brought together into a coherent policy approach. This could be done informally through arrangements for joint Commonwealth/State negotiations with universities. A stronger policy could be to establish a single funding pool to which both the Commonwealth and the States contribute, which would facilitate direct negotiations between the Commonwealth and State Governments on one hand, and universities on the other. A single funding pool could be administered by a jointly established health workforce funding agency in each State which would have the full purchasing responsibility for health professional education. These arrangements would, of course, be more complex to implement but would be much more powerful mechanisms for reform. Coordination of state and Commonwealth activity would also facilitate engagement with the private sector. Given the different roles of the public and private sectors in health delivery, it is important to develop structures to engage the private sector more directly in educating the future health workforce which will be employed in that sector. Involvement of the private sector in the skills planning process would also help to ensure that demand from that sector is taken into account in supply decisions, overcoming one of the early weaknesses experienced in the similar arrangements in the United Kingdom.
Universities might benefit from more systematic planning arrangements for health professional education. At present universities often have problems negotiating clinical education arrangements with health agencies, where there are few incentives on agencies to assume responsibility for education of the next generation of health professionals. A quid pro quo for increased university accountability to State or Commonwealth health departments would be an increased responsibility on either the Commonwealth or the State governments (or both) to ensure the adequacy of clinical education arrangements for universities. A direct relationship would also make it more likely that skills upgrading would remain within the purview of universities rather than bypassing them. Finally, distinct education funding arrangements for the health sector would mean that government could allocate more funding per health student without creating a precedent for increases in other disciplines.
Conclusion
There are a number of contemporary problems of health professional education. Many of these problems have been identified for decades, but there have been few incentives to achieve change and/or the structural mechanisms for change have hitherto not been present. Past opportunities have been missed, eg. in the negotiation of the 2003–2008 Australian Health Care Agreement. The current heightened policy awareness of the need for workforce reform provides a new opportunity for change. Discussion, planning and experimentation should commence now to provide a sounder conceptual and evidence base to ensure that opportunities are not missed to incorporate reform proposals in the 2008–2013 Australian Health Care Agreement.
In this paper I have outlined incremental steps that could be used to facilitate change in health workforce policy in Australia. The general tenor of the changes are that they provide for increased accountability of universities. But the costs of these changes do not fall only on universities. Introducing new mechanisms to hold universities accountable for adequacy of health professional education in turn means that governments themselves are more clearly accountable for the adequacy of health professional education, and shortages in any discipline would be more clearly seen to be as a result of government decisions. Governments (both Commonwealth and State) benefit from the ability to blame shift to other participants in the health workforce policy arena.
New structures for the health workforce and for health workforce planning are clearly necessary in Australia. In this paper I have outlined a possible win/win scenario for policy reform to address these needs.
==== Refs
Duckett SJ 'Health workforce design in the 21st century' Aust Health Rev 2005 29 201 210 15865571
Simoens S Hurst J Matching supply with demand for the services of physicians and nurses 2004 Towards high-performing health systems, The OECD Health Project: Paris
Simoens S Villeneuve M Hurst J Tackling nurse shortages in OECD countries 2005 OECD Health Working Papers (19): Paris
Buchan J Calman L Skill-mix and policy change in the health workforce: Nurses in advanced roles 2004 OECD Health Working Papers (17): Paris
Laurant M Reeves D Hemens R Braspenning J Grol R Sibbald B 'Substitution of doctors by nurses in primary care (Review)' The Cochrane Library 2005
Sibbald B Shen J McBride A 'Changing the skill-mix of the health care workforce' J Health Serv Res Policy 2004 9 28 38 15006226 10.1258/135581904322724112
Buchan J Edwards N 'Nursing numbers in Britain: the argument for workforce planning' BMJ 2000 320 1067 1070 10764372 10.1136/bmj.320.7241.1067
Reinhardt UE Physician productivity and the demand for health manpower 1975 Ballinger Publishing Company: Cambridge, Mass
Pressman J Wildavsky A Implementation: How great expectations in Washington are dashed in Oakland 1973 California University Press
de Bie J Cuperus-Bosma JM Gevers JKM van der Wal G 'Reserved procedures in Dutch hospitals: knowledge, experiences and views of physicians and nurses' Health Policy 2004 68 373 384 15113648 10.1016/j.healthpol.2003.11.001
Marginson S 'Steering from a distance: Power relations in Australian higher education' Higher Education 1997 34 63 80 10.1023/A:1003082922199
|
15987520
|
PMC1200557
|
CC BY
|
2021-01-04 16:38:28
|
no
|
Aust New Zealand Health Policy. 2005 Jun 29; 2:14
|
utf-8
|
Aust New Zealand Health Policy
| 2,005 |
10.1186/1743-8462-2-14
|
oa_comm
|
==== Front
BMC ImmunolBMC Immunology1471-2172BioMed Central London 1471-2172-6-191604278410.1186/1471-2172-6-19Research ArticleA role for the Tec family kinase ITK in regulating SEB induced Interleukin-2 production in vivo via c-jun phosphorylation Ragin Melanie J [email protected] Jianfang [email protected] Andrew J [email protected] Avery [email protected] Pathobiology Graduate Program, Center for Molecular Immunology & Infectious Disease and Department of Veterinary Science, The Pennsylvania State University, University Park, PA 16802, USA2 Immunobiology Option of the Integrated Bioscience Graduate Program, Center for Molecular Immunology & Infectious Disease and Department of Veterinary Science, The Pennsylvania State University, University Park, PA 16802, USA3 Center for Molecular Immunology & Infectious Disease and Department of Veterinary Science The Pennsylvania State University, University Park, PA 16802, USA2005 22 7 2005 6 19 19 20 12 2004 22 7 2005 Copyright © 2005 Ragin et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Exposure to Staphylococcal Enterotoxin B (SEB), a bacterial superantigen secreted by the Gram-positive bacteria Staphyloccocus aureus, results in the expansion and eventual clonal deletion and anergy of Vβ8+ T cells, as well as massive cytokine release, including Interleukin-2 (IL-2). This IL-2 is rapidly secreted following exposure to SEB and may contribute to the symptoms seen following exposure to this bacterial toxin. The Tec family kinase ITK has been shown to be important for the production of IL-2 by T cells stimulated in vitro and may represent a good target for blocking the production of this cytokine in vivo. In order to determine if ITK represents such a target, mice lacking ITK were analyzed for their response to SEB exposure.
Results
It was found that T cells from mice lacking ITK exhibited significantly reduced proliferative responses to SEB exposure in vitro, as well as in vivo. Examination of IL-2 production revealed that ITK null mice produced reduced levels of this cytokine in vitro, and more dramatically, in vivo. In vivo analysis of c-jun phosphorylation, previously shown to be critical for regulating IL-2 production, revealed that this pathway was specifically activated in SEB reactive Vβ8+ (but not non-reactive Vβ6+) T cells from WT mice, but not in Vβ8+ T cells from ITK null mice. However, toxicity analysis indicated that both WT and ITK null animals were similarly affected by SEB exposure.
Conclusion
These data show that ITK is required for IL-2 production induced by SEB in vivo, and may regulate signals leading IL-2 production, in part by regulating phosphorylation of c-jun. The data also suggest that perturbing T cell activation pathways leading to IL-2 does not necessarily lead to improved responses to SEB toxicity.
==== Body
Background
Superantigens (SAGs) are microbial toxins of bacterial and viral origin with the ability to activate 5–20% of the T cell population, causing T cell activation, cytokine release and systemic shock [1,2]. Most SAGs share the ability to simultaneously bind the class II major histocompatibility complex molecules and the variable region of the T cell receptor β-chain, without the need to be processed by antigen presenting cells [1,2]. Thus SEB can interact directly with MHC class II molecules on APCs and activate T cells bearing the proper TcR Vβ chains. The result of this interaction is large-scale stimulation of any T cell that expresses the proper TCR Vβ chain. A number of studies have shown that when mice are challenged with a SAG such as Staphylococcal Enterotoxin B (SEB), toxicity results from massive induction of cytokines derived from T-helper-type-1 (TH1) type cells such as IL-2, IFN-γ, TNF-α and TNF-β [1,3]. This cytokine production is accompanied by expansion of the numbers of SEB reactive T cells, followed by cell death and the induction of functional "anergy" [4-6].
In order for a T cell to be activated antigen presenting cells (APC) must present antigen, either an antigenic peptide or SAG. In vitro, this interaction with the TcR has been shown to lead to the activation of a number of tyrosine kinases, including the Src family kinase Lck, the Syk family kinase Zap-70 and the Tec family kinase ITK (for review see [7-10]). This then results in the activation of a number of signaling pathways including members of the MAPK family of kinases, ERK, JNK and p38, followed by transcription factor activation [10]. In vivo, activation of these T cells lead to the induction of cytokine secretion within hours of SEB exposure [1,3].
ITK is expressed primarily in T cells, NK cells and mast cells [11-13]. In T cells, it is rapidly activated following TcR crosslinking in vitro [14,15]. Mice lacking ITK exhibit reduced proliferation and IL-2 production in vitro, and reduced T cell differentiation in vitro and in vivo, with TH2 cell differentiation preferentially affected [15-19]. The observed reduced proliferation of ITK null T cells in vitro is IL-2 dependent as it could be rescued by the addition of exogenous IL-2 [20]. However, these animals are not entirely immunocompromised, with residual responses against LCM, Vaccinia and VS viruses [21]. We have tested whether ITK null mice are susceptible to SEB induced IL-2 secretion. We show here that mice lacking ITK have much reduced IL-2 production and T cell expansion in response to SEB in vitro and in vivo. We also show that SEB induced the activation of the JNK MAPK pathway in responding T cells in vivo, and that ITK null T cells were defective in the activation of this pathway in vivo. However, toxicity analysis indicated that both WT and ITK null animals were similarly affected by SEB exposure. Our data suggest that ITK is required for full IL-2 secretion following SEB exposure, and that this may be due to the regulation of the JNK pathway by ITK in vivo. However, reducing T cell signals does not necessarily lead to better physiological responses to SEB exposure.
Results
ITK deficient T cells proliferate less efficiently than WT T cells in response to varying concentrations of SEB in vitro
It has previously been reported that T cells from mice lacking ITK exhibit much reduced proliferation in response to anti-CD3 or TcR antibodies in vitro [15,16,18]. By contrast, these T cells did not have any defects and actually had higher proliferative responses following anti-CD28/PMA stimulation [20]. These data do not provide clear answers as to whether ITK would present a good target for inhibition of SEB induced T cell activation in vivo, and in particular, in the cytokine response observed following SAG exposure. We therefore first tested whether ITK null T cells could respond to SEB stimulation by proliferation in vitro. Pooled splenocytes and lymph node cells from WT or ITK null animals were cultured in vitro in the presence of varying concentrations of SEB (5 μg/ml, 0.5 μg/ml, 0.05 μg/ml, 0.005 μg/ml, and 0.0005 μg/ml), and proliferation analyzed over a 5-day period. We found that WT cells responded normally by proliferating between days 2 and 3, with maximum proliferation observed at day 5 and the 5 μg/ml dose of SEB (Fig. 1a). Analysis of the dose response at 5 days indicated that the WT cells responded in a dose dependent fashion (Fig. 1b). By contrast, cells from ITK mice showed very little response, regardless of the dose of SEB used for stimulation (Fig. 1a, b). We next determined if the same results would be obtained in purified T cell populations from these mice. T cells were purified from WT and ITK null mice, and stimulated with varying dose of SEB, in the presence of APCs. Analysis of the data over the 5 days indicated that in the presence of the highest concentration of SEB tested (10 μg/ml), WT T cells responded more robustly than ITK null T cells (see Fig. 1c). Indeed, at all the doses tested, our analysis indicated that ITK null T cells responded less than WT T cells on day 3, 4 and 5 (Fig. 1d–f), while proliferation was equivalent when PMA and Ionomycin was used to bypass TcR signals (data not shown). These data indicate that ITK null T cells have significantly reduced proliferative responses to SEB in vitro.
ITK null T cells secrete less IL-2 in response to SEB in vitro
Interleukin-2 is one of the cytokines induced by exposure to SEB [22]. ITK has been shown to regulate IL-2 secretion in vitro by T cells stimulated anti-CD3/TcR antibodies or in response to peptide antigen [15,18,23,24]. We therefore determined the levels of IL-2 in supernatants from cells similarly stimulated with SEB in vitro (Fig. 2). While WT T cells responded to SEB by secretion of IL-2 within 24 hrs. of stimulation in a dose dependent manner (see Fig. 2a for 5 μg/ml dose), starting at the 0.5 μg/ml SEB concentration (Fig. 2bi), ITK null T cells made very little IL-2, and the small amount of IL-2 that was made was only at the highest concentration of SEB tested (5 μg/ml). ITK null T cells consistently secreted less IL-2 over the 5-day period and dose range (Fig. 2b). These data indicate that as previously reported for TcR or antigenic stimulation, ITK regulates the production of IL-2 in vitro in response to SEB stimulation [15,18,23,24].
Reduced expansion of Vβ8+CD4+ population in ITK null animals in response to SEB exposure in vivo
While these data indicate that ITK regulates T cell proliferation in response to SEB in vitro, it is possible that within the context of antigen presentation and optimal co-stimulatory signals in vivo, ITK null T cells may exhibit a better response. Exposure of animals to SEB results in an expansion of the SEB reactive T cell population with a peak of around 2 days post exposure [2,4]. In order to determine if the reduced proliferative responses seen with the ITK null T cells in vitro was also seen in vivo, we exposed WT and ITK null animals to SEB and determined the percentages of Vβ8 and Vβ6 positive cells in the CD4+ population by flow cytometry after 2, 4 and 5 days as a measure of T cell expansion. T cells bearing Vβ8 TcR chains are responsive to SEB stimulation while those that bear Vβ6 TcR chains are not, so we analyzed these two populations of T cells. As previously reported, in WT animals, we observed an approximately 4 fold expansion in Vβ8 CD4+ T cells in spleen and lymph nodes following SEB exposure (as compared to exposure to PBS, fig 3a,b), while Vβ6 CD4+ T cell populations were largely unchanged (data not shown). By contrast, there was statistically significantly less expansion of ITK null Vβ8+CD4+ T cells following SEB exposure, although these cells still expanded (Fig. 3a,b) As previously reported, the percentages of T cells bearing Vβ6 did not change over this period in either the WT or ITK null mice (data not shown and [4]). Thus the absence of ITK results in reduced T cell expansion in vivo in response to SEB.
Defective SEB induced IL-2 secretion in vivo in ITK null mice
T cells from ITK null mice proliferate less in vitro and in vivo, and secrete significantly less IL-2 in vitro in response to SEB. We therefore determined if they would also secrete less IL-2 following SEB stimulation in vivo. Groups of mice were exposed to SEB i.v. and at various time points their serum was taken and analyzed for IL-2 production. We found that in comparison to in vitro IL-2 production, when WT mice were exposed to SEB in vivo they produced IL-2 within 1 hr., which peaked at around 2 hrs. (Fig. 4). In addition, consistent with the in vitro stimulation, ITK deficient mice secrete significantly less IL-2 in response to SEB in vivo (Fig. 4). Based on these data we conclude that in vivo IL-2 production occurs earlier than observed in vitro, and that ITK null T cells continue to exhibit defects in IL-2 production in response to SEB in vivo, even in the presumed presence of adequate costimulatory signals in vivo.
Defective phosphorylation of c-jun induced by SEB in ITK null T cells
The JNK MAPK pathway has been shown to be essential for IL-2 production upon stimulation of T cells by the TcR and CD28 in vitro [25-28]. This pathway leads to phosphorylation of c-jun and activation of an AP-1 transcription factor complex that is required for IL-2 transcription [28]. While activation of the JNK pathway leading to phosphorylation and activation of c-jun has been demonstrated in T cells following TcR and CD28 crosslinking in vitro, as well as by peptide antigen stimulation in vivo, it is not clear whether the SAG SEB activates this pathway in vivo [29,30]. We therefore tested whether SEB could induce phosphorylation of c-jun in WT T cells stimulated with SEB in vivo. To do this, we adapted a method initially used by Jenkins and colleagues to analyze phosphorylation of c-jun following in vivo exposure of antigen. In this protocol, mice were exposed to SEB, then cells from lymph nodes, spleen and blood rapidly isolated and fixed, and antibodies specific for phosphorylated c-jun used to analyze its phosphorylation. In addition, cells were stained with antibodies specific to Vβ8 or Vβ6 to detect SEB responsive and non-responsive T cells respectively. Flow cytometry was employed to detect phosphorylation of c-jun [30]. Figure 5 demonstrates that 1 hr. after intravenous exposure to SEB, Vβ8+ SEB reactive T cells in spleen and lymph nodes contain higher levels of phosphorylated c-jun (cf. Figs. 5a, b iii & iv). Similar results were found in animals exposed to SEB i.p. (data not shown). By contrast, T cells non-reactive to SEB in the same animals, those bearing Vβ6+ TcRs, did not have any increase in phosphorylated c-jun (cf. Figs. 5a, b, i & ii). This demonstrates that in the same animal, only those T cells that interact with and can be activated by SEB respond by phosphorylation of c-jun, while at the same time, those T cells that are not reactive are not activated, demonstrating specificity. Similarly, animals injected with PBS showed no such change in phosphorylated c-jun in either the SEB reactive or non-reactive T cell populations, indicating that this was an SEB mediated event (Fig. 5a–b). Other controls including secondary reagents alone demonstrate the specificity of antibody staining (data not shown).
We next determined whether T cells lacking ITK could induce c-jun phosphorylation in response to SEB activation in vivo. ITK null mice were exposed to SEB as described above, and phosphorylation of c-jun determined in Vβ8+ T cell population (SEB reactive) or Vβ6+ T cell population (non-reactive) in the same animals. These experiments show that in the absence of ITK, SEB does not lead to phosphorylation of c-jun in either T cell population (Fig. 6a–d). Thus, the absence of ITK results in the lack of SEB phosphorylation of c-jun, which could affect the ability of these T cells to produce IL-2. To determine if this lack of c-jun phosphorylation in ITK null T cells in response to SEB reflected global nonresponsiveness in these cells, we determined whether these cells were indeed activated by SEB exposure. To do this, we measured increases in CD69 expression, a cell surface protein that is rapidly expressed by activated T cells in a Ras/MAPK dependent manner [31-33]. Similar experiments were performed as those examining c-jun phosphorylation, except that we used specific antibodies against CD69 to determine if the SEB reactive cells had been activated. Figure 7 demonstrates that Vβ8+CD4+ T cells but not Vβ6+CD4+ T cells from both WT and ITK null mice exhibited increases in CD69 expression 2 hours following SEB exposure, indicating that in both cases, the T cells were responding to SEB activation (Fig. 7). However, SEB induced CD69 expression on a higher percentage of Vβ8+CD4+ SEB responding T cells in WT mice than in ITK null mice (73.5 +/- 12% in WT vs. 43.6 +/- 0.5% in ITK null). However, there was no difference in the MFI of CD69 expression WT responding T cells vs. ITK null responding T cells (362.5 _+/- 6.3 in WT vs. 375 +/- 11.3 in ITK null). In addition, there was no difference in the expression of CD25 on WT or ITK null T cells (data not shown). Thus the lack of SEB induced c-jun phosphorylation in ITK null T cells seems to reflect the specific role of ITK in regulating this event in response to SEB exposure in vivo.
Similar toxicity of SEB on WT and ITK null mice
Exposure of mice to SEB and LPS results in toxicity culminating in death [1]. We therefore analyzed the effect of exposure to SEB/LPS on ITK null mice. However, in contrast to the results with IL-2 secretion and T cell expansion, both mice were affected similarly (Table 1). Thus the same proportion of mice became sick following exposure to SEB and LPS, suggesting that although the absence of ITK affects IL-2 secretion following SEB exposure, its absence does not affect the ability of SEB to induce sickness in these mice.
Discussion
It is well established that exposure to SEB results in large scale cytokine production, which in part is responsible for symptoms of exposure to this toxin [1,2,35]. T cells have been shown to be largely responsible for this excessive cytokine production [35,36]. Current suggestions for pharmacologically blocking the symptoms of SEB exposure include T cell signal transduction inhibitors such as CsA and Perfenidone, however, these agents have significant side effects [37,38]. Here we present a promising target for inhibition IL-2 production induced by SEB exposure, the tyrosine kinase ITK. We show that mice lacking ITK have significantly reduced T cell expansion and IL-2 secretion upon exposure to SEB in vivo. Furthermore, we show that the SEB induced signaling pathway leading to c-jun phosphorylation, an indication of JNK pathway activation, is significantly reduced in ITK null T cells, although these T cells could respond to SEB activation by upregulating CD69, and there was no difference in the overall toxicity of SEB/LPS in these mice compared to WT mice.
Previous work performed in vitro suggests that ITK is required for the anti-TcR/CD3 antibody mediated activation of the transcription factor NFAT, via regulation of calcium influx into TcR stimulated T cells [19,39]. ITK null T cells also exhibit reduced AP-1 DNA binding activity when stimulated in vitro with anti-CD3 antibodies [19]. It has also been shown that JNK activation lies in part downstream of ITK following TcR crosslinking in vitro [23]. However, these experiments were performed in vitro and it is not clear that effects seen in vitro would reflect what happens in vivo, since other cell-cell interactions may allow for activation of pathways that are not seen using in vitro activation with antibodies. Our data however, suggest that indeed, ITK lies downstream of the TcR and is required for IL-2 secretion and c-jun phosphorylation in vivo. Corroborating a role for c-jun activation in IL-2 production, mice carrying a dominant negative c-jun have much reduced IL-2 secretion in vitro [28] (mice lacking c-jun die during embryogenesis [40]). Similarly, JNK has been implicated in IL-2 secretion by T cells in vitro [25-27]. Our data lend support to this idea that the JNK-c-jun pathway is involved in the production of IL-2 by T cells in vivo.
Curiously, the immunosuppresant CsA has been shown to inhibit the effects of SEB exposure in mice if delivered prior to SEB exposure [36], however it does not inhibit SEB induced effects in monkeys if delivered at the same time as SEB [38]. The effects of CsA on SEB induced symptoms has been suggested to be due to its effects on inhibiting IL-2 and other cytokine production due to inhibition of activation of the transcription factor NFAT [36,41], however, CsA also inhibits activation of the JNK pathway following TcR/CD3 and CD28 stimulation [29,30], and so CsA pretreatment may act to prevent early T cell activation of these pathways, thus blocking cytokine production and protecting mice from the effects of subsequent SEB exposure.
Mice lacking ITK exhibit reduced T cell responses in vitro, but also have reduced percentages of CD4+ T cells (approximately 60–70% of WT, [16]). While this would be expected to reduced the overall T cell response in vivo to SEB exposure, if these cells were able to respond to SEB, we would expect an equivalent reduction in the amount of IL-2 secreted in vivo in response to SEB. We would also expect to see an equivalent reduction in proliferation in vitro. However, in vivo, we observed much reduced IL-2 secretion, and reduced expansion of Vβ8+CD4+ T cells, although this was not as dramatic as the reduction in IL-2 secretion. Indeed, it has been observed that SEB induced T cell expansion in IL-2 null mice, indicating that IL-2 is not necessary for expansion of T cells in vivo following SEB exposure. However, even with the reduced T cell responses observed in ITK null mice, they still suffer from SEB/LPS induced toxicity similar to that seen in the WT mice. It should be noted however, that C57BL/6 mice are much less sensitive to SEB induce lethal shock than Balb/c mice [34]. Thus we observed very few deaths in our experiments, and ITK may protect Balb/c mice from actual death. We are currently crossing these mice onto the Balb/c background to determine if this is the case.
Conclusion
In conclusion, we have show that ITK is required for IL-2 production induced by SEB in vivo and in vitro. We have also shown that ITK may regulate signals leading IL-2 production in part by regulating phosphorylation of c-jun. However, mice lacking ITK exhibit similar responses to SEB toxicity. The data suggest that the inability of T cell lacking ITK to produce IL-2 cannot be overcome by SAG stimulation, and that perturbing T cell activation pathways leading to IL-2 production does not necessarily lead to improved responses to SEB toxicity.
Methods
Mice
Wild type mice and ITK-deficient mice (C57BL/6 background, [16]) between 8 and 10 weeks were used in all experiments. RAG1-/- mice were a kind gift of Dr. Eric Harvill (Penn State University) and were similarly used between 6–10 weeks of age. The Institutional Animal Care and Use Committee of The Pennsylvania State University approved all experiments.
In vivo expansion assays
WT and ITK deficient mice were injected intraperitoneally (i.p.) with 50 μg SEB in PBS (Sigma-Aldrich, St. Louis, MO) and after 2 days, were sacrificed and splenocytes stained with antibodies specific for Vβ8 (SEB reactive T cells) or Vβ6 (non-reactive T cells) and CD4 directly conjugated to FITC and PE respectively (BDPharmingen, San Diego, CA). Alternatively, mice were injected i.p. with 50 μg SEB and eye-bled every 2 days for 5 days. Following lysis of red blood cells, the remaining cells were stained as described above for the splenocytes. Lymphocytes were identified by flow cytometry by their forward and side scatter characteristics.
Intracellular staining
We used a protocol reported by Zell et al to determine the phosphorylation status of c-jun, as a measure of activation of the JNK-c-jun pathway in SEB responding T cells. WT and ITK deficient mice were injected i.p. or intravenously (i.v.) with 50 μg SEB and allowed to survive for an 1 hr. Similar results were found with both routes of administration. Animals were sacrificed and spleens, lymph nodes, and blood rapidly isolated, and dounced in 2% paraformaldehyde to rapidly fix the cells as previously described [30]. Cells were stained with antibodies to Vβ8 or Vβ6 directly conjugated to PE (BDPharmingen, San Diego, CA), then permeabilized with saponin. Cells were then stained for intracellular phosphorylated-c-jun with a monoclonal antibody against phosphorylated c-jun (IgG1, Cell Signaling, Beverly, MA), followed by biotinylated rabbit anti-mouse IgG1 and streptavidin conjugated to FITC. Controls included secondary reagents alone. Upregulation of CD69 by SEB in vivo was performed using similar approaches, except that T cells were not fixed or permeabilized prior to staining with PE-conjugated anti-CD69 monoclonal antibody (BDPharmingen, San Diego, CA). This was followed by analysis by flow cytometry, with post analysis performed using the WinMDI program 2.8 (The Scripps Research Institute, La Jolla, CA).
In vitro proliferation
Lymph nodes and spleens were isolated from WT and ITK deficient mice and pooled. Red blood cells were removed using ACK lysis buffer, and cells resuspended in complete RPMI. Cells were then left unstimulated or stimulated with varying concentrations of SEB (5 μg/ml, 0.5 μg/ml, 0.05 μg/ml, 0.005 μg/ml, and 0.0005 μg/ml) at a concentration of 2 × 106 cells/ml. These cells were incubated at 37°C for 5 days and pulsed once a day for 5 days with 0.5 μCi tritiated thymidine per well for 12 hrs. before harvest. T cells were purified from WT and ITK null mice using MiniMACS separation columns (Miltenyi Biotec Inc, Auburn, CA), and plated at 2 × 106 cells/ml and stimulated as above, in the presence of splenocytes from RAG null mice pre-treated with mitomycin C. All experiments were done in triplicate.
In vitro cytokine analysis
Pooled lymphocytes and splenocytes from WT and ITK deficient mice were incubated with varying concentrations of SEB (5 μg/ml, 0.5 μg/ml, 0.05 μg/ml, 0.005 μg/ml, and 0.0005 μg/ml) for 5 days as described above for in vitro proliferation. Supernatants were sampled once a day over this period and analyzed for IL-2 using ELISAs according to the manufacturer's recommendations (BDPharmingen, San Diego, CA).
In vivo cytokine analysis
WT and ITK deficient mice were injected with 50 μg SEB i.v., and serum isolated from cardiac blood from mice 1, 2, 4, 8, 12, and 24 hrs. post SEB exposure and used to perform an ELISA to determine IL-2 levels. PBS injected mice served as controls, and there was no change in IL-2 secretion in these animals.
Statistics
Values were compared using student's t test and considered significant if p < 0.05.
List of abbreviations used
ITK Interleukin-2 inducible T cell Kinase
IL-2 Interleukin-2
SEB Staphylococcal Enterotoxin B
SAG Superantigen
Th1 T-helper 1 cells
Th2 T-helper 2 cells
Syk Spleen tyrosine kinase
Zap-70 Zeta chain Associated protein-70
Lck Lstra cell kinase
MAPK Mitogen Activated Protein Kinase
ERK Extracellular signal Regulated Kinase
JNK c-jun N-terminal kinase
NK Natural Killer cells
LCM Lymphocytic Choriomeningitis
VS Vesicular Stomatitis
PMA Phorbol Myristic Acid
PBS Phosphate Buffered Saline
NFAT Nuclear Factor of Activated T cells
AP-1 Activator protein 1
CsA Cyclosporin A
Authors' contributions
AA designed the experiments. MJR and JH performed experiments leading to figure 1. MJR performed experiments leading to figures 2, 3, 4, 5, 6, 7. AA and AJH provided financial support and supervision. MJR, AJH and AA wrote the manuscript. All authors read and approved the final manuscript.
Acknowledgements
We thank members of the August lab and the Center for Molecular Immunology & Infectious Disease at Penn State for helpful comments and suggestions. We also thank Elaine Kunze and Susan Magargee in the Center for Quantitative Cell Analysis at Penn State for excellent technical help. We also thank Dr. Dan Littman (New York University Medical School, NY, NY) for kindly providing us with the ITK null mice.
This work was supported in part by a Johnson & Johnson Focused Giving Grant (to A.A.), the American Heart Association (0330036N), and Public Health Service Grants AI-51626 (to A.A.) and AI-46261 (to A.J.H.). MJR is a Ford Foundation Scholar.
Figures and Tables
Figure 1 ITK regulates proliferation in vitro in response to SEB. Splenocytes and lymph node cells from WT (filled circles) and ITK deficient (open circles) mice were stimulated in vitro with the indicated concentrations of SEB, and proliferative response determined. Time course of proliferation over the indicated doses of SEB a) 5 μg/ml SEB, b) dose curve of 0.0005, 0.005, 0.05, 0.5, & 5 μg/ml SEB analyzed at 5 days post stimulation. Purified WT (filled circles) and ITK deficient (open circles) T cells were stimulated in vitro over the indicated time with c) 10 μg/ml SEB, or a dose curve of 0.01, 0.1, 1, and 10 μg/ml SEB and analyzed at d) 3 days, e) 4 days or f) 5 days post stimulation. * indicates statistically significant difference with a p < 0.05.
Figure 2 ITK regulates IL-2 secretion in vitro in response to SEB. Splenocytes and lymph node cells from WT (filled circles) or ITK deficient (open circles) mice were stimulated in vitro with the indicated concentrations of SEB and IL-2 levels secreted in supernatants determined. a) Time course of IL-2 secretion at 5 μg/ml SEB dose. b) IL-2 secretion over the indicated times and doses. (i) Day 1; (ii) Day 2; (iii) Day 3; (iv) Day 4 and (v) Day 5. * indicates statistically significant difference with a p < 0.05.
Figure 3 Reduced T cell expansion in response to SEB exposure in vivo. a) WT (filled circles) or ITK deficient mice (open circles) were injected with 50 μg SEB i.p. and peripheral blood lymphocytes analyzed after 2, 4 and 5 days as in the materials and methods for the percentage of Vβ8+CD4+ T cells as indicated (Vβ6+CD4+ T cells showed no change, data not shown). b) The same data plotted as fold increase in percentage of Vβ8+CD4+ T cells in peripheral blood. * indicates a statistically significant difference with a p < 0.05.
Figure 4 ITK null mice secrete less IL-2 in response to SEB exposure in vivo. WT (filled circles) and ITK deficient (open circles) mice were injected i.v. with 50 μg SEB and blood sampled after 1, 2, 4, 8, 12 and 24 hrs. post injection. IL-2 in serum was determined by ELISA.
Figure 5 SEB induces activation of the JNK pathway specifically in responding T cells in vivo. WT mice were injected with 50 μg SEB or PBS i.v. and sacrificed after 1 hr. Spleen and lymph nodes were harvested and analyzed for the presence of phosphorylated c-jun in the Vβ8+ and Vβ6+ T cell populations. a) Spleen: (i) phosphorylation of c-jun in Vβ6+ T cells from PBS injected mice. (ii) phosphorylation of c-jun in Vβ6+ T cells from SEB injected mice. (iii) phosphorylation of c-jun in Vβ8+ T cells from the same PBS injected mice in (i). (iv) phosphorylation c-jun in Vβ8+ T cells from the same SEB injected mice in (ii). b) Lymph Node: (i) phosphorylation of c-jun in Vβ6+ T cells from PBS injected mice. (ii) phosphorylation of c-jun in Vβ6+ T cells from SEB injected mice. (iii) phosphorylation of c-jun in Vβ8+ T cells from the same PBS injected mice in (i). (iv) phosphorylation c-jun in Vβ8+ T cells from the same SEB injected mice in (ii). Arrow points to the population of cells responding to SEB stimulation by activation of the JNK pathway by inducing c-jun phosphorylation.
Figure 6 ITK is required for SEB mediated induction of c-jun phosphorylation in responding T cells in vivo. WT (top panels) or ITK null (bottom panels) mice were injected with 50 μg SEB or PBS i.v. and sacrificed after 1 hr. Splenocytes were then analyzed for the presence of phosphorylated c-jun in the Vβ8+ and Vβ6+ T cell populations. a) Vβ6+ T cells from WT mice; b) Vβ8+ T cells from WT mice; c) Vβ6+ T cells from ITK null mice; d) Vβ8+ T cells from ITK null mice.
Figure 7 ITK is not necessary for SEB induced upregulation of CD69 inresponding T cells in vivo. WT (top panels) or ITK null (bottom panels) mice were injected with 50 μg SEB or PBS i.v. and sacrificed after 2 hrs. Splenocytes were then analyzed for the presence of CD69 on the Vβ8+ and Vβ6+ T cell populations. a) Vβ6+ T cells from WT mice; b) Vβ8+ T cells from WT mice; c) Vβ6+ T cells from ITK null mice; d) Vβ8+ T cells from ITK null mice.
Table 1 Effect of SEB on health and survival of WT and ITK null mice. Mice (WT or ITK null) were injected with the indicated amount of LPS, followed 4 hours later by 50 μg SEB, both delivered Ip. Alternatively, mice were injected with 20 mg. D-Gal and 50 μg SEB at the same time. All injections were delivered intraperitoneally. Mice were then monitored for the presence of ruffled fur, mucous in feces and lethargy. *mice had ruffled fur, mucous in feces and were lethargic. **mouse died over the course of the experiment. ***mouse had to be euthanized due to severity of sickness.
Mice Challenge # Sick/n* # Dead/n
WT 50 μg SEB + 150 μg LPS 14/15 1/15**
ITK-/- 50 μg SEB + 150 μg LPS 11/12 1/12***
WT 50 μg SEB + 20 mg D-Gal 10/11 1/11**
ITK-/- 50 μg SEB + 20 mg D-Gal 11/11 0/11
==== Refs
Dinges MM Orwin PM Schlievert PM Exotoxins of Staphylococcus aureus. Clin Microbiol Rev 2000 13 16 34 10627489
Kotzin BL Leung DY Kappler J Marrack P Superantigens and their potential role in human disease. Adv Immunol 1993 54 99 166 8397479
Miethke T Wahl C Heeg K Echtenacher B Krammer PH Wagner H T cell-mediated lethal shock triggered in mice by the superantigen staphylococcal enterotoxin B: critical role of tumor necrosis factor. J Exp Med 1992 175 91 98 1730929 10.1084/jem.175.1.91
MacDonald HR Baschieri S Lees RK Clonal expansion precedes anergy and death of V beta 8+ peripheral T cells responding to staphylococcal enterotoxin B in vivo. Eur J Immunol 1991 21 1963 1966 1907921
Kawabe Y Ochi A Selective anergy of V beta 8+,CD4+ T cells in Staphylococcus enterotoxin B-primed mice. J Exp Med 1990 172 1065 1070 1976734 10.1084/jem.172.4.1065
Rellahan BL Jones LA Kruisbeek AM Fry AM Matis LA In vivo induction of anergy in peripheral V beta 8+ T cells by staphylococcal enterotoxin B. J Exp Med 1990 172 1091 1100 2212944 10.1084/jem.172.4.1091
Takesono A Finkelstein LD Schwartzberg PL Beyond calcium: new signaling pathways for Tec family kinases. J Cell Sci 2002 115 3039 3048 12118060
August. A Fischer. AM Hao. S Mueller. C Ragin. M The Tec Family of tyrosine kinases in T cells, amplifiers of T cell receptor signals. Int J Biochem & Cell Biol 2002 34 1184 1189 12127569 10.1016/S1357-2725(02)00068-7
Miller AT Berg LJ New insights into the regulation and functions of Tec family tyrosine kinases in the immune system. Curr Opin Immunol 2002 14 331 340 11973131 10.1016/S0952-7915(02)00345-X
Nel AE T-cell activation through the antigen receptor. Part 1: signaling components, signaling pathways, and signal integration at the T-cell antigen receptor synapse. Erratum in: J Allergy Clin Immunol 2002 Jul;110(1):25. J Allergy Clin Immunol 2002 109 758 770 11994696 10.1067/mai.2002.124259
Gibson. S August. A Kawakami. Y Kawakami. T Dupont. B Mills. GB The EMT/ITK/TSK (EMT) tyrosine kinase is activated during TcR signaling: LCK is required for optimal activation of EMT. J Immunol 1996 156 2716 2722 8609388
Heyeck. SD Berg. LJ Developmental regulation of a murine T-cell specific tyrosine kinase gene, Tsk. Proc Natl Acad Sci U S A 1993 90 669 673 8421704
Yamada. N Kawakami. Y Kimura. H Fukamachi. H Baier. G Altman. A Kato. T Inagaki. Y Kawakami. T Structure and expression of novel protein-tyrosine kinases, Emb and Emt, in hematopoietic cells. Biochem Biophys Res Commun 1993 192 231 240 8476425 10.1006/bbrc.1993.1404
August. A Gibson. S Kawakami. Y Kawakami. T Mills. GB Dupont. B CD28 is associated with and induces the immediate tyrosine phosphorylation and activation of the Tec family kinase ITK/EMT in the human Jurkat leukemic T-cell line. Proc Natl Acad Sci USA 1994 91 9347 9351 7524075
Liu KQ Bunnell SC Gurniak CB Berg LJ T cell receptor-initiated calcium release is uncoupled from capacitative calcium entry in Itk-deficient T cells. J Exp Med 1998 187 1721 1727 9584150 10.1084/jem.187.10.1721
Liao. XC Littman. DR Altered T cell receptor signaling and disrupted T cell development in mice lacking Itk. Immunity 1995 3 757 769 8777721 10.1016/1074-7613(95)90065-9
Mueller C August A Attenuation of Immunological Symptoms of Allergic Asthma in Mice Lacking the Tyrosine Kinase ITK. J Immunol 2003 170 5056 5063 12734350
Schaeffer. EM Debnath. J Yap. G McVicar. D Liao. XC Littman. DR Sher. A Varmus. HE Lenardo. MJ Schwartzberg. PL Requirement for Tec kinases Rlk and Itk in T cell receptor signaling and immunity. Science 1999 284 638 641 10213685 10.1126/science.284.5414.638
Schaeffer. EM Yap. GS Lewis. CM Czar. MJ McVicar. DW Cheever. AW Sher. A Schwartzberg. PL Mutation of Tec family kinases alters T helper cell differentiation. Nat Immunol 2001 2 1183 1188 11702066 10.1038/ni734
Liao. XC Fournier. S Killeen. N Weiss. A Allison. JP Littman. DR Itk negatively regulates induction of T cell proliferation by CD28 costimulation. J Exp Med 1997 186 221 228 9221751 10.1084/jem.186.2.221
Bachmann MF Littman DR Liao XC Antiviral immune responses in Itk-deficient mice J Virol 1997 71 7253 7257 9311799
Uchiyama T Kamagata Y Yan XJ Kawachi A Fujikawa H Igarashi H Okubo M Relative strength of the mitogenic and interleukin-2-production-inducing activities of staphylococcal exotoxins presumed to be causative exotoxins of toxic shock syndrome: toxic shock syndrome toxin-1 and enterotoxins A, B and C to murine and human T cells. Clin Exp Immunol 1989 75 239 244 2784735
Miller AT Berg LJ Defective Fas ligand expression and activation-induced cell death in the absence of IL-2-inducible T cell kinase. J Immunol 2002 168 2163 2172 11859102
Wilcox HM Berg LJ Itk phosphorylation sites are required for functional activity in primary T cells. J Biol Chem 2003 278 37112 37121 12842872 10.1074/jbc.M304811200
Sabapathy K Hu Y Kallunki T Schreiber M David JP Jochum W Wagner EF Karin M JNK2 is required for efficient T-cell activation and apoptosis but not for normal lymphocyte development. Curr Biol 1999 9 116 125 10021384 10.1016/S0960-9822(99)80065-7
Nishina H Bachmann M Oliveira-dos-Santos AJ Kozieradzki I Fischer KD Odermatt B Wakeham A Shahinian A Takimoto H Bernstein A Mak TW Woodgett JR Ohashi PS Penninger JM Impaired CD28-mediated interleukin 2 production and proliferation in stress kinase SAPK/ERK1 kinase (SEK1)/mitogen-activated protein kinase kinase 4 (MKK4)-deficient T lymphocytes. J Exp Med 1997 186 941 953 9294148 10.1084/jem.186.6.941
Sabapathy K Kallunki T David JP Graet I Karin M Wagner E c-Jun NH2-terminal Kinase (JNK)1 and JNK2 have similar and stage-dependent roles in regulating T cell apoptosis and proliferation. J Exp Med 2001 193 317 328 11157052 10.1084/jem.193.3.317
King LB Tolosa E Lenczowski JM Lu F Lind EF Hunziker R Petrie HT Ashwell JD A dominant-negative mutant of c-Jun inhibits cell cycle progression during the transition of CD4(-)CD8(-) to CD4(+)CD8(+) thymocytes. Int Immunol 1999 11 1203 1216 10421778 10.1093/intimm/11.8.1203
Su. B Jacinto. E Hibi. M Kallunki. T Karin. M Ben-Neriah. Y JNK is involved in signal integration during costimulation of T lymphocytes. Cell 1994 77 727 736 8205621 10.1016/0092-8674(94)90056-6
Zell. T Khoruts. A Ingulli. E Bonnevier. JL Mueller. DL Jenkins. MK Single-cell analysis of signal transduction in CD4 T cells stimulated by antigen in vivo. Proc Natl Acad Sci U S A 2001 98 10805 10810 11535838 10.1073/pnas.191567898
Denny MF Patai B Straus DB Differential T-cell antigen receptor signaling mediated by the Src family kinases Lck and Fyn. Mol Cell Biol 2000 20 1426 1435 10648627 10.1128/MCB.20.4.1426-1435.2000
Taylor-Fishwick DA Siegel JN Raf-1 provides a dominant but not exclusive signal for the induction of CD69 expression on T cells. Eur J Immunol 1995 25 3215 3221 8566003
D'Ambrosio D Cantrell DA Frati L Santoni A Testi R Involvement of p21ras activation in T cell CD69 expression. Eur J Immunol 1994 24 616 620 7907294
Stiles BG Campbell YG Castle RM Grove SA Correlation of temperature and toxicity in murine studies of staphylococcal enterotoxins and toxic shock syndrome toxin 1. Infect Immun 1999 67 1521 1525 10024605
Anderson MR Tary-Lehmann M Staphylococcal enterotoxin-B-induced lethal shock in mice is T-cell-dependent, but disease susceptibility is defined by the non-T-cell compartment. Clin Immunol 2001 98 85 94 11141330 10.1006/clim.2000.4960
Marrack P Blackman M Kushnir E Kappler J The toxicity of staphylococcal enterotoxin B in mice is mediated by T cells. J Exp Med 1990 171 455 464 2303780 10.1084/jem.171.2.455
Hale ML Margolin SB Krakauer T Roy CJ Stiles BG Pirfenidone blocks the in vitro and in vivo effects of staphylococcal enterotoxin B. Infect Immun 2002 70 2989 2994 12010989 10.1128/IAI.70.6.2989-2994.2002
Komisar JL Weng CF Oyejide A Hunt RE Briscoe C Tseng J Cellular and cytokine responses in the circulation and tissue reactions in the lung of rhesus monkeys (Macaca mulatta) pretreated with cyclosporin A and challenged with staphylococcal enterotoxin B. Toxicol Pathol 2001 29 369 378 11442023 10.1080/019262301316905336
Fowell. DJ Shinkai. K Liao. XC Beebe. AM Coffman. RL Littman. DR Locksley. RM Impaired NFATc Translocation and Failure of Th2 Development in Itk Deficient CD4 T Cells. Immunity 1999 11 399 409 10549622 10.1016/S1074-7613(00)80115-6
Johnson RS van Lingen B Papaioannou VE Spiegelman BM A null mutation at the c-jun locus causes embryonic lethality and retarded cell growth in culture. Genes Dev 1993 7 1309 1317 8330736
Clipstone NA Crabtree GR Identification of calcineurin as a key signalling enzyme in T-lymphocyte activation. Nature 1992 357 695 697 1377362 10.1038/357695a0
Kneitz B Herrmann T Yonehara S Schimpl A Normal clonal expansion but impaired Fas-mediated cell death and anergy induction in interleukin-2-deficient mice. Eur J Immunol 1995 25 2572 2577 7589128
|
16042784
|
PMC1200558
|
CC BY
|
2021-01-04 16:03:31
|
no
|
BMC Immunol. 2005 Jul 22; 6:19
|
utf-8
|
BMC Immunol
| 2,005 |
10.1186/1471-2172-6-19
|
oa_comm
|
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-711598516110.1186/1471-2458-5-71Case ReportInternational public health research involving interpreters: a case study from Bangladesh Pitchforth Emma [email protected] Teijlingen Edwin [email protected] Department of Health Sciences, University of Leicester, 22–28 Princess Road West, Leicester, LE1 6TP, UK2 Edwin van Teijlingen, Department of Public Health and Dugald Baird Centre for Research on Women's Health, University of Aberdeen, Medical School, AB25 2ZD, UK2005 28 6 2005 5 71 71 1 9 2004 28 6 2005 Copyright © 2005 Pitchforth and van Teijlingen; licensee BioMed Central Ltd.2005Pitchforth and van Teijlingen; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Cross-cultural and international research are important components of public health research, but the challenges of language barriers and working with interpreters are often overlooked, particularly in the case of qualitative research.
Methods
A case-study approach was used to explore experiences of working with an interpreter in Bangladesh as part of a research project investigating women's experiences of emergency obstetric care.
The case study
Data from the researcher's field notes provided evidence of experiences in working with an interpreter and show how the model of interviewing was adapted over time to give a more active role to the interpreter. The advantages of a more active role were increased rapport and "flow" in interviews. The disadvantages included reduced control from the researcher's perspective. Some tensions between the researcher and interpreter remained hard to overcome, irrespective of the model used. Independent transcription and translation of the interviews also raised questions around accuracy in translation.
Conclusion
The issues examined in this case study have broader implications for public health research. Further work is needed in three areas: 1) developing effective relationships with interpreters; 2) the impact of the interpreter on the research process; and 3) the accuracy of the translation and level of analysis needed in any specific public health research. Finally, this paper highlights the importance to authors of reflecting on the potential impact of translation and interpretation on the research process when disseminating their research.
==== Body
Background
Cross-cultural and international research can be illuminating in the public health field, but some important aspects of conducting such research remain under-evaluated. Issues of research practice relating to language are particularly important.[1] To date, much attention has been focused on the conduct of quantitative survey-type research, including the adaptation of questionnaires to achieve equivalence when translating generic health-related measures from one language into another.[1,2] Gaining conceptual equivalence or comparability of meaning in questionnaires is difficult, especially when collecting data in one language and reporting in another.[1,2] Problems of language and cultural translation have real impacts on research outcomes. For example, a review of prevalence data in UK surveys on tobacco and alcohol in ethnic minority groups showed that inadequate cross-cultural adaptation was a potential explanation for discrepancies in reported prevalence.[1]
Conducting qualitative research raises distinctive issues. Unlike questionnaire-based surveys, it may be much more difficult to plan precisely what is going to be said and how. This is partly the case because the interview is a dialogue and certain comments made by interviewees will lead the interviewer to ask different follow-up questions to some interviewees than to others. It is also partly due to the fact that "stories being narrated are constructed in the moments of the interview to the extent that neither the interviewers nor the interviewees can predict the details of what is going to be discussed in advance" (Nunkoosing 2005: 703).[1]
Qualitative research in cross-cultural contexts often relies on interpreters. While there has been a general call to make decisions regarding the use of interpreters and translation in research more explicit using interpreters in public health research is under-researched. [4,8] Several recommendations do exist in other fields and two suggestions are (a) that clarification of roles between the researcher and interpreter before the interview can help to avoid potential problems; [9] and (b) that co-operative working is more likely to be achieved where the researcher appreciates the interpreter's role as actively participative. [10]
This paper examines experiences of working with a lay interpreter during public health qualitative research in Bangladesh. Using a case-study approach, it presents two possible models for working with interpreters used in the study and considers the impact of translation on both the research process and findings.
Methods
A case study is a research methodology that focuses on the circumstances, dynamics and complexity of a single case or small number of cases.[11] This case study considers the dynamics in cross-cultural research involving an interpreter in addition to the researcher and participants. A case study approach allows detailed examination of the research process and aspects of research practice. [12] An important role for case studies can be the revelation of phenomena that would otherwise be cut off from those at which the work is aimed,[13] and in this instance, involves important consideration of the effects of using an interpreter on the research process and findings.
Case study
Ethical approval was gained from the Bangladesh Medical Research Council for our research project investigating experiences of women utilizing emergency obstetric care at a large teaching hospital. The research involved a questionnaire administered orally with women as they were admitted to hospital and a more in-depth interview at their home following discharge. The focus of this case study is on the post-discharge interviews; the purpose of this paper is not to present the findings from these interviews, but to examine in detail the methods and effects of working with an interpreter on the research process.
The aim of the interviews was to gather details of the women's 'journey of care' from the decision to go to hospital through to approximately four weeks after discharge. Women and their families were asked in particular about organisational issues, costs incurred and raising the finances required for their treatment episode. Such organisational issues were often the responsibility of family members or neighbours and these people were identified and invited to participate in the interview with the woman who had undergone hospital treatment.
The researcher (EP) had limited understanding of Bangla and employed a lay interpreter for the data collection. The interpreter worked with EP on a full-time basis for a period of six months. The background characteristics of the researcher, interpreter and main participants are presented in Table 1.
Table 1 Characteristics of researcher, interpreter and participants
The researcher The interpreter The participants
Female Female Female
Aged 25 Aged 25 Aged 15–45
Unmarried Unmarried Married
No children No children Children
Ph.D. student Master's education Limited formal education
Non-poor Non-poor Poor
Christian Muslim Muslim/Hindu
Scottish Bangladeshi Bangladeshi
No experience working with interpreter No experience working as interpreter Unknown experience of research
The interpreter was identified through a research and evaluation unit at a large non-government organisation in Bangladesh. She had excellent spoken and written English but was not trained as an interpreter, although she did have experience of research interviewing and had a health-related postgraduate qualification. The researcher and the interpreter spent considerable time discussing how best to carry out the interviews. Following previous advice, this included discussion over the respective roles of the interpreter and researcher. For example, the interpreter was asked to interpret all of the participants' answers and the researcher's questions. The researcher stressed that it was important to hear participants' responses even if the question or anticipated answer seemed obvious to the interpreter. The interview schedule was devised and written in English and time was spent ensuring that it would be culturally acceptable to the interviewees.
Interview process
Figure 1 illustrates the model of interviewing that was adopted for the initial interviews. This reflects a passive model, in which the researcher asked questions through the interpreter, who would then interpret the response from the participant to the researcher. This method was adopted because it enabled the researcher to follow the entire interview and to ask further questions, as required.
Figure 1 Passive interpreter model.
During the interviews it became clear that there were a number of disadvantages to this approach. The interpretation back and forth made the interviews time consuming. The interviews often brought the women and their families away from other work and this model of interviewing seemed to reduce the focus on the participants, as they had to wait while information was interpreted. This model also involved some tensions between the researcher and the interpreter, including occasions when the interpreter did not interpret either the question to the participant or the answer back to the researcher. There were other such occasions when the interpreter clearly felt a question was obvious or did not want to ask it and would sigh loudly or raise her eyebrows. A benefit of qualitative interviewing is that the interviews can flow like a conversation rather than a structured question and answer situation – a guided conversation with a purpose.[14] This flow was restricted as the interpreting interrupted the dialogue between participants and the interpreter and in practice the interviews often became disjointed.
The researcher and interpreter reflected on the progress of the interviews and it was felt that a new model of interviewing should be adopted. The interviews were semi- structured and the interpreter had quickly become familiar with the aims of the interview and interview schedule. The model of interviewing was subsequently changed to an active model, so that the interpreter carried out most of the interview (Figure 2), with the aim of improving the flow of interviews and making the interviews less burdensome. The researcher was always present and at key points the interpreter would summarise the interview to the researcher to allow additional questions to be asked. With the participants' permission, all of the interviews were audio-taped and transcribed so that the full interview transcript was available to the researcher.
Figure 2 Active interpreter model.
This second model did allow interviews to flow more like a conversation, enabling the interpreter to build a better rapport with participants and reduced the time burden for participants. Allowing the interpreter to take more control in the interview also reduced some of the tensions inherent in the first model, but it did so by reducing the control and involvement of the researcher (situated outside the circle in Figure 2). Of course, in either model the researcher has reduced control compared to direct interviewing between researcher and participant.
Some differences arose irrespective of the model used. For example, the researcher and interpreter differed as to how the role of the researcher-interpreter team should be perceived by the respondents. The researcher felt that it would be important to emphasise that they were not associated with the hospital so that the women and their families felt that they could talk freely about their experiences.[15] It was hoped that the researcher was not associated with any form of authority.[16] The interpreter, however, placed less importance on this, introducing herself as a nutritionist working for a non-government organization, backed up with the name of a well-known doctor. The interpreter also placed less emphasis on introductions and informed consent. She felt that if families did not want to talk to us they would not do so. The researcher and interpreter had, in addition, discussed and (apparently) agreed that it was not appropriate to be judgemental about decisions made by families. However the researcher, with growing understanding of Bangla, was aware of occasions when the interpreter would take issue with the participants, e.g. when a woman reported that she had not attended antenatal check-ups during her pregnancy.
Transcribing, interpreting and translating
With permission, all of the interviews were audio taped and transcribed in English. This was done by the same interpreter. A thematic analysis of the transcripts was planned and as a quality-control measure to check accuracy and validity, four interviews were transcribed by a bilingual interpreter in the UK. Comparisons were to be made within and across themes and to show common and differing experiences among women.[17,18]
The transcripts from the original and independent translations were compared and contrasted. Some differences were noted but overall the researchers were reassured that they were similar enough for the purpose of this public health study. Any differences tended to be in over-interpretation of women's own words, level of precision and emphasis.
The extract below shows, for example, that there may have been a tendency on the part of the lay interpreter to "interpret" the women's words rather than directly translate. In the original transcripts the word "forceps" is used but in the independent translation, the description of a metal cup on the baby's head is more likely to have been the words used by the woman:
"There the doctors also tried but they failed to get my delivery. They tried to deliver by forceps also" (original translation)
Compared to the less technical:
"The doctors tried to take the baby out by using a metal cup on the baby's head but they were unsuccessful" (independent translation)
There were also occasions when the different transcripts did not vary considerably in content but the emphasis may have been interpreted differently. For example the tone of the doctors is not portrayed in the same way in the following two examples.
"One doctor came to me to say that it is not possible for us, you have to take your patient. After a while the nurses also said the same thing. Then they took me to Dhaka Medical College Hospital" (original translation)
Compare this with the following:
"One doctor said, 'we can't do it here. Take her away!' Later seven nurses tried, but then we won't be able to do it. Then my father took me away" (independent translation)
Similarly, the example below shows the differences in the level of detail included in the transcripts. The interpreter asked:
"Was the ambulance arranged by Red Crescent Hospital or by your own relatives?"
The original translation of the reply was:
"By Red Crescent Hospital" (original translation)
Whilst the independent translation was more subtle:
"At first they told us to make our own way – but after my uncle spoke to them, they arranged to bring me in the ambulance" (independent translation)
However, both excerpts show that the Red Crescent Hospital arranged the ambulance but the detail included in the independent translation gives greater insight into the process involved.
We are very well aware that different academic disciplines might approach the detail of transcribing and/or analysing differently. For example, a student of Psychology might be interested in the 'pauses' and 'hmms' and 'sighs' in the conversation to establish how emotionally challenging the conversation is for the woman, whilst a student of Women's Studies might be interested in the power differences between the female interviewees and their male family members, whilst a student of Management Studies might focus on the way health care professionals manage the woman's situation within their limited resources. In Public Health we are often interested in groups of people rather than individual stories or events. As such, in more applied studies like this one, detailed transcription and some differences in the translations are not necessarily as significant as they may be in other disciplines.
Discussion
This case study shows some of the challenges that can be faced when conducting qualitative research with an interpreter. The problems identified are not necessarily new, but we would argue that the advice that many of them can be avoided by clarifying the role of the researcher and interpreter might be somewhat naïve. For example, it has been suggested that the respective roles of the researcher and interpreter and rules of research governance could be established before the interview. In practice, however, we found the reduced control inherent in relying on an interpreter meant that the agreed procedure was not always followed. Using the interpreter in a traditional passive role introduced tensions into the research interviews and was burdensome for the participants. The alternative 'active-interpreter' model that was adopted gave the impression of overcoming these tensions but did so by effectively excluding the researcher for part of the interview. Again, the view that the researcher can avoid such problems by seeing the interpreter's role as actively participative seemed difficult to implement.
This paper has used a case study approach to illustrate the issues involved in using an interpreter in cross-cultural research. Common criticisms of case studies are that they provide little basis for scientific generalization and that there can be difficulty in assessing the importance of relationships, which may be simply idiosyncratic to one particular case.[18,20] The nature of case studies, however, means that the extrapolation of findings does not depend on the representativeness of the case, but on the clarity of the theoretical reasoning.[21] The generalisability of the findings in this case study would need to be established; it is likely that the nature and context of the research and interviews would have important effects. That said, case studies are well suited to areas where previous research is lacking and in this case highlights that more practical advice is needed on developing effective relationships between the researcher-interpreter team.
In this respect our case study demonstrates vividly the importance of reflexivity in conducting this type of research. Reflexivity is widely encouraged within qualitative research as a means for researchers to reflect upon, critically examine and analytically explore the nature of the research to demonstrate assumptions about gender or other relationships which are built into the research.[22] The process of translation and use of interpreters is an important part of such reflexive methodology.[10] It can be argued that research involving an interpreter can add additional layers of bias within the interview process.[23] Jentsch showed diagrammatically how the background characteristics, psychological factors and behavioural factors of the researcher, interpreter and respondents may interact.[9] The background characteristics and behaviour may have an impact on psychological factors (such as attitude and expectations), which in turn influence behaviour.
Taking the background characteristics of all three parties in our case study (Table 1), there are clear differences but also some similarities. First and foremost, all were the same gender. To do research in a traditional society such as Bangladesh requires cultural sensitivity from the part of the researcher and research on such as female oriented topic also requires a female researcher and interpreter. [24] The interpreter shared some characteristics with the researcher and some with the respondents. An advantage of using an interpreter, born and brought up in Bangladesh, was that she was able to act as a 'cultural interpreter', giving a greater insight into the interviews with the women and Bangladesh more generally. The respondents could also relate immediately to the interpreter which helped facilitate the interviews.[9]
Issues of sameness and difference have been highlighted in the literature by drawing on the influence of 'insider' and 'outsider' status.[25-27] Gender, racial identity, social class and shared experience can affect the research process and willingness of respondents to talk to the researchers.[25] It has been argued that the relationship between interviewer and participant should be non-hierarchical to best understand their life experiences. [28] The simple comparison of background characteristics in our case study shows that, aside from gender, the researcher could definitely be considered an 'outsider' but this need not be detrimental to the research and it was felt that the differences between all three parties may have been beneficial. The obvious differences in background characteristics such as wealth and educational attainment gave opportunity for those involved to articulate their feelings about their different life experiences.[29]
In addition to the effect of an interpreter on the research process it is important to make explicit decisions about the use of translations and presentation of findings. The use of only one interpreter was seen as advantageous as it helps ensure reliability in translation.[30] The independent translation of a small number of interviews was then used to assess validity.[30,31] The case study has highlighted the differences rather than similarities between the transcripts and, overall, it was judged that, for this research project, the differences were not sufficiently significant to impact on the thematic analysis undertaken. In contrast, it could have raised significant problems had our study been to analyse (1) how women describe the decision-making process, for example, around transfer to hospital; or (2) the lay understanding of maternity service provisions. The differences in language that were highlighted did not effect our interpretation of the findings. Interestingly, they suggested that the original interpreter was more likely to medicalise terms and was less detailed. It is not that one translation should be considered right or wrong, but the value is in appreciating differences in interpretation in order to discuss different possible perspectives on the research findings.[31]
Finally, it is recognised in medical sociology that differences in interpretation between researcher and interpreter are not simply differences in translation, but are part of a wider phenomenon, namely that concepts of health and illness are socially constructed. [32-34] In other words, a particular woman in Bangladesh having certain obstetric complications is not a taken-for-granted fact based on medical scientific evidence, but it is a renegotiation of medical knowledge within a cultural and social context. This has an impact on the disease under debate, but also in the way one deals with it in terms of health care provision.[35]
Conclusion
Issues of cross-language data collection should be seen as a challenge and not as an obstacle. [1] This paper presents two different models of practice for working with an interpreter. The 'passive-interpreter' and 'active-interpreter' model each had advantages and disadvantages as discussed. Despite efforts to clarify roles between the researcher and interpreter some areas of tension were hard to overcome. Greater attention should be paid to the effect of an interpreter on the research process. As Temple and Young reminded us: "The translator always makes her mark on the research, whether this is acknowledged or not…".[36] Developments should include more practical advice to enable researchers and interpreters to develop more effective relationships.
In terms of the research findings this study highlights the importance of considering the relationship between researcher, interpreter and respondents and the inclusion in reflexive methodology. The challenges faced when using translation in qualitative research will depend upon the level of depth in analysis and accuracy in translation required. For some studies a less accurate, but perhaps easier to read translation will be sufficient. This should be considered at the outset of study design so that the level of accuracy achieved does not compromise the level of analysis undertaken. The type of analysis planned and study design should drive the level of accuracy in translation rather the level in accuracy limiting the possible analysis.
In conclusion, we would like to recommend that public health researchers working with interpreters and translations pay more attention to: 1) developing effective relationship with interpreter; 2) the effect of the interpreter on the research process; and 3) the accuracy of the translation and level of analysis needed any specific public health research.
Competing interests
The author(s) declare that they have no competing interests.
Author's contributions
EP carried out the data collection and analysis for this study as part of her Ph.D. research. EvT was joint supervisor of this public health research project and participated at all stages of the study. Both authors have written several drafts and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The researchers would like to thank Professor Wendy Graham for her significant role in this research, Tamsin Moushumi Hossain for her work as interpreter and the women and families who participated in the study. Finally, to the two reviewers (Jeanine Blackford and Alexander Bischoff) for their comments and Mary Dixon-Woods for her help with redrafting.
Financial support for this study was received from the University of Aberdeen, Phil Strong Memorial Prize (British Sociological Association's Medical Sociology Group) and the Carnegie Trust for Scotland.
==== Refs
Hunt S Bhopal R Self reports in research with non-English speakers Br Med J 2003 327 352 3 12919965
Bowden A Fox-Rushby JA A systematic and critical review of the process of translation and adaptation of generic health-related quality of life measures in Africa, Asia, Eastern Europe, the Middle East, South America Soc Sci Med 2003 57 1289 1306 12899911 10.1016/S0277-9536(02)00503-8
Fukuhara S Bito S Green J Hsiao A Kurokawa K Translations, adaptation and validation of the SF-36 Health Survey for use in Japan J Clin Epidemiol 1998 51 1037 44 9817121 10.1016/S0895-4356(98)00095-X
Birbili M Translating from one language to another Social Research Update 2000 31
Bassnett S Translation Studies 1991 second edition London: Routledge
Bhopal R Vettini A Hunt S Wiebe S Hanna L Amos A Review of prevalence data in, and evaluation of methods for cross cultural adaptation of, UK surveys on tobacco and alcohol in ethnic minority groups Online First BMJ doi:10.1136/bmj.37963.426308.9A (2 January 2004)
Nunkoosing K The problem with interviews Qual Health Res 2005 15 698 706 15802544 10.1177/1049732304273903
Temple B Edwards R Interpreters/translators and cross-language research: reflexivity and border crossings Int J Qual Meth 2002 1 Article 1
Jentsch B The 'interpreter effect': rendering interpreters visible in cross-cultural research and methodology J Eur Soc Policy 1998 8 275 289
Bradby H Translating culture and language: a research note on multilingual settings Sociol Health Illn 2002 24 842 855 10.1111/1467-9566.00321
Bowling A Research Methods in Health: investigating health and health services 1997 Buckingham: Open University Press
Mason J Qualitative Researching 2002 2 SAGE: London
Platt J Bryman A, Burgess R What can case studies do?, Qualitative Research 1999 London: SAGE 160 179
Teijlingen van E Ireland J Research interviews in midwifery RCM Midwives J 2003 6 260 263
Richards H Emslie C The 'doctor' or the 'girl from the university'? Considering the influence of professional roles on qualitative interviewing Fam Pract 2000 17 71 75 10673494 10.1093/fampra/17.1.71
Jones EL Bulmer M and Warwick D The courtesy bias in South-East Asian surveys Social Research in Developing Countries 1983 London: John Wiley & Sons
Rubin HJ Rubin IS Qualitative Interviewing: the art of hearing data 1995 SAGE: Thousand Oaks
Holliday A Doing and writing qualitative research 1997 Blackwell Science: Oxford
Mitchell JC Bryman AB, Burgess RG Case and situation analysis Qualitative Research 1999 London: SAGE 180 220
Flyvbjerg B Seale C, Giampietri G, Jaber F, Silverman D Five misunderstandings about case-study research Qualitative Research Practice 2004 SAGE: London 420 434
Eisenhardt KM Bryman A, Burgess R Building theories from case study research Qualitative Research 1999 London: SAGE 135 179
Cutcliffe JR Reconsidering reflexivity: introducing the case for intellectual entrepreneurship Qual Health Res 2003 13 136 148 12564268 10.1177/1049732302239416
Fontana A Frey JH Denzin N, Lincoln YS Interviewing: the art of science Handbook of qualitative research 1994 Tousand Oaks: SAGE
Blanchet T Lost innocence, stolen childhood 2001 Dhaka: The University Press Ltd
Bhopal K Researching South Asian women: issues of sameness and difference in the research process J Gend Stud 2001 10 279 286 10.1080/09589230120086485
Dyck I Lynam JM Anderson JM Women talking: treating knowledge through difference in cross-cultural research Womens Stud Int Forum 1995 18 611 626 10.1016/0277-5395(95)00083-6
Herod A Reflections on interviewing foreign elites: praxis, positionality, validity, and the cult of the insider Geoforum 1999 30 313 327 10.1016/S0016-7185(99)00024-X
Oakley A Roberts H Interviewing women: a contradiction in terms Doing Feminist Research 1981 London: Routledge & Kegan Paul
Miller J Glassner B Silverman D The "inside" and the "outside": finding realities in interviews Qualitative research: theory, method and practice 2004 London: SAGE
Twinn S An exploratory study examining the influence of translation on the validity and reliability of qualitative data in nursing research J Adv Nur 1997 26 418 423 10.1046/j.1365-2648.1997.1997026418.x
Temple B Watch your tongue: issues in translation and cross-cultural research Sociol 1997 31 607 618 10.1177/0038038597031003016
Turner BS Regulating Bodies: Essays in Medical Sociology 1992 London: Routledge
Armstrong D Albrecht GL, Fitzpatrick R and Scrimshaw SC Social Theorizing about health and illness Handbook of Social Studies in Health & Medicine 2000 London: Sage
Lupton D Medicine as Culture: Illness Disease and Body in Western Societies 1994 London: sage
DeVries R Benoit C Teijlingen van E Wrede S Eds Birth by Design: Pregnancy, Midwifery Care and Midwifery in North America and Europe 1994 New York: Routledge
Temple B Young A Qualitative research and translation dilemmas Qual Res 2004 4 161 178 10.1177/1468794104044430
|
15985161
|
PMC1200559
|
CC BY
|
2021-01-04 16:28:58
|
no
|
BMC Public Health. 2005 Jun 28; 5:71
|
utf-8
|
BMC Public Health
| 2,005 |
10.1186/1471-2458-5-71
|
oa_comm
|
==== Front
Cancer Cell IntCancer Cell International1475-2867BioMed Central London 1475-2867-5-191597210910.1186/1475-2867-5-19Primary ResearchU94 alters FN1 and ANGPTL4 gene expression and inhibits tumorigenesis of prostate cancer cell line PC3 Ifon Ekwere T [email protected] Alan LY [email protected] Warren [email protected] Kathleen [email protected] Sharon [email protected] Sumitra [email protected] Wai-Yee [email protected] John [email protected] Leonard Jason [email protected] Department of Microbiology and Immunology, Georgetown University Medical Center, 3900 Reservoir Road, NW, Washington, D.C. 20057, USA2 Laboratory of Clinical Genomics, NICHD, National Institutes of Health, 9000 Rockville Pike, Bethesda, MD 20892, USA3 Department of Pediatrics, Georgetown University Medical Center, 3800 Reservoir Road, NW, Washington, D.C. 20057, USA4 Department of Cell Biology, Georgetown University Medical Center, 3800 Reservoir Road, NW, Washington, D.C. 20057,5 Department of Biochemistry & Molecular Biology, Georgetown University Medical Center, 3800 Reservoir Road, NW, Washington, D.C. 20057, USA;2005 22 6 2005 5 19 19 13 5 2004 22 6 2005 Copyright © 2005 Ifon et al; licensee BioMed Central Ltd.2005Ifon et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Insensitivity of advanced-stage prostate cancer to androgen ablation therapy is a serious problem in clinical practice because it is associated with aggressive progression and poor prognosis. Targeted therapeutic drug discovery efforts are thwarted by lack of adequate knowledge of gene(s) associated with prostate tumorigenesis. Therefore there is the need for studies to provide leads to targeted intervention measures. Here we propose that stable expression of U94, a tumor suppressor gene encoded by human herpesvirus 6A (HHV-6A), could alter gene expression and thereby inhibit the tumorigenicity of PC3 cell line. Microarray gene expression profiling on U94 recombinant PC3 cell line could reveal genes that would elucidate prostate cancer biology, and hopefully identify potential therapeutic targets.
Results
We have shown that stable expression of U94 gene in PC3 cell line inhibited its focus formation in culture, and tumorigenesis in nude mice. Moreover gene expression profiling revealed dramatic upregulation of FN 1 (fibronectin, 91 ± 16-fold), and profound downregulation of ANGPTL 4 (angiopoietin-like-4, 20 ± 4-fold) in U94 recombinant PC3 cell line. Quantitative real-time polymerase chain reaction (QRT-PCR) analysis showed that the pattern of expression of FN 1 and ANGPTL 4 mRNA were consistent with the microarray data. Based on previous reports, the findings in this study implicate upregulation of FN 1 and downregulation of ANGPTL 4 in the anti tumor activity of U94. Genes with cancer inhibitory activities that were also upregulated include SERPINE 2 (serine/cysteine protease inhibitor 2, 7 ± 1-fold increase) and ADAMTS 1 (a disintegrin-like and metalloprotease with thrombospondin type 1 motif, 7 ± 2-fold increase). Additionally, SPUVE 23 (serine protease 23) that is pro-tumorigenic was significantly downregulated (10 ± 1-fold).
Conclusion
The dramatic upregulation of FN 1 and downregulation of ANGPTL 4 genes in PC3 cell line stably expressing U94 implicate up-regulation of FN 1 and downregulation of ANGPTL 4 in anti tumor activity of U94. Further studies are necessary to determine functional roles of differentially expressed genes in U94 recombinant PC3 cell line, and hopefully provide leads to potential therapeutic targets in prostate cancer.
==== Body
Background
Prostate cancer is the most common form of malignancy in US males. An estimated 29,900 fatalities out of 230,110 new cases are expected in the year 2004 [1]. Androgen ablation is currently the mainstay in prostate cancer therapy, but its efficiency is marred by the relapse of some advanced-stage prostate cancer cells into an androgen refractory state [2,3]. Advanced-stage prostate cancer progression is usually aggressive and correlates with poor prognosis [2,4,5]. Hence, insensitivity to androgen ablation by advanced-stage prostate cancer invariably constitutes a major problem in clinical therapy. Therefore there is an urgent need for the development of targeted therapeutic strategies in advanced-stage prostate cancer.
Knowledge of the genes that are associated with prostate cancer is important for designing an effective therapeutic strategy. However, present knowledge of the molecular biology of prostate cancer is inadequate to define etiologic genes [5-7]. Consequently, current therapeutic strategies in prostate cancer are inefficient [8-20], and an effective targeted therapy remains elusive. This situation prompted our laboratory to embark on studies to provide alternative leads for the development of efficacious and targeted anti prostate cancer agent(s). In our approach, we investigated the anti tumor activity of U94 protein (U94) in prostate cancer cell line, PC3.
U94 is a 1473 bp gene located in the HD12 fragment of human herpesvirus 6A (HHV-6A), strain U1102 [21]. U94 encodes a 490 amino acid protein that is not found in other herpesviruses [21,22], and U94 is expressed at very low levels [23,24]. Recent reports suggest that U94 is a latency gene, and modulates viral DNA replication [23-26]. Moreover, structural homology of U94 to Rep 78/68 from adeno-associated virus type 2 (AAV-2) [21,27] suggests that there might be functional similarities between these proteins. Strong evidence in support of functional similarities between U94 and Rep 78/68 is the observation that U94 complemented the replication of an AAV-2 mutant that was deficient in Rep 78/68 [28]. Additionally, recent reports show that U94 also inhibits gene transcription [29], which is a biological function of its homologue Rep78/68. However, U94 may affect gene transcription differently than Rep 78/68, because U94 activates human immunodeficiency virus 1 (HIV-1) long terminal repeat (LTR) promoter in fibroblast cell lines [28] and inhibits HIV-1 LTR in T-cell lines [29], whereas Rep 78/68 inhibits HIV-1 LTR promoter in both fibroblast cell lines and T-cell lines [28].
Previous studies demonstrated that U94 suppressed transformation by oncogenes [22,29]. Data from these studies showed that an NIH 3T3 cell line stably expressing U94 gene suppressed transformation by the oncogene H-ras, when compared to the parental NIH 3T3 cell line treated under similar conditions [29]. We were motivated by the findings in previous studies to determine the anti tumor potential of U94 in the human prostate tumor cell line PC3.
In this paper we report that the expression of U94 protein in PC3 cells inhibited foci formation (Figure 2; Table 1), and the tumorigenicity of recombinant PC3 cell line in athymic nude mice (Figure 3). Moreover, gene expression analyses (Figures 4 and 5), and QRT-PCR (Table 2) revealed dramatic upregulation of FN 1 (~91-fold) and profound downregulation of ANGPTL 4 (~20-fold) in 2 separate recombinant PC3 cell lines stably expressing U94. Our study also demonstrated the differential expression pattern of several other genes in the presence of U94. This is the first study to report the inhibitory potential of U94 on the tumorigenicity of advanced-stage prostate cancer cell line PC3.
Figure 2 Inhibition of focus formation by PC3 cell line stably expressing U94 protein. PC3 cell line transfected with plasmid containing U94 DNA was used for studies. The controls were PC3 cell line transfected with vector cassette or parental PC3 cell line. 1 × 106 cells/ 75 cm2 culture flask was grown to confluence, and focus formation was examined 10 days after. Panel A: PC3 cells (negative control); Panel B: PC3 cells transfected with vector cassette (vector control); Panel C: PC3 cells stably expressing U94 protein (test). Control cells formed foci (Panel A and Panel B) consisting of rounded refractive cells piling on top of each other. Notice that the expression of U94 protein (Panel C) inhibited focus formation. (Magnification: 20X on Olympus CK2 microscope, Olympus Optical Co. Ltd. Japan).
Figure 3 Tumorigenicity of U94 recombinant PC3 cell line in nude mice. Stable G418 resistant cell lines were generated by transfection of PC3 cells with either pRc-RSV (vector control) or pRc-U94 (test). Confluent cells (5 × 106 cells/100 μl), in culture medium without antibiotics or serum were inoculated behind the neck into athymic nude mice (Ncr nu/nu), and monitored for tumor production. Tumor size was measured on days 7, 14, 17, 21 and 28 post inoculation. Data from two animal experiments (experiment 1: n = 3 per group; and experiment 2, n = 4 per group) were pooled. The average tumor volume in cubic centimeters was plotted against time in days. The error bars represent standard deviation. Notice the significant reduction in tumor size in animals inoculated with U94 recombinant PC3 cell line. A repeated measures analysis of variance demonstrated a significant difference (P < 0.05) in tumor volume between test and control animals. Additionally, paired Student's t-test showed a significant difference in average tumor size of test animals in comparison to control animals.
Figure 4 Upregulated genes in PC3 cell line stably expressing U94. Two clones of G418 resistant PC3 cell lines transfected with plasmid pBk-U94 (tests 1 and 2) and a clone transfected with plasmid pBK-CMV (reference) were used for cDNA microarray studies. Each experiment was performed in triplicate and the results are mean ± SD. Notice the dramatic upregulation of FN 1 (91 ± 16-fold). A subset of other genes was also upregulated, but genes of interest in this study (> 6-fold change) include: SERPINE 2 (7 ± 1-fold); and ADAMTS 1 (7 ± 2-fold).
Figure 5 Downregulated genes in PC3 cell line stably expressing U94. Data presented here were generated as reported in Figure 4. Notice the pronounced down-regulation of angiogenic gene, ANGPTL4 (20 ± 4-fold). Additional genes that were significantly downregulated, and with a fold change > 6, include: SPUVE 23 (10 ± 1-fold); TGM 2 (8 ± 2-fold).
Table 1 Inhibition of focus formation by prostate cancer cell line PC3 expressing U94 protein.
Cell line No. of foci/75 cm2flask
Parental PC3 cell line 80
Vector transfected PC3 cell line 69
PC3 cell line stably expressing U94 2
Clonal PC3 cell lines, transfected with plasmid containing U94 gene and geneticin (G418) resistant cassette, were used for studies. PC3 cell line transfected with vector cassette or parental PC3 cell line served as controls. G418 resistant clones were sub-cultured and grown to confluence. Focus formation was detected as dense foci of actively growing and refractive cells. The experiment was performed in duplicate, and the average number of foci was reported. Notice that stable expression of U94 drastically inhibited focus formation.
Table 2 Fold changes of differentially expressed genes in PC3 cell line stablyexpressing U94.
PC3/U94 Clones FNI SERPINE2 ADAMTS1 ANGPTL4 SPUVE23
Microarray
1 & 2 91 ± 16 7 ± 1 7 ± 2 -20 ± 4 -10 ± 1
QRT-PCR
1 183 ± 27 3 ± 1 4 ± 1 -76 ± 13 -6 ± 1
2 467 ± 33 7 10 ± 1 -67 ± 5 -9
Differential expression of genes was determined by microarray, and QRT-PCR was used to confirm microarray data. Microarray data were computed from array analyses presented in Figures 4 and 5. For QRT-PCR, a subculture of PC3 cell lines used for microarray was used for studies: 2 clones of U94 transfected PC3 cells and vector transfected PC3 cells. The protocol for total RNA extraction was same as for microarray studies, except that the RNA extracts were treated with DNase 1. QRT-PCR was performed in triplicate, using SYBR® Green I chemistry, on 7900 HTS Sequence Detection System (Applied Biosystems, Foster City, CA) according to manufacturer's instructions. The results are presented as average of 3 experiments ± SD. Notice that Microarray data and QRT-PCR data show similar trends; moreover the trend of QRT-PCR results was reproducible in 2 clones of U94 recombinant PC3 cell lines.
Results
Previously, we have demonstrated that U94 inhibited gene transcription and also transformation by oncogenes [22,29]. Several reports have implicated the malfunction of transcription regulatory factors [30-38] as well as the activities of oncogenes [39-48] as etiologic factors in prostate tumorigenesis. Hence we wanted to determine whether U94 could exert inhibitory activity on the tumorigenesis of PC3 cell line.
Expression and intracellular localization of U94 protein
First, we wanted to determine whether U94 could be expressed in PC3 cell line. We transfected PC3 cell line with plasmid pBKU94, which contained U94 DNA insert and a selectable geneticin (G418)-resistant vector cassette. pBKCMV vector transfected PC3 cells served as control. Immunoblot analyses, using the U94 polyclonal antibody AB679 as depicted in Figure 1, showed that U94 (56 kDa) protein was expressed in the nuclear fraction (lane 3) and not the cytoplasmic fraction (lane 2) of stably transfected PC3 cell line. No immunoreactivity was detected in the nuclear fraction of vector transfected PC3 cell line (lane 1). Figure 1 lane 4 shows the high molecular weight Rainbow marker.
Figure 1 Immunoblot of U94 protein. Nuclear and cytoplasmic protein was extracted from confluent PC3 cell line, PC3 cell line transfected with vector cassette (controls) or PC3 cell line transfected with plasmid containing U94 DNA (test). 100 μg sample protein was loaded per lane and separated by SDS-PAGE through a 10% Tris-glycine gel (Novex; Invitrogen, Gaithersburg, MD). The results showed the 56 kDa U94 protein in the nuclear extract of U94 recombinant cells. There was no immunoreactivity in all cytoplasmic extracts, and nuclear extracts of control cells. U94 protein was detected using the polyclonal anti-U94 antibody, AB679 (Amersham, England). The positions of 46 kDa and 66 kDa markers, and the U94 protein are indicated on the right.
Inhibition of focus formation by U94 protein expression
In order to monitor the effect of U94 on tumor formation, we investigated focus formation by PC3 cell line as an index of a neoplastic phenotype. Focus formation was observed as dense foci of intensive cell growth in culture, consisting of refractive cells that rounded up and piled on top of each other [49]. Three PC3 cell lines were used in this study: U94 transfected, vector cassette transfected, and parental PC3 cell line. For each cell line, 1 × 106 cells/ 60 mm culture dish was seeded and grown to confluence. Focus formation was examined 10 days post confluence. The result of this study (Table 1) showed a drastic reduction in focus formation by PC3 cells expressing U94: the number of foci were reduced ~35-fold and 40-fold in comparison with the control vector transfected and parental PC3 cell lines, respectively. Figure 2 shows large and widespread foci in the culture of control vector transfected and parental PC3 cell lines. The culture of recombinant PC3 cell line expressing U94 protein showed only few foci, grossly reduced in size. Our findings suggest that U94 may exhibit anti tumor activity in vitro.
Expression of U94 inhibits tumorigenicity of PC3 cell line in athymic nude mice
In order to determine whether U94 inhibits the tumorigenicity of PC3 cell line in vivo, we inoculated 5 × 106 PC3 cells (U94 transfected cells as test, or vector cassette transfected cells as control) subcutaneously behind the neck, into athymic nude mice. Animals were examined for tumor formation on days 7, 14, 17, 21, and 28 after inoculation. Our result showed that tumor formation was inhibited in mice that were inoculated with PC3 cell line stably expressing U94 protein (Figure 3). The control animals that were inoculated with PC3 cell line transfected with vector cassette developed tumors, and tumor size increased progressively with time as shown in Figure 3. Statistical analysis, using a repeated measures analysis of variance (ANOVA), demonstrated that tumor volume in test and control animals were significantly different (P < 0.05). A comparison of tumor volume between test and control mice, using Paired Student's t-test to supplement ANOVA, further showed that the average tumor volume of test and control animals were significantly different (P < 0.05) on each day tumor volumes were determined. These findings demonstrate that U94 significantly (P < 0.05) inhibited the tumorigenicity of PC3 cell line in athymic nude mice, and corroborate our data (Figure 2) from focus formation assay.
Microarray gene expression profiling in PC3 cell line stably expressing U94
We performed microarray gene expression profiling on recombinant PC3 cell line stably expressing U94 to determine whether U94 affected expression of genes involved in tumorigenesis. We used two clones of U94 recombinant PC3 cell lines as test samples, and PC3 cell line transfected with plasmid vector as our reference. The effect of U94 on gene expression was analyzed using two-color comparative fluorescence assays on glass slide microarrays containing ~6,000 cancer related genes. Our data demonstrated the differential expression of 78 genes: 31 genes were upregulated (Figure 4) while 47 genes were downregulated (Figure 5). These results show the mean values for two clones of U94 recombinant PC3 cell lines. Notably, the microarray results revealed dramatic upregulation of FN 1 (91 ± 16-fold), and profound downregulation of ANGPTL 4 (20 ± 4-fold) in PC3 cell lines stably expressing U94 protein. Although a majority of the differentially expressed genes showed a 2–3 fold change in expression level in the presence of U94, we decided to consider for further studies only genes with ≥ 6-fold change. The microarray data is deposited in the Gene Expression Omnibus of the NCBI, and is available at the NCBI web site .
Quantitative real-time PCR
We performed QRT-PCR to confirm the microarray data. In one clone of U94 recombinant PC3 cell line, the QRT-PCR data (Table 2) showed that the changes in expression levels of FN 1 and ANGPTL 4 mRNA were 183 ± 27-fold increase and 76 ± 13-fold decrease, respectively. In the second clone of U94 recombinant PC3 cell line, the corresponding changes in expression levels were 467 ± 32-fold increase and 67 ± 5-fold decrease, respectively. The observed differences in the fold changes between the microarray and QRT-PCR data may be due, at least in part, to differences in detection sensitivity of the two techniques, as well as the subtle differences in experimental conditions and physiological conditions in the microenvironment of the cells in culture. Nevertheless, the trend observed from QRT-PCR data was consistent with the trend from microarray data. Additionally, the results from both techniques showed elevation of SERPINE 2 and ADAMTS 1 expressions (Table 2 and Figure 4), and downregulation of SPUVE 23 (Table 2 and Figure 5). TGM-2 (transglutaminase 2) showed 8 ± 2-fold decrease by microarray, but we did not perform QRT-PCR.
Discussion
In the present study, we have demonstrated for the first time that U94 protein inhibited focus formation and tumorigenicity of the prostate cancer cell line PC3. This study is particularly interesting because PC3 cell line is a derivative of advanced-stage prostate cancer metastasis to bone and is insensitive to androgen ablation therapy. Insensitivity to androgen ablation therapy is associated with aggressive progression of the cancer, and ultimately fatal in less than 24 months [2,6]. Therefore the anti tumor activity of U94 in PC3 cell line is novel and interesting, and may have a translational application.
The impetus for our study on anti tumor activity of U94 in PC3 cell line was given by previous findings [22,29] that U94 suppressed transformation by oncogenes. Apparently, U94 shares this functional activity with its homologue Rep 78/68 of AAV-2 [24]. However, the mechanism(s) of transformation suppressor activity is not understood. A previous report showed that U94 lost its activity when translation termination linkers were inserted at codons 25, 125 and 245 of its nucleotide sequence [29]. This finding implicates U94 protein expression in anti tumor activity in recombinant PC3 cell line. Therefore we performed immunoblot analysis and demonstrated that U94 protein (56 kDa) was expressed, and localized to the nucleus (Figure 1, lane 3) in PC3 cell line. Nuclear localization of U94 protein suggests that U94 might exert activity, probably on gene expression, in the nucleus of PC3 cell line. This view is in consonance with previous findings [22,29] that U94 inhibited gene expression. A previous study [29] showed that U94 suppressed the P97 promoter, which controls the expression of the E6 and E7 transforming genes of human papillomavirus 16 (HPV-6). Therefore we suspect that the tumor suppressor activity of U94 in our study was exerted by inhibition of gene expression. It is interesting to note that the expression of U94 protein does not affect the growth pattern of NIH 3T3 cell line [29]. This observation is supported by another report [25] that lymphoid cells stably expressing U94 had the same morphology and growth characteristics as parental cell line. Thus previous findings suggest that U94 is not toxic to cells, and we speculate that the same would be true for the PC3 cell line.
In the current investigation, we examined the effect of stable expression of U94 protein on focus formation, a malignant phenotype, by recombinant PC3 cell line. Focus formation by PC3 cell line stably expressing U94 was inhibited drastically (~30- to 40-fold) in comparison to that of control parental and vector transfected cell lines (Table 1). As shown in Figure 2, widespread and large foci were formed by control PC3 cell lines in contrast to background foci formed by U94 recombinants. It is possible that the few foci observed in the background originated from spontaneous transformation and/ or leakage during clonal selection. Furthermore, our studies demonstrated that the anti cancer activity of U94 was sustainable in vivo as tumor development was significantly (P < 0.05) inhibited in mice that were treated with U94 recombinant PC3 cell line (Figure 3). Previous studies linked prostate malignancy to the activities of oncogenes [39-48]. Therefore the inhibition of focus formation and tumorigenicity in our study supports the hypothesis that U94 probably inhibited oncogenic activities in prostate cancer cell line PC3, and thereby exhibited anti cancer activity. Our microarray data identified FN 1 that was dramatically elevated (~91-fold) (Figure 4) and ANGPTL 4 that was profoundly reduced (~20-fold) (Figure 5) as genes of interest in this study. Up-regulation of FN 1 in the current study was actually unexpected because previous reports [22,29] suggested that U94 inhibited gene expression. In contrast, our findings suggested that U94 actually altered gene expression in PC3 cell line positively or negatively. Although it is not known how U94 mediated gene expression, our data is interesting because ANGPTL 4 is pro-angiogenic [50], and reduced expression could negatively impact tumorigenesis. Additionally, previous reports [51-54] implicated elevated FN 1 and/ or its derivatives in paradigms of tumor inhibition.
Fibronectin (FN) is a major component of extracellular matrix (ECM), where it is assembled as insoluble polymers, and is present in the blood as a soluble dimer [55]. Fibronectin 1 (FN 1) is a homologue of FN, and contains a self-assembly domain, which induces FN 1-FN 1 polymerization [56-58]. Therefore the terms FN 1 and FN are used interchangeably in regard to polymerization in this report.
FN 1-FN 1 interaction is reported [59,60] to induce conformational changes that increase the binding ability of FN 1 to receptor(s). We speculate that the tremendous upregulation of FN 1 transcription in the presence of U94 led to elevated translation and secretion of protein product in PC3 cells. The increased level of FN 1 protein in turn accelerated FN 1-FN 1 polymerization [56-58]. We suspect that polymeric FN 1 binding to PC3 cell surface mitigated malignant signaling [61]. The potential anti-malignancy activity of FN 1 is evident from a recent report [62] that exogenous FN 1 could reverse transformed phenotype. Hence, FN 1 interaction with PC3 cell surface might have contributed, at least in part, to inhibition of focus formation in vitro (Figure 2) and tumorigenesis in vivo (Figure 3).
Strong support for in vivo anti tumor activity of polymeric FN 1 is given by a previous report [53] demonstrating that systemic administration of polymeric FN 1 exhibited anti tumor activity in mice bearing various types of tumors. Further support is provided by a recent report [59] showing that anastellin, a component of FN 1 that is capable of inducing FN 1-FN 1 polymerization, displays anti angiogenic and anti metastastic properties in vivo. Additionally, other workers [52] have demonstrated that peptides of FN 1 exert anti tumor activity in vivo. These findings provide a rational explanation for the observation in this study that elevated FN 1 expression in U94 recombinant PC3 cell line is associated with anti tumor activity. Taken together, previous reports [52,53,59] lend credence to our view that binding of polymeric FN 1 to PC3 cells surface induces the inhibition of focus formation in culture and inhibition of tumorigenicity in mice. However, in addition to FN 1, our results also implicated ANGPTL 4 in the anti tumor activity of U94 in PC3 cell line.
ANGPTL 4 was shown in a chicken chorioallantoic membrane assay to induce a strong pro-angiogenic response, independent of VEGF gene [50]. Since angiogenesis is implicated in vascular development, and neovascularization is the hallmark of tumor progression [63,64], an inhibitor of angiogenesis could greatly impact tumor therapy. In the current study we have demonstrated that ANGPTL 4 was profoundly inhibited (downregulated about 20-fold) in U94 recombinant PC3 cell line. It is therefore expected that downregulation of ANGPTL 4 would exert a negative effect on vascular development, and thereby inhibit PC3 cell line tumorigenicity in vivo. Although it is not clear how U94 mediates the expression of ANGPTL 4, recent reports [65-67] show that angiogenesis is regulated by ECM signals. Interestingly, other reports [53,59] suggest that the anti-angiogenic property of polymeric FN 1 is mediated by induction of ECM signals. Therefore, it appears that there may be a casual or causal relationship between anti tumor activity of polymeric FN 1 and the inhibition of ANGPTL 4 in U94 recombinant PC3 cell line. Since ANGPTL 4 supports vascular development [50], we speculate that ANGPTL 4 did not mediate the inhibition of focus formation by PC3 cell line in this study.
In addition to FN 1 and ANGPTL 4, we also chose for further studies a subset of other genes that expressed differentially > 6-fold. Genes in this category included SERPINE 2 (elevated ~7-fold), ADAMTS 1 (upregulated ~7-fold) and SPUVE 23 (downregulated ~10-fold). SERPINE 2 encodes a serine proteinase inhibitor, and was recently implicated in anti cancer activity [68]. ADAMTS 1 is an active metalloproteinase associated with ECM [69]. It is essential for normal growth [70], but also displays anti-angiogenic activity [71]. In consonance with previous reports [68,71], data from the current study suggest that SERPINE 2 and ADAMST 1 probably exerted anti tumor activity. Previous studies [72,73] showed that the expression of SPUVE 23, a serine protease, is associated with increased malignant potential. Therefore we propose that downregulation of SPUVE 23 in U94 recombinant PC3 cell line is tantamount to anti tumor activity.
In conclusion, the findings in this study have suggested that U94 exhibits anti tumor potential in PC3 cell line. The dramatic elevation of FN 1 expression and reduction of ANGPTL 4 expression in U94 recombinant PC3 cell line can be interpreted as evidence of the mechanism of U94 anti tumor activity. Therefore data from our study seem to support the anti tumor hypothesis of FN 1 previously reported by other workers [51,52,54,55,74]. Moreover, this report identifies ANGPTL 4 and SPUVE 23 as potential therapeutic targets in prostate tumorigenesis. Hopefully, further studies on the microarray data reported herein might elucidate the complex genetic alterations that underlie advanced-stage prostate tumorigenesis, and thereby provide leads for defining targeted therapeutic strategies for advanced-stage prostate cancer.
Materials and methods
Cells and transfection
PC3 cell line was purchased from American Type Culture Collection (ATCC, Manassas, VA, USA) and plasmid U94 DNA was prepared as previously described [22,29]). All cells were cultured in HAM's F12 medium (Cell gro/Mediatech, VA, USA) supplemented with 2 mM glutamine, 100 U of penicillin-streptomycin per ml (Invitrogen, Gaithersburg, MD, USA), and 10% Fetal Bovine Serum (FBS, HyClone, Logan, UT, USA), at 37°C and 5% CO2. Plasmid U94 DNA was cloned into the HindIII site of pRc-RSV vector (Invitrogen, Gaithersburg, MD, USA) or HindIII/ BamHI site of pBK-CMV (Stratagene, Cedar Creek, TX, USA) vector. Both pRc-RSV and pBK-CMV vectors contain a geneticin (G418; Mediatech Inc, Herndon, VA, USA) selectable marker. U94 DNA sequence in the constructs was confirmed by DNA sequencing. The pBK-U94 construct was specifically used in experiments that necessitated strong expression of U94 protein e.g. immunoblotting, because previous findings showed that U94 mRNA and protein were expressed at very low levels [23,24,75]. All plasmid DNA were prepared by double-banded cesium chloride gradient ultracentrifugation. U94 construct (pRc-U94 or pBK-U94), or plasmid vector cassette (pRc-RSV or pBK-CMV) was used for transfection of PC3 cells. For transfections, 5.5 × 105 PC3 cells were plated in 60 mm culture dish, and grown over-night (50%-70% confluence). Transfection was performed by the calcium phosphate-based ProFection Mammalian Transfection method (Promega, Madison, WI, USA) in accordance with manufacturer's protocol. Stably transfected PC3 cells were selected with G418 (600 μg/ml), and expanded to establish U94 recombinant PC3 cell line. Clonal selection was performed on G418 resistant healthy colonies using a clonal cylinder. In order to minimise culture driven genetic changes [76], transfected cells were discarded after 8 passages.
Protein extraction, and immunoblot analysis
Nuclear fraction from cellular extract was prepared as described previously [77]. Confluent PC3 cell line (107-1.5 × 107 cells), stably expressing U94 and resistant to G418 antibiotic (Cellgro, Herndon, VA, USA) was freshly prepared by trypsinization, washed with DMEM (Cellgro/Mediatech, VA, USA), suspended in 50 ml DMEM (in sterile 50 ml centrifuge tube), and incubated at 37°C/ 5% CO2 for 2 hours. Cells were centrifuged at 200 × g for 5 minutes, and resuspended in 0.5 ml phosphate buffered saline (PBS, Biofluids, Rockville, MD, USA) in microfuge tube. The cell pellet from another round of centrifugation was resuspended in 400 μl of Buffer A (10 mM HEPES; 10 mM KCl; 0.1 mM EDTA; 0.1 mM EGTA; 1 mM DTT; o.5 mM PMSF; 1% v/v aprotinin). After incubation at 40°C for 15 minutes, cells were lysed by adding 0.6% Nonidet P-40, mixed by inverting tube 10 times, and the nuclei was obtained by centrifugation at 200 × g for 5 minutes. The supernatant was used as the cytoplasmic fraction. The nuclei were resuspended gently in ice-cold 100 μl of Buffer B (20 mM HEPES; 0.4 M NaCl; 1 mM EDTA; 1 mM DGTA; 1 mM DDT; 1 mM PMSF; 1% v/v aprotinin; 10% glycerol), using a wide bore pipette. The nuclei lysate was incubated in a rotary shaker for 30 minutes at 4°C, and then centrifuged at 12000 × g for 10 minutes. To aliquots of the clear supernatant in microfuge tubes, 0.025 mg/ml leupeptin was added before storage at -80°C. The control cell lines: vector transfected and resistant to G418, and parental PC3, were similarly treated.
Protein determination in the extracts was performed using BCA protein assay kit (Pierce, Rockford, IL, USA). Following sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE), resolved proteins were electroblotted onto polyvinylidene difluoride (PVDF) membrane. The membrane was blocked in 5% non-fat milk solution on a rocker for 30 min, and rinsed quickly in Tris/sodium chloride/EDTA/Tween 20 (TNET, 0.2 M Tris pH 7.5; 0.05 M EDTA; 1.0 M NaCl; 1% Tween 20) wash solution. Then the membrane was washed twice in TNET on a rocker for 10 minutes, before it was probed with U94 primary antibody AB679 (Rabbit antiserum, 1:1000 dilution in TNET; Amersham) on a rocker for 1 hour. This was followed by three 10-minutes washes in TNET before anti rabbit-HRP-tagged secondary antibody (1:10,000 dilution in TNET; Amersham) was added and incubated for 1 hour. Three 10-minutes washes in TNET were performed on a rocker, before the detection of immunoreactive proteins using ECL (Amersham, England) reagent.
Focus formation assay
Two clones of U94 transfected cell lines were used for studies. Vector transfected and parental PC3 cell lines were used as controls. Cells were plated at 1 × 106 cells/ 75 cm2 culture flask in duplicate and grown to confluence. Focus formation was visually detected by observing dense foci of intensive cell growth, consisting of refractive cells that rounded up and piled on top of each other [49]. The number of foci in each flask was noted on the 10th day after the cells were confluent. Average counts of foci in duplicate flasks were determined for each cell type.
Tumorigenicity assay
The tumorigenicity of PC3 cell line stably expressing U94 was tested in athymic Ncr nu/nu mice. The control animals were treated with vector transfected PC3 cell line. In all cases 5 × 106 cells were inoculated subcutaneously behind the neck, into athymic nude mice as earlier described [78]. The mice were monitored every 2 or 3 days for the appearance of tumors, and tumor volume was measured on days 7, 14, 17, 21 and 28 post inoculation. Tumor sizes were evaluated by tumor volume (length × width × height, in cm). In all cases confluent cells were used. There were two animal experiments. In the first (n = 3 per group), data entries were made on days 0, 7, 17, 21, and 28, while in the second (n = 4 per group) entries were made on days 0, 7, 14, and 17 post inoculation. Data were pooled from the two experiments and reported. The Animal Welfare Committee, Georgetown University, approved the protocol for the animal studies.
Total RNA extraction and purification
Two clones of PC3 cell line stably expressing U94 (test samples 1 and 2) and PC3 cell line transfected with vector cassette (reference sample) were used for studies. Cells were grown to confluence and total RNA was extracted using Trizol reagent (Invitrogen, Gaithersburg, MD) following manufacturer's instructions. The RNA was cleaned up using the RNeasy® mini columns (Qiagen, Valencia, CA, USA) following manufacturers' instructions. RNA content and quality was initially determined by OD260and OD280 measurements. RNA samples showing an OD260/280 ratio higher than 1.8 was used for microarray hybridization and QRT-PCR. RNA content and integrity was reassayed in duplicate using the Bioanalyzer 2100 (Agilent, Germantown, MD, USA).
Microarray analysis
Gene expression analysis was performed using a 6 k human cDNA microarray fabricated with ~6000 cancer related genes. Fifty micrograms of total RNA from test and reference samples were separately reversed transcribed using the MicroMax™ Direct cDNA Labeling Kit (Perkin Elmer Life Sciences, Boston, MA, USA) into Cy3 and Cy5 labeled cDNA targets. Cy3-labeled targets prepared from test samples 1 and 2 were hybridized with Cy5-labeled cDNA targets from reference sample onto separate microarrays. A dye-swapping experiment was performed with cDNA targets from test sample 1 labeled with Cy5 and cDNA targets from reference sample labeled with Cy3 in order to eliminate any experimental bias owing to the differences in incorporation efficiency of the 2 fluorescent dyes. Thus the microarray hybridization was carried out in triplicate. Labeled test and reference cDNAs were pooled and purified using Microcon YM-100 filter units (Millipore Corp., Bedford, MA, USA), and co-hybridized onto the microarray at 65°C for 14 hours in the dark. Each microarray was washed at room temperature in 45 ml of the respective wash buffer with the following composition and for the specified duration: 1x SSC/ 0.2% SDS for 5 minutes; 0.5x SSC/ 0.01% SDS for 15 minutes; 0.06x SSC/ 0.01% SDS for 15 minutes; 0.06x SSC for 15 minutes. The washed microarrays were spun at 1000 rpm for 4 minutes before scanning at 5 micron resolution using the ScanArray 5000XL (Packard Biosciences, Billerica, MA, USA). Signals generated from Cy3 and Cy5 channels on each microarray were background subtracted and normalized to the total signals of all spots by LOWESS method, and analyzed by ScanArray Express software (Perkin Elmer Life Sciences, Boston, MA, USA). Data were represented as a fold change of fluorescence intensity of a gene from test sample versus reference sample. A fluorescence intensity ratio of U94/vector transfected targets ≥ 2 represented up-regulation of gene; while ≤ 0.5 represented down-regulation. Average values and standard deviation for triplicate experiments were determined. Genes were considered to be differentially expressed only if they displayed the same trend of change in expression in each of the triplicate experiments. Gene annotation information was based on the human Unigene Cluster Build #161 (5th of June 2003; NCBI)
Quantitative real-time polymerase chain reaction (QRT-PCR)
To verify the expression pattern of the differentially expressing genes identified from microarray experiments, QRT-PCR was performed as described previously (79). Equal amounts of total RNA from test and reference cell lines were treated with DNase 1 (Invitrogen, Gaithersburg, MD, USA), and reverse transcribed using random hexamers and SuperScript II (Invitrogen, Gaithersburg, MD, USA) to prepare the first strand cDNA samples for QRT-PCR analyses. The RT product was diluted 5-fold, and 1 μl is equivalent to 1x concentration. Gene specific primers (Table 3) were designed by Primer Express Version 2.0 (Applied Biosystems, Foster City, CA, USA) according to the sequence information provided for the cDNAs on the microarray. The primers were BLASTed against the non-redundant and EST mouse sets from NCBI to confirm specificity. QRT-PCR was performed in triplicate using SYBR® Green I chemistry on 7900 HTS Sequence Detection System (Applied Biosystems, Foster City, CA, USA) according to manufacturer's instructions. The temperature cycle for QRT-PCR was set up as following: 50°C for 2 minutes; 95°C for 10 minutes; 95°C 15 seconds and 60°C for 1 minute for 40 cycles. A final dissociation cycle running at 95°C for 15 seconds, 60°C for 15 seconds and 95°C for 15 seconds was set up for monitoring the specificity of amplification. The relative standard curve method was used for quantifying gene expression level, in which the CT values of a series of fixed amounts of test sample (or reference sample) cDNAs (0.01x , 0.1x and 1x as described above) were plotted against these amounts of cDNAs. The CT value for a gene at 0.1x concentration in the reference sample (or test sample) was fitted onto the standard curve to obtain the respective expression level. A smaller CT value indicates a higher expression level, and vice versa. Genes showing CT values ≥ 40 were considered to be non-expressing. The final gene expression data were reported after normalising to that of 18S RNA.
Table 3 Sequences of Forward and Reverse primers for gene amplification in QRT-PCR.
Genes Forward Primer Reverse primer
FN 1 5-GTGTGACCCTCATGAGGCAAC-3 5-CTGGCCTCCAAAGCATGTG-3
SERPINE 2 5-CACATCAGCACCAAGACCATAGAC-3 5-TGCCAAGAACTTTCAGCGG-3
ADAMST 1 5-CCAGCGTATCTTGCCAGTAACC-3 5-TTTGCAACTGGCAGTTTACTCTG-3
ANGPTL 4 5-CCACTTGGGACCAGGATCAC-3 5-CGGAAGTACTGGCCGTTGAG-3
SPUVE 23 5-CCCAGTCTACCCTCAATTTAGCC-3 5-GCAGTGGAGTTCCCTTATGACAC-3
Gene specific primers for QRT-PCR were designed using Primer Express Version 2.0 (Applied Biosystems, Foster City, CA) according to the sequence information provided for the cDNAs on the microarray.
Statistical analysis
A repeated measures analysis of variance (ANOVA), supplemented by Paired Student's t-test, was used to evaluate the differences in tumor volume between U94 treated and control vector treated mice. A value of p < 0.05 was considered statistically significant. SAS software (v8.2, SAS Institute, Cary, NC, USA) was used for ANOVA. Experimental data, where applicable, are represented as mean ± SD.
Authors' contributions
ETI performed Cell culture, molecular biology studies, immunoassays, participated in statistical analysis, and drafted the manuscript; ALYP carried out the microarray hybridization and data analysis; WJ performed the QRT-PCR analysis; SM carried out the tumorigenicity studies in mice; WYC performed RNA extraction, data analysis and participated in coordination of the studies; KC and SZ assisted in some of the molecular biology studies, and JC participated in coordination of the studies; LJR conceived of the study, and coordinated the studies. All authors read and approved the final manuscript.
Competing interests
The author(s) declare that they have no competing interests.
Acknowledgements
ETI was supported by a Minority Supplement (CMBB/NCI/NIH) to Public Health Service (PHS) grant CA 78120 from the National Institutes of Health (NIH; Bethesda, MD, USA). This work was supported in part by PHS/NIH grant CA 78120, and a Contract from the National Foundation for Cancer Research (Bethesda, MD, USA) awarded to LJR. Assistance with the tumorigenicity assays was provided by the Lombardi Cancer Research Center Animal Care Facility. Special thanks to: Dr. Yan A. Su, Department of Pathology, Loyola University Medical Center, Maywood, IL, USA for fabricating glass slide microarrays for NICHD/NIH; and Michael Sheridan, Sc. D., Consulting Epidemiologist, Inova Health System & Director, Epidemiology & Biostatistics, Department of Medicine, Inova Fairfax Hospital, VA, USA, for performing the ANOVA.
==== Refs
American Cancer Society Cancer Statistics Presentation
Ficazzola MA Taneja SS Prospects for gene therapy in human prostate cancer Molecular Medicine Today 1998 4 494 504 9857369 10.1016/S1357-4310(98)01334-3
Buttyan R Shabsigh A Perlman H Colombel M Regulation of Apoptosis in the Prostate Gland by Androgenic Steroids Trends Endocrinol Metab 1999 10 47 54 10322394 10.1016/S1043-2760(98)00104-0
Chakravarti A Zhai GG Molecular and genetic prognostic factors of prostate cancer World J Urol 2003 21 265 274 12910365 10.1007/s00345-003-0362-z
Karan D Lin M Johansson SL Batra SK Current Status of the Molecular Genetics of Human Prostatic Adenocarcinomas Int J Cancer 2003 103 285 293 12471610 10.1002/ijc.10813
Abate-Shen C Shen MM Molecular genetics of prostate cancer Genes and Dev 2000 14 2410 2434 11018010 10.1101/gad.819500
Kumazawa T Tsuchiya N Wang L Sato K Kamoto T Ogawa O Nakamura A Kato T Habuchi T Microsatellite polymorphism of steroid hormone synthesis gene CYP11A1 is associated with advanced prostate cancer Int J Cancer 2004 110 140 144 15054879 10.1002/ijc.20070
Flaherty KT Malkowicz SB Vaughn DJ Phase I study of weekly liposome-encapsulated doxorubicin in patients with advanced, androgen-independent prostate cancer Am J Clin Oncol 2004 27 136 139 15057152 10.1097/01.coc.0000054888.02055.4E
Joshi B Li L Taffe BG Zhu Z Wahl S Tian H Ben-Josef E Taylor JD Porter AT Tang DG Apoptosis Induction by a Novel Anti-Prostate Cancer Compound, BMD188 (a Fatty Acid-containing Hydroxamic Acid), Requires the Mitochondrial Respiratory Chain Cancer Res 1999 59 4343 4355 10485482
Goodin S Rao KV DiPaola RS State-of-the-Art Treatment of Metastatic Hormone-Refractory Prostate Cancer Oncologist 2002 7 360 370 12185298 10.1634/theoncologist.7-4-360
Gulley J Dahut W Novel clinical trials in androgen-independent prostate cancer Clin Prostate Cancer 2002 1 51 57 15046713
Nimmanapalli R Perkins CL Orlando M O'Bryan E Nguyen D Bhalla KN Pretreatment with Paclitaxel Enhances Apo-2 Ligand/Tumor Necrosis Factor- related Apoptosis-inducing Ligand-induced Apoptosis of Prostate Cancer Cells by Inducing Death Receptors 4 and 5 Protein Levels Cancer Res 2001 61 759 763 11212279
Chay CH Cooper CC Hellerstedt BA Pienta KJ Antimetastatic drugs in prostate cancer Clin Prostate Cancer 2002 1 14 19 15046708
Allay JA Steiner MS Zhang Y Reed CP Cockroft J Lu Y Adenovirus p16 gene therapy for prostate cancer World J Urol 2000 18 111 120 10854145 10.1007/s003450050182
Arlen PM Figg WD Gulley J Cox MC Linehan WM Dahut W National Cancer Institute intramural approach to advanced prostate cancer Clin Prostate Cancer 2002 1 153 162 15046690
Raghow S Hooshdaran MZ Katiyar S Steiner MS Toremifene prevents prostate cancer in the transgenic adenocarcinoma of mouse prostate model Cancer Res 2002 62 1370 1376 11888907
Steiner MS Wang Y Zhang Y Zhang X Lu Y p16/MTS1/INK4A suppresses prostate cancer by both pRb dependent and independent pathways Oncogene 2000 19 1297 1306 10713671 10.1038/sj.onc.1203428
Sternberg C Overview of international collaborative group prostate cancer trials Crit Rev Oncol Hematol 2002 43 153 158 12191737
Collis SJ Khater K DeWeese TL Novel therapeutic strategies in prostate cancer management using gene therapy in combination with radiation therapy World J Urol 2003 21 275 289 12920560 10.1007/s00345-003-0363-y
Timme TL Satoh T Tahir SA Wang H Teh BS Butler EB Miles BJ Amato RJ Kadmon D Thompson TC Therapeutic targets for metastatic prostate cancer Curr Drug Targets 2003 4 251 261 12643475 10.2174/1389450033491127
Thompson BJ Efstathiou S Honess RW Acquisition of the human adeno-associated virus type-2 rep gene by human herpesvirus type-6 Nature 1991 351 78 80 1851252 10.1038/351078a0
Araujo JC Doniger J Kashanchi F Hermonat PL Thompson J Rosenthal LJ Human Herpesvirus 6A suppresses both transformations by H-ras and Human Immunodeficiency Virus type 1 promoters J Virol 1995 69 4933 4940 7609062
Mori Y Dhepakson P Shimamoto T Ueda K Gomi Y Tani H Matsuura Y Yamanishi K Expression of human herpesvirus 6B rep within infected cells and binding of its gene product to the TATA-Binding protein in vitro and in vivo J Virol 2000 74 6096 6104 10846093 10.1128/JVI.74.13.6096-6104.2000
Dhepakson P Mori Y Jiang YB Huang HL Akkapaiboon P Okuno T Yamanishi K Human herpesvirus-6 rep/U94 gene product has single-stranded DNA-binding activity J Gen Virol 2002 83 847 854 11907335
Rotola A Ravaioli T Gonelli A Dewhurst S Cassai E Di Luca D U94 of human herpesvirus 6 is expressed in latently infected peripheral blood mononuclear cells and blocks viral gene expression in transformed lymphocytes in culture Proc Natl Acad Sci USA 1998 95 13911 13916 9811900 10.1073/pnas.95.23.13911
Turner S DiLuca D Gompels UA Characterization of a human herpesvirus 6 variant A 'amplicon' and replication modulation by U94-Rep 'latency gene' Journal of Virological Methods 2002 105 331 341 12270665 10.1016/S0166-0934(02)00130-1
Gompels UA Nicholas J Lawrence G Jones M Thomson BJ Martin ME Efstathiou S Craxton M Macaulay HA The DNA sequence of human herpesvirus-6: structure, coding content, and genome evolution Virology 1995 209 29 51 7747482 10.1006/viro.1995.1228
Thomson BJ Weindler FW Gray D Schwaab V Heilbronn R Human herpesvirus 6 (HHV-6) is a helper virus for adeno-associated virus type 2 (AAV-2) and the AAV-2 rep gene homologue in HHV-6 can mediate AAV-2 DNA replication and regulate gene expression Virology 1994 204 304 311 8091661 10.1006/viro.1994.1535
Araujo JC Doniger J Stöppler H Sadaie MR Rosenthal LJ Cell lines containing and expressing the human herpesvirus 6A ts gene are protected from both H-ras and BPV-1 transformation Oncogene 1997 14 937 943 9050993 10.1038/sj.onc.1200899
Kibel AS Faith DA Bova GS Isaacs WB Loss of heterozygosity at 12P12-13 in primary and metastatic prostate adenocarcinoma J Urol 2000 164 192 196 10840458 10.1097/00005392-200007000-00059
Guo Y Sklar GN Borkowski A Kyprianou N Loss of the cyclin-dependent kinase inhibitor p27(Kip1) protein in human prostate cancer correlates with tumor grade Clin Cancer Res 1997 3 2269 2274 9815624
Tsihlias J Kapusta LR DeBoer G Morava-Protzner I Zbieranowski I Bhattacharya N Catzavelos GC Klotz LH Slingerland JM Loss of cyclin-dependent kinase inhibitor p27Kip1 is a novel prognostic factor in localized human prostate adenocarcinoma Cancer Res 1998 58 542 548 9458103
Yang RM Naitoh J Murphy M Wang HJ Phillipson J deKernion JB Loda M Reiter RE Low p27 expression predicts poor disease-free survival in patients with prostate cancer J Urol 1998 159 941 945 9474188 10.1097/00005392-199803000-00088
Park MS Rosai J Nguyen HT Capodieci P Cordon-Cardo C Koff A p27 and Rb are on overlapping pathways suppressing tumorigenesis in mice Proc Natl Acad Sci USA 1999 96 6382 6387 10339596 10.1073/pnas.96.11.6382
Chi SG deVere White RW Muenzer JT Gumerlock PH Frequent alteration of CDKN2 (p16(INK4A)/MTS1) expression in human primary prostate carcinomas Clin Cancer Res 1997 3 1889 1897 9815578
Jarrard D Modder J Fadden P Fu V Sebree L Heisey D Schwarze S Friedl A Alterations in the p16/pRb cell cycle checkpoint occur commonly in primary and metastatic human prostate cancer Cancer Lett 2002 185 191 12169393 10.1016/S0304-3835(02)00282-3
Jarrard DF Sarkar S Shi Y Yeager TR Magrane G Kinoshita H Nassif N Meisner L Newton MA Waldman FM Reznikoff CA p16/pRb pathway alterations are required for bypassing senescence in human prostate epithelial cells Cancer Res 1999 59 2957 2964 10383161
Alexander K Hinds PW Requirement for p27(KIP1) in retinoblastoma protein-mediated senescence Mol Cell Biol 2001 21 3616 3631 11340156 10.1128/MCB.21.11.3616-3631.2001
Colombel M Symmans F Gil S O'Toole KM Chopin D Benson M Olsson CA Korsmeyer S Buttyan R Detection of the apoptosis-suppressing oncoprotein Bcl-2 in hormone-refractory human prostate cancers Am J Pathol 1993 143 390 400 7688182
Apakama I Robinson MC Walter NM Charlton RG Royds JA Fuller CE Neal DE Hamdy FC Bcl-2 overexpression combined with p53 protein accumulation correlates with hormone-refractory prostate cancer Brit J Cancer 1996 74 1258 1262 8883414
Furuya Y Krajewski S Epstein JI Reed JC Isaacs JT Expression of Bcl-2 and the progression of human and rodent prostatic cancers Clin Cancer Res 1996 2 389 398 9816182
McDonnell TJ Navone NM Troncoso P Pisters LL Conti C von Eschenbach AC Brisbay S Logothetis CJ Expression of bcl-2 oncoprotein and p53 protein accumulaion in bone marrow metastases of androgen independent prostate cancer J Urol 1997 157 569 574 8996359 10.1097/00005392-199702000-00042
Chaudhary KS Abel PD Lalani EN Role of the Bcl-2 gene family in prostate cancer progression and its implications for therapeutic intervention Environ Health Perspect 1999 107 49 57 10229706
Buttyan R Sawczuk IS Benson MC Siegal JD Olsson CA Enhanced expression of the c-myc protooncogene in high-grade human prostate cancers Prostate 1987 11 327 337 2446300
Phillips MEA Ferro MA Smith PJB Davies P Intranuclear androgen receptor deployment and protooncogene expression in human diseased prostate Urol Int 1987 42 115 119 2441504
Nag A Smith RG Amplification, rearrangement, and elevated expression of c-myc in the human prostatic carcinoma cell line LNCaP Prostate 1989 15 115 122 2678039
Peehl DM Oncogenes in Prostate Cancer Cancer 1993 71 1159 1164 8428339
Dhanasekaran SM Barrette TR Ghosh D Shah R Varambally S Kurachi K Pienta KJ Rubin MA Chinnaiyan AM Delineation of prognostic biomarkers in prostate cancer Nature 2001 412 822 826 11518967 10.1038/35090585
Muralidhar S Doniger J Mendelson E Araujo JC Kashanchi F Azumi N Brady JN Rosenthal LJ Human cytomegalovirus mtrII oncoprotein binds to p53 and down-regulates p53-activated transcription J Virol 1996 70 8691 8700 8970996
Le-Jan S Amy C Cazes A Monnot C Lamande N Favier J Philippe J Siboney M Gasc JM Corvol P Germain S Angiopoietin-like 4 is a proangiogenic factor produced during ischemia and in conventional renal cell carcinoma Am J Pathol 2003 162 1521 1528 12707035
Yi M Ruoslahti E A fibronectin fragment inhibits tumor growth, angiogenesis, and metastasis Proc Natl Acad Sci USA 2001 98 620 624 11209058 10.1073/pnas.98.2.620
Zhang GM Yang Y Huang B Xiao H Li D Feng ZH Experimental study on therapeutic effect of in vivo expression of Cell I-Hep II recombinant polypeptide of fibronectin on murine H22 hepatocellular carcinoma World J Gastroenterol 2003 9 1940 1945 12970880
Yi M Sakai T Fassler R Ruoslahti E Antiangiogenic proteins require plasma fibronectin or vitronectin for in vivo activity Proc Natl Acad Sci USA 2003 100 11435 11438 13679585 10.1073/pnas.1635112100
Pasqualini R Bourdoulous S Koivunen E Woods VL JrRuoslahti EA Polymeric form of fibronectin has antimetastatic effects against multiple tumor types Nat Med 1996 2 1197 1203 8898745 10.1038/nm1196-1197
Hynes RO The dynamic dialogue between cells and matrices: implications of fibronectin's elasticity Proc Natl Acad Sci USA 1999 96 2588 2590 10077553 10.1073/pnas.96.6.2588
Schwarzbauer JE Identification of the fibronectin sequences required for assembly of a fibrillar matrix J Cell Biol 1991 113 1463 1473 2045422 10.1083/jcb.113.6.1463
Hocking DC Sottile J McKeown-Longo PJ Fibronectin's III-1 module contains a conformation-dependent binding site for the amino-terminal region of fibronectin J Biol Chem 1994 269 19183 19191 8034677
Aguirre KM McCormick RJ Schqarzbauer JE Fibronectin self-association is mediated by complementary sites within the amino-terminal one-third of the molecule J Biol Chem 269 27863 27868 7961716
Briknarovà K Åkerman ME Hoyt DW Ruoslahti E Ely KR Anastellin, an FN 3 Fragment with Fibronectin Polymerization Acitivity, Resembles Amyloid Fibril Precursors J Mol Biol 2003 332 205 215 12946358 10.1016/S0022-2836(03)00890-8
Pickford AR Smith SP Staunton D Boyd J Campbell ID The hairpin structure of the 6F11F22F2 fragment from human fibronectin enhances gelatin binding The EMBO Journal 2001 20 1519 1529 11285216 10.1093/emboj/20.7.1519
Clark EA Brugge JS Integrins and signal transduction pathways: the road taken Science 1995 268 233 239 7716514
Brenner KA Corbett SA Schwarzbauer JE Regulation of fibronectin matrix assembly by activated Ras in transformed cells Oncogene 2000 19 3156 3163 10918570 10.1038/sj.onc.1203626
Folkman J Angiogenesis in cancer, vascular, rheumatoid and other disease Nat Med 1995 1 27 31 7584949 10.1038/nm0195-27
Risau W Mechanism of angiogenesis Nature 1997 386 671 674 9109485 10.1038/386671a0
Varner JA The role of vascular cell integrins alpha v beta 3 and alpha v beta 5 in angiogenesis EXS 1997 79 361 90 9002227
Ruoslahti e The RGD story: a personal account Matrix Biology 2003 22 459 465 14667838 10.1016/S0945-053X(03)00083-0
Hynes RO A reevaluation of integrins as regulators of angiogenesis Nat Med 2002 8 918 921 12205444 10.1038/nm0902-918
Vitale M Matola TD Rossi G Laezza C Fenzi G Bifulco M Prenyltransferase inhibitors induce apoptosis in proliferating thyroid cells through a p53-Independent, CrmA-Sensitive, and caspase-3-like protease-dependent mechanism Endocrinology 1999 140 698 704 9927296 10.1210/en.140.2.698
Kuno K Terashima Y Matsushima K ADAMTS-1 is an active metalloproteinase associated with the extracellular matrix J Biol Chem 1999 274 18821 18826 10373500 10.1074/jbc.274.26.18821
Shindo T Kurihara H Kuno K Yokoyama H Wada T Kurihara Y Imai T Wang Y Ogata M Nishimatsu H Moriyama N Oh-hashi Y Morita H Ishikawa T Nagai R Yazaki Y Matsushima K ADAMTS-1: a metalloproteinase-disintegrin essential for normal growth, fertility, and organ morphology and function J Clin Invest 2000 105 1345 1352 10811842
Vazquez F Hastings G Ortega MA Lane TF Oikemus S Lombardo M Iruela-Arispe ML METH-1, a human ortholog of ADAMTS-1, and METH-2 are members of a new family of proteins with angio-inhibitory activity J Biol Chem 1999 274 23349 23357 10438512 10.1074/jbc.274.33.23349
Pineiro-Sanchez ML Goldstein LA Dodt J Howard L Yeh Y Chen W Identification of the 170-kDa Melanoma membrane-bound gelatinase (seprase) as a serine integral membrane protease J Biol Chem 1997 272 7595 7601 9065413 10.1074/jbc.272.12.7595
Vacca A Ria R Presta M Ribatti D Iurlaro M Merchionne F Tanghetti E Dammacco F alpha (v) beta (3) integrin engagement modulates cell adhesion, proliferation, and protease secretion in human lymphoid tumor cells Exp Hematol 2001 29 993 1003 11495705 10.1016/S0301-472X(01)00674-9
Abeysinghe HR Cao Q Xu J Pollock S Veyberman Y Guckert NL Keng P Wang N THY1 expression is associated with tumor suppression of human ovarian cancer Cancer Genetics and Cytogenetics 2002 143 125 132 12781446 10.1016/S0165-4608(02)00855-5
Rapp JC Krug LT Inoue N Dambaugh TR Pellett PE U94 the human herpesvirus 6 homolog of the Parvovirus nonstructural gene, is highly conserved among isolates and is expressed at low mRNA levels as a spliced transcript Virology 2000 268 504 516 10704358 10.1006/viro.1999.0163
Beheshti B Park PC Sweet JM Trachtenberg J Jewett MA Squire JA Evidence of chromosomal instability in prostate cancer determined by spectral karyotyping (SKY) and interphase fish analysis Neoplasia 2001 3 62 69 11326317 10.1038/sj.neo.7900125
Olnes MI Kurl RN Isolation of nuclear extracts from fragile cells: a simplified procedure applied to thymocytes Biotechniques 1994 17 828 829 7840956
Muralidhar S Pumfery AM Hassani M Sadaie MR Kishishita M Brady JN Doniger J Medveczky P Rosenthal LJ Identification of kaposin (open reading frame K12) as a human herpesvirus 8 (Kaposi's sarcoma-associated herpesvirus) transforming gene J Virol 1998 72 4980 4988 9573267
Pang AL Taylor HC Johnson W Alexander S Chen Y Su YA Li X Ravindranath N Dvm M Rennert OM Chan WY Identification of differentially expressed genes in mouse spermatogenesis J Androl 2003 24 899 911 14581517
|
15972109
|
PMC1200560
|
CC BY
|
2021-01-04 16:24:16
|
no
|
Cancer Cell Int. 2005 Jun 22; 5:19
|
utf-8
|
Cancer Cell Int
| 2,005 |
10.1186/1475-2867-5-19
|
oa_comm
|
==== Front
Global HealthGlobalization and Health1744-8603BioMed Central London 1744-8603-1-101604276510.1186/1744-8603-1-10ReviewMedicines and vaccines for the world's poorest: Is there any prospect for public-private cooperation? Scheffler Richard M [email protected] Vikram [email protected] Director, The Nicholas C. Petris Center on Health Care Markets and Consumer Welfare, University of California, Berkeley, USA2 Distinguished Professor of Health Economics & Public Policy, University of California, Berkeley, USA3 University of California, Berkeley, 140 Warren Hall, MC7360 Berkeley, CA 94720-73604 PhD Student, Department of Economics, University of California, Berkeley, USA2005 21 7 2005 1 10 10 22 2 2005 21 7 2005 Copyright © 2005 Scheffler and Pathania; licensee BioMed Central Ltd.2005Scheffler and Pathania; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
This paper reviews the current status of the global pharmaceutical industry and its research and development focus in the context of the health care needs of the developing world. It will consider the attempts to improve access to critical drugs and vaccines, and increase the research effort directed at key public health priorities in the developing world. In particular, it will consider prospects for public-private collaboration. The challenges and opportunities in such public-private partnerships will be discussed briefly along with a look at factors that may be key to success. Much of the focus is on HIV/AIDS where the debate on the optimal balance between intellectual property rights (IPR) and human rights to life and health has been very public and emotive.
==== Body
Introduction
Infectious diseases continue to place a great burden on the people in the developing world [1]. These diseases are for the most part controlled in developed countries. Since the global pharmaceutical industry is mostly grounded in developed countries, infectious diseases are not the prime focus of research and development (R&D). An important exception is HIV/AIDS therapy. This is a pressing matter for both developed and developing countries. But even in this case, the drug cocktails and disease management protocols are designed and priced to suit customers in the developed world. Pharmaceutical firms like all corporations aim to maximize shareholder value – their R&D and pricing decisions reflect this imperative. However, the needs and the paying capacity of the rich markets in the developed world are very different from those in the developing world. Consequently, the R&D agenda does not reflect most public health priorities in developing countries. Further, innovations in drugs and vaccines that are of potentially great benefit to developing countries are priced such that they are out of reach for most people in these countries [2]. Many find this situation deeply immoral and not in the best long-term interests of the world as a whole.
This paper reviews the current status of the global pharmaceutical industry and its R&D focus in the context of the health care needs of the developing world. It will consider the attempts to improve access to critical drugs and vaccines, and increase the research effort directed at key public health priorities in the developing world. In particular, it will consider prospects for public-private collaboration. The challenges and opportunities in such public-private partnerships will be discussed briefly along with a look at factors that may be key to success. Much of the focus is on HIV/AIDS where the debate on the optimal balance between intellectual property rights (IPR) and human rights to life and health has been very public and emotive.
The Pharmaceutical Industry
The global pharmaceutical market was estimated at about US$406 billion in 2002. Figure 1 shows the geographical distribution. The US, Europe and Japan account for 77% of the market although they account for less than 15% of the global population [3]. In contrast, Sub-Saharan Africa that accounts for almost 25% of the global burden (measured in DALYs) of disease accounts for only 1% of the global spending.
Figure 1 Global Pharmaceutical Market 2002.
Among the developed countries, the US dominates. It accounts for 38% of all global spending. The US market is huge and very important, and not just because it is the most populous of developed countries. Due to a relative absence of price controls, the unit realizations of pharmaceutical companies are higher in the US.
Figure 2 shows the relative prices of a basket of all patented drugs in the US as compared to select OECD countries in 1999 [4]. Prices in the US are often twice as high. Thus, the US market is crucial important to the overall profitability of the industry, accounting for 60% of the global profits of the industry [5]. Therefore, the needs of the US market figure prominently in the priorities of decision makers in the industry and dictate much of the R&D agenda. In fact, until recently Europe has an edge in R&D spending and outcomes. Now the US has now come to dominate pharmaceutical R&D. In 2001, it spent over $30 billion on R&D as compared to $20 billion in Europe. In 1993–97, Europe launched 81 new molecular entities (NME) and America 48. But in 1998–2002, the respective figures were 44 and 85, almost an exact reversal [6]. In 2001, the major items on the R&D agenda were disorders linked to the Central Nervous System (26% of US R&D spending), Cancer, Endocrine & Metabolic Diseases (22%), and the cardiovascular system (18%.) Spending on research on developing an AIDS vaccine accounted for less than 1% of global R&D [7].
Figure 2 US is the Key Market.
This skew in R&D focus is exacerbated by the nature of the R&D process. It is a long and uncertain road from the laboratory to the marketplace. Only 1 in 5000 promising molecules makes it to the product stage. On average, each new drug costs US$800 million in R&D costs [8]. It takes almost 12 years on average to get through all the stages of drug development. Most drugs do not contribute to profits; the industry depends on a handful of the so-called 'blockbuster' drugs. Examples of these drugs include the cholesterol lowering Lipitor by Pfizer and the anti-diabetic drug Glucophage by Bristols-Myers-Squibb. Blockbuster drugs have a rising share of the total market – from 6% in 1991 to 45% in 2001.
The rising uncertainty of payoffs from R&D is partly compensated by an increase in the effective patent life. Patent life has increased from about 8 years for drugs discovered in 1980 to as much as 13–15 years for recent discoveries [9]. Patent life is increased by: reducing the time spent in the approval process and testing, and extensions of patents. On the other hand, breakthrough drugs attract competition in form of slightly differentiated products even during the patent period; so pure monopoly is restricted effectively to 1–5 years. Post-patent, generics offer stiff competition and prices are marked down sharply.
In spite of the risky and expensive nature of the R&D process, the pharmaceutical industry as a whole is very profitable. Figure 3 shows that average profit margin for the industry from 1970 to 2002 [10]. It can be seen that the margin for the Fortune 500 drug firms has consistently exceeded the margin for all Fortune 500 companies taken together. It is possible that the higher profits are compensation for higher risk [11]. Also, pharmaceutical companies do not capitalize R&D costs. Instead, these are expensed as current costs. Therefore, successful drugs can generate high profits in accounting terms as the R&D costs for such drugs have already been expensed in the past.
Figure 3 Profits in the Drug Industry.
Nevertheless, Table 1 shows that profits for the leading firms are well in excess of their current R&D outlays. The overall picture is one of an industry in robust health. This is a far cry from the picture often portrayed by industry advocates of an imperiled sector that would be pushed over the brink by price controls in developing countries.
Table 1 Top 5 by Drug Sales, 2002
Company Pharma. Revenue* ($ bn) Net Profit ($ bn) R&D ($ bn)
Pfizer 28.29 9.13 5.18
GSK 27.06 6.96 4.11
Merck 20.13 7.15 2.67
AstraZeneca 17.84 2.84 3.07
Johnson & Johnson 17.15 6.6 3.96
*Most firms have income from other products as well; so total income exceeds pharmaceutical income
Current Status in Access to Drugs
In this section, the current status in access to life saving drugs in developing countries is reviewed briefly. The focus is on access to HIV/AIDS drugs since the issue is topical, of great concern to major stakeholders, and illustrates the key issues. The World Health Organization estimates that 6 million AIDS patients can benefit immediately from Anti-Retroviral (ARV) therapy. However, only 300,000 are estimated to be currently accessing ARV [12]. This despite a sharp fall in the price of drugs in recent years. Table 2 shows the average price in US$ of ARV in May 2002 as compared to May 2001 in a few South American countries. Note that the price fell by as much as 85% in Haiti over this one-year period [13].
Table 2 Prices of Drugs in Selected Countries. Average Price in US$ of One year treatment with Anti Retro Viral (ARV: 3TC+AZT+EFV)
May 2001 May 2002
Argentina 5182 1339
Brazil 1416 1307
El Salvador 6251 5583
Haiti 12569 1484
Mexico 3914 3820
Average 5506 2499
More recently, even lower prices have been quoted. Indian generic makers have quoted a price of $140 for a year's supply of generics for ARV while leading Western firms have quoted $500 per year for branded drug [14]. These prices are a small fraction of the market price in developed countries where ARV costs several thousand dollars a year. Table 3 shows the price difference between Brazil and Spain for commonly used anti-retroviral drugs in 2002 [15]. Brazilian prices are a fraction of the Spanish prices that often tend to be among the lowest in the OECD. Indian producers are offering generics at prices lower than those in Brazil. Clearly, branded and generics drugs are being offered at substantially lower prices in many developing countries.
Table 3 Price Variation in Select AIDS Drugs
Drug Unit Brazil Spain
Aciclovir 250 mg (in vial) 1.25 4.87
Didanosine (ddl) 100 mg (tablet) 0.39 1.29
Efavirenz (EFZ) 200 mg (capsule) 0.85 3.3
Lamivudine (3TC) 150 mg (tablet) 0.29 2.7
Zidovudine (AZT) 100 mg (capsule) 0.13 0.79
Indicative Ex-Work Prices in 2001, US$ Source: Sources and Prices of Selected Drugs and Diagnostics for People Living with HIV/AIDS, WHO/UNICEF/UNAIDS/MSF, 2003
Why are prices falling so fast? There are a number of reasons. First, there has been powerful advocacy by civil society, international development agencies, and many developing countries. While much of the push has been to lower prices on branded drugs and permit use of cheaper generics, there is also a growing effort to increase R&D in vaccines and drugs suited to needs in developing countries. Public-private partnerships to promote such research are an interesting new development and are covered in greater detail below. Second, there is active use of compulsory licensing by many developing countries. The World Trade Organization permits low-income countries to grant licenses for low-cost manufacture of patented drugs if these are deemed as essential to respond to serious public health threats. Third, there is also increasing use of parallel importation wherein countries import from the cheapest international source including generic manufacturers. Finally, countries have taken steps to reduce or eliminate import duties on drugs and pool procurement orders to gain bargaining leverage through higher volumes [16].
It should be noted though that even $140 per year is about half the per capita income in many African countries. Given the sheer scale of the epidemic, paying for treatment of all AIDS patients is beyond the capacity of these states; significant external funding is required to sustain such programs. One also has to add the investments and running costs required for upgrading and maintaining infrastructure to deliver AIDS care. Table 4 shows the relative cost-effectiveness of HIV interventions in the African setting. Clearly, prevention programs still deliver the best bang for the buck and should not be neglected in any AIDS control program [17]. Prevention should be a key priority for governments and donors.
Table 4 Estimates of the Cost-Effectiveness of HIV Interventions
Intervention in Africa, 2000 Cost per life year saved
Prevention Blood Screening $3.35
STD control & management for sex workers $3.95
Voluntary Counseling & Testing $22.03
Short-course ARV treatment for pregnant women (proposed) $140*
ARV treatment Generic Drugs (proposed) $140*
Patented Drugs $560**
Full price in 2000 $10707.09
*Clinton Foundation deal with Indian generic makers; **Price to Africa of HAART from big firms Source: Masaki E. et al. "Cost-effectiveness of HIV Interventions for Resource Poor Countries: Setting Priorities for HIV/AIDS management
Industry Concerns: The Case for Property Rights
The brief discussion above shows that there is significant progress in reducing prices of ARV in developing countries. Pressure from advocacy groups and competition from cheap generics appear to have 'coerced' major pharmaceutical firms into marking down the prices of branded drugs in low income countries. But there are mounting industry concerns in this regard. Property Rights are considered the bedrock of a capitalist system and key to economic growth. Intellectual Property Rights (IPRs) are viewed as reward to innovation, and central to recouping R&D costs. Without such protection, firms warn that the incentive to invest in R&D is much attenuated with future generations around the world being the major losers [18].
The industry also has other major concerns. It fears that there will be legal or illegal re-importation of drugs into the rich markets given the huge price differential. This process can be seen unfolding in the fast growing volume of drug imports from Canada by US consumers. Further, the industry fears public pressure for price controls in key markets as consumers in developed countries become aware of low prices elsewhere. There is also fear of a domino effect: developing countries could demand lower prices for all patented drugs, not just for AIDS drugs. Finally, firms fear that without a deepening commitment to IPR, cheap generics will continue to dominate these potentially large and lucrative markets of the future.
The pharmaceutical firms' primary mission is to maximize shareholder value. They are partially justified in saying that they should not be made to pay for remedying the problem of inequitable access to drugs, a problem that is fundamentally driven by global economic inequity.
On the other hand, it can be argued that the right to life trumps the right to property. While the two rights are not mutually exclusive, they can be in immediate conflict as in the case of expensive life saving HIV/AIDS drugs and millions of impoverished AIDS patients around the world. Further, manufacturing and selling medicine is not quite the same as selling cars. Part of the mission of a pharmaceutical firm is to cure people. Also, exclusivity in the rights to ideas is questionable: new ideas build on existing knowledge. Often, private R&D uses freely available input in academic journals and conference proceedings; inputs that come from publicly funded universities and government laboratories. As a practical matter, continued foot dragging on the issue of global access to medicines may be poor corporate strategy. It generates adverse publicity, and animosity in developing countries that are destined to grow into the large markets of the future.
Thus far it seems that increasing access to drugs through lower prices has come largely through coercion of drug companies. However, there are recent systematic efforts to bring companies on board in public-private partnerships.
Public-Private Partnerships
There are numerous partnerships that have sprung up in recent years. Among the notable ones are the Alliance for Microbicide Development, the Clinton Foundation HIV/AIDS Initiative, the Global Alliance for TB Drug Development, the International AIDS Vaccine Initiative, the Malaria Vaccine Initiative and the Medicines for Malaria Venture [19].
Main Characteristics of the Partnerships
The partnerships share many structural characteristics [20]. They are usually constituted as independent legal entities. This aids in transparency and accountability. Further they may be viewed as relatively nonpartisan since they do not come encumbered with historical baggage. They have multiple partners from academia, industry, civil society, rich and poor countries, governments, and international agencies. The seed funding and some or much of the administrative costs is provided by public and philanthropic agencies. The pharmaceutical industry furnishes valuable in-kind contributions such as laboratory space, scientists' time, and access to databases. A key feature is that most of the partnerships recognize the basic validity of Intellectual Property Rights with some caveats. Indeed, this is crucial to gaining cooperation from the pharmaceutical companies.
Keys to Successful Partnerships
Most partnerships are only a few years old and it is premature to pronounce a verdict on their effectiveness. However, even in this short time frame many have started to make major strides. Three select examples are discussed below. It is already possible to identify key success factors. Most partnerships share a mix of these factors although each may bring a unique proposition to the table to entice partners. First, many partnerships have charismatic leaders and spokespersons, e.g. the Clinton Foundation that is backed by former US president Bill Clinton and the icon of the anti-apartheid struggle, Nelson. Mandela. Second, the partnerships do not merely have strong advocacy skills but also display a keen business sense. An example is the business savvy Medicines for Malaria Venture (MMV.) This is very useful in dealing credibly with the large pharmaceutical firms. Third, the partnerships, perhaps by definition have to be adept at relationship management. They have to accommodate and influence the disparate agendas of multiple partners. For instance, the International AIDS Vaccine Initiative (IAVI) has 25 partners and operations in 22 countries. Fourth, partnerships often have technical expertise. They either employ clinical, epidemiological and laboratory experts or more likely have access to their services. Fifth, most partnerships have a sharp focus on one disease in their mission and operations. This in turn allows them to build up in-depth knowledge of the disease, epidemiological trends, the current status of R&D, and the market size and trends.
The Clinton Foundation: Leveraging Charismatic Leadership
The foundation's HIV/AIDS Initiative is focused on supporting large-scale prevention & treatment in Caribbean & African countries. It develops country-level 'business' plans; and then presents these to donors and partners to mobilize resources. It has been very successful in reducing drug prices. It has been able to procure WHO-endorsed generics for ARV for as low as US$140 per year. Such low prices are now available to over 100 countries. In return, countries have to guarantee payment & secure drug distribution.
Medicines for Malaria Venture: Demonstrating Business Savvy
Widespread drug resistance to older drugs has hampered malaria control in developing countries. Until recently, there was little R&D in new drugs or vaccines, as the disease had been all but eradicated in the developed world. MMV is a global public-private partnership of academia, government research groups, and pharmaceutical firms [21]. It develops and manages "virtual" R&D i.e. it does not own the physical infrastructure or employ many scientists but it leases these resources from companies. Clearly, this calls for considerable business skills. MMV aims for 1 new drug every 5 years at a cost of about US$150 million. This is significantly lower than the average cost of US$800 million for a new drug. The reduced costs are due to effective use of in-kind contributions from companies, simpler animal models and clinical trials, and pro bono governance and management.
International AIDS Vaccine Initiative: Managing Relationships
The IAVI is focused on developing a vaccine to prevent HIV/AIDS in developing countries. It is involved in the entire gamut of operations to develop and test a vaccine – ranging from basic laboratory research to clinical tests. Its partners range from private laboratories to community-based organizations in developing countries that help recruit volunteers for clinical trials. In all, the IAVI has 25 partners and operations in 22 countries. It is the second largest supporter of AIDS vaccine R&D; it has committed US$100 million as of end-2003. The IAVI has five candidate vaccines under trial [22]. The IAVI retains the property rights to any future vaccine.
What Is in it for Pharmaceutical Companies?
Why are pharmaceutical companies willing to participate in these partnerships? There are many reasons. First, in some cases they can retain the property rights to new medicines or vaccines developed. This is subject to their commitment to sell these products at marginal cost in developing countries. But they are free to make large profits in rich markets. Second. there are spin-off benefits from R&D. For instance, new knowledge gleaned from malaria R&D is potentially applicable in other products. Third, companies gain understanding of, and access to new markets. Fourth, smaller biotech firms can get into the spotlight, with higher visibility leading to more funding, and potentially big orders. Finally, companies can project themselves as good corporate citizens. Cooperation is usually a better option than legal confrontation and adverse publicity, and losing markets to generic manufacturers.
Conclusion & Future Priorities
What are the major conclusions? It appears that the resource gap that is perceived as the main obstacle to access to drugs is shrinking. In part, this is due to an increased flow of resources from bilateral and multilateral agencies and private donors. But it is the rapidly falling price of drugs that has really helped reduce the resource gap. Providing ARV to millions of AIDS patients in developing countries at market prices is near impossible but becomes much more feasible at US$140 per year. Another key development is the emergence of focused public-private partnerships. The industry is being brought on board gradually. There are huge potential benefits if even a fraction of the industry's vast resources – laboratories, scientists, and databases can be harnessed to look for solutions to developing country health needs. As the partnerships strive to make the drug companies allies in the war on disease in developing countries, coordination across multiple partners will be a key challenge.
Coverage is still low – only a small fraction of patients are receiving drugs. Further, too little resources are devoted to R&D designed to address developing country needs that account for a huge portion of the global burden of disease. Advocacy groups should maintain pressure for lower prices. In this context, it is worth noting again that the industry is in robust financial health. The advocacy groups should also keep up pressure for increased funding from public and private donors. As the resource gap shrinks, strengthening public health infrastructure in developing countries will become a key priority. It is crucial to develop and put in place robust care delivery mechanisms that ensure smooth and secure flow of drugs, and maximize adherence to treatment protocols. Finally in the case of HIV/AIDS prevention should not be neglected. It is still the most cost-effective intervention, and therefore the most sustainable one in the long run.
==== Refs
For instance, 98% of the 10.5 million deaths among children under the age of 5 in 2002 were in developing countries. For details of the geographical distribution of the morbidity and mortality, see the World Health Report 2003
Only 5% of those who require anti-retroviral therapy in the developing world have access to it. Less than 50,000 of 4 million AIDS patients in Sub-Saharan Africa are benefiting from ARV. See the World Health Report 2003
Patented Medicine Review Board of Canada, Annual Report 1999
Economist. Trouble with Cheap Drugs Jan 29, 2004
Ibid
See the joint report by Drugs for Neglected Diseases and Medicins Sans Frontieres, "A Survey of Private Sector Drug Research and Development"
Prusoff William "One Scientist's Story" The New York Times March 19, 2001
Hunt Michie "Prescription Drugs and Intellectual Property Protection" National Institute for Health Care Management Issues Brief 2000
Scherer FM "Pricing, Profits and Technological Progress in the Pharmaceutical Industry" The Journal of Economic Perspectives 1993 7 97 115
World Health Report 2003
Price data from PAHO
Drug prices are computed from a NY Times report "Clinton Plan to Provide AIDS Drugs to Poor Countries' October 23, 2003
See joint report by WHO/UNICEF/UNAIDS/MSF, "Sources and Prices of Selected Drugs and Diagnostics for People Living with HIV/AIDS" 2003
For a good discussion of strategies to reduce prices see, "Surmounting Challenges: Procurement of Antiretroviral Medicines in Low- and Middle-Income Countries: The Experience of Medicins Sans Frontieres," a joint report by Access in Essential Medicines, WHO, and UNAIDS
Masaki E "Cost-effectiveness of HIV Interventions for Resource Poor Countries: Setting Priorities for HIV/AIDS Management"
For a summary of industry concerns, see the FAQ
Global Forum for Health Research, Initiative on Public-Private Partnerships for Health 2003
Wheeler Craig Berkley Seth "Initial lessons from public-private partnerships in drug and vaccine development" Bulletin of the World Health Organization 2001 79 728 734 11545329
|
16042765
|
PMC1200561
|
CC BY
|
2021-01-04 16:39:02
|
no
|
Global Health. 2005 Jul 21; 1:10
|
utf-8
|
Global Health
| 2,005 |
10.1186/1744-8603-1-10
|
oa_comm
|
==== Front
Health Qual Life OutcomesHealth and Quality of Life Outcomes1477-7525BioMed Central London 1477-7525-3-431602950410.1186/1477-7525-3-43ResearchInjection Drug Use Quality of Life scale (IDUQOL): A validation study Hubley Anita M [email protected] Lara B [email protected] Anita [email protected] Measurement Evaluation and Research Methodology, Dept of ECPS, 2125 Main Mall, The University of British Columbia, Vancouver, BC, Canada2 Division of Internal Medicine, Department of Medicine, Faculty of Medicine, University of British Columbia, Vancouver, BC, Canada3 Centre for Health Outcome and Evaluation Sciences, St. Paul's Hospital, Vancouver, BC, Canada2005 19 7 2005 3 43 43 4 5 2005 19 7 2005 Copyright © 2005 Hubley et al; licensee BioMed Central Ltd.2005Hubley et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Existing measures of injection drug users' quality of life have focused primarily on health and health-related factors. Clearly, however, quality of life among injection drug users is impacted by a range of unique cultural, socioeconomic, medical, and geographic factors that must also be considered in any measure. The Injection Drug User Quality of Life (IDUQOL) scale was designed to capture the unique and individual circumstances that determine quality of life among injection drug users. The overall purpose of the present study was to examine the validity of inferences made from the IDUQOL by examining the (a) dimensionality, (b) reliability of scores, (c) criterion-related validity evidence, and (d) both convergent and discriminant validity evidence.
Methods
An exploratory factor analysis using principal axis factoring in SPSS 12.0 was conducted to determine whether the use of a total score on the IDUQOL was advisable. Reliability of scores from the IDUQOL was obtained using internal consistency and one-week test-retest reliability estimates. Criterion-related validity evidence was gathered using variables such as stability of housing, sex trade involvement, high-risk injection behaviours, involvement in treatment programs, emergency treatment or overdose over the previous six months, hospitalization and emergency treatment over the subsequent six month period post data collection. Convergent and discriminant validity evidence was gathered using measures of life satisfaction, self-esteem, and social desirability.
Results
The sample consisted of 241 injection drug users ranging in age from 19 to 61 years. Factor analysis supports the use of a total score. Both internal consistency (alpha = .88) and one-week test-retest reliability (r = .78) for IDUQOL total scores were good. Criterion-related, convergent, and discriminant validity evidence supports the interpretation of IDUQOL total scores as measuring a construct consistent with quality of life.
Conclusion
The findings from this study provide initial evidence to support the use of the IDUQOL total score. The results of the study also suggest the IDUQOL could be further strengthened with additional attention to how some IDUQOL domains are described and satisfaction is measured.
Drug UseFactor AnalysisPsychometricsQuality of LifeReliabilityValidity
==== Body
Background
Existing measures of injection drug users' (IDUs) quality of life (QoL) have focused primarily on health and health-related factors. The Opiate Treatment Index, the only standardized instrument designed specifically for IDUs, is essentially a symptom checklist [1]. The Nottingham Health Profile [2,3] focuses exclusively on health. The MOS surveys (MOS SF-36, MOS SF-20 and MOS-HIV) have been used in IDU populations but because they are constructed to measure the range of health in the general population (with the exception of MOS-HIV), IDUs score very poorly [4,5]. It makes intuitive sense that IDUs have lower physical and psychological health relative to the general population. It is not surprising that IDU scores tend to be clustered at the low end of the distribution and that instruments devised for the general population may not be particularly sensitive to change in the IDU population. For example, some studies found that, in working with HIV patients with a history of injection drug use, some scales measuring the physical aspects of QoL were relatively insensitive to change and that the effects of drug use tended to overshadow the impact of HIV on health [6,7]. Among crack smokers, most SF-36 subscales did not reflect the adverse health effects of crack cocaine use and therefore appeared to have limited applicability with this population [8].
A meta-analysis of existing QoL studies indicated that QoL and health status are distinct constructs that should not be used interchangeably [9]. Of the instruments used with IDUs, only the MOS series examine QoL domains other than health. With the exception of the MOS-HIV (which was adapted for use with HIV patients), the MOS domains were chosen to measure QoL of the general population. Existing QoL tools do not measure the QoL of drug users in a culturally-sensitive fashion [10]. Problems arise with both the item content and methods of administration. These measures clearly do not take into account the full complexity of drug dependence or account for the individual factors that may compromise effective administration. The social context in which drug injectors live is likely a key component of their QoL and most measures do not capture the chronic long-term impact of drug use on diverse domains such as social, psychological, physical and occupational realms [11]. Even instruments such as the MOS-HIV that are devised for HIV-infected individuals are often not applicable to actively using IDUs because the effects of drug use tend to overshadow the impact of HIV [6].
QoL assessment continues to be widely used in clinical trials and observational studies of health and disease to evaluate clinical interventions, treatment side effects, and disease impact over time [12]. It has become evident that population-sensitive approaches that consider the many components of an individual's life that are deemed critical to his/her QoL are needed. Clearly, QoL among IDUs encapsulates a range of unique cultural, socioeconomic, political, medical, and geographic factors that must be considered in measuring QoL. With these considerations in mind, the Injection Drug User Quality of Life (IDUQOL) scale, an instrument that captures the unique and individual circumstances that determine QoL among IDUs, was developed [13]. To be able to use an instrument with confidence, however, one needs to be able to provide evidence of the validity – that is, the meaningfulness, usefulness, and appropriateness – of the inferences to be made from scores obtained on the instrument with a given population and in a given context [14-16]. The overall purpose of the present study was to examine the validity of inferences made from the IDUQOL. Several lines of construct validity evidence were examined: (a) essential unidimensionality supporting use of an IDUQOL total score, (b) internal consistency and test-retest reliability of IDUQOL scores, (c) criterion-related validity evidence, and (d) both convergent and discriminant validity evidence.
Methods
Sample
Participants consisted of a sub-sample of individuals participating in the Vancouver Injection Drug User Study (VIDUS), a longitudinal study of the incidence of HIV among IDUs in Vancouver, Canada. The research design and methods of the VIDUS have been previously described [17]. In brief, this open cohort study was initiated in 1996 to clarify the socio-demographic and behavioural determinants of HIV sero-conversion among this group. Eligibility for initial enrolment required current injection drug use (injected at least once within the last month) and evidence of recent injection was required by inspection of needle tracks. Potential participants also were required to reside in the Lower Mainland of British Columbia and provide informed consent. Most participants (82%) were recruited through word of mouth and street outreach programs. The remaining participants were referred by the needle-exchange program (5%), other storefront agencies (10%), and clinics (3%). Participants who have stopped injecting after the baseline visit are still eligible for follow-up. Trained interviewers administer a survey instrument every 6 months. Participants are asked about their demographics, needle sharing, drug using behaviour, sexual behaviours, access to clean needles and syringes, access to health care, service needs, and medical service use (e.g., self-reported visits to primary care/outpatient clinics, Emergency Department, detoxification, methadone maintenance, ambulance use and hospital admissions). Participants were reimbursed $20 CDN for each study visit, at which time referrals were provided for medical care, HIV/AIDS care, available drug and alcohol treatment and counselling as needed. The VIDUS study participants may not be representative of all IDUs because those in the lowest socioeconomic group are overrepresented in this study sample. However, it is this group that is most in need of innovative interventions.
In the present study, a total of 250 individuals were recruited in the order in which they appeared for their regularly scheduled appointment for VIDUS. A subsequent appointment for the quality of life study was scheduled and participants were paid $10 CDN in each session of the present study. Data from nine participants were excluded because of missing data or because they were deemed, at the time of data collection, to be too impaired to focus on the research tasks. The final sample consisted of 241 IDUs ranging in age from 19 to 61 years (M = 39.4, SD = 9.5 years). There were more males (63%) than females (37%) and most participants (85%) had completed high school. There were no significant socio-demographic or drug using behaviour differences between the 250 recruited individuals and the other VIDUS participants.
The first 50 participants were invited to return for a second session within 6–8 days to collect test-retest reliability data. All 50 participants returned for the second session as scheduled. In the test-retest group, 58% were male and 42% were female. These participants ranged in age from 22 to 59 years (M = 41.7, SD = 9.2). Most of the participants (90%) had completed high school.
Measures
Injection Drug User Quality of Life Scale (IDUQOL)
The present version of the IDUQOL consists of 21 life domains and builds on the original version first published by Brogly et al. [13]. Many of these domains (e.g., Being Useful, Drugs, Drug Treatment, Harm Reduction and Neighbourhood Safety) are particularly relevant to the physical, social, psychological, occupational, and geographical reality of IDUs' lives. Life domains are each represented on a 5 by 5 inch card, with the name of the domain and a simple representative picture on the front of the card and a description of the domain on the back of the card. Graphic representations were used so that this measure would be more accessible to respondents who do not speak English as a first language or have low literacy skills.
Although the administration of the IDUQOL permitted respondents to subjectively weight the importance of the life domains to his/her quality of life, a review of the literature on importance ratings and weighting [18] as well as an empirical comparison of the utility of weighted versus unweighted scores with the IDUQOL showed that weighting does not improve upon the use of simpler unweighted scores [19]. Thus, unweighted scores are used in the present study, wherein the respondent simply assigned a satisfaction rating for each life domain using a 7-point Likert-type scale ranging from 1 (very dissatisfied) to 7 (very satisfied) and illustrated with seven stylised frowning and smiling faces. Again, visual representation was included as a guide for respondents with limited English or literacy skills. Domain scores were summed and averaged to obtain an overall quality of life score ranging from 1 (very dissatisfied) to 7 (very satisfied).
Satisfaction with Life Scale (SWLS)
The SWLS is a 5-item global measure of life satisfaction [20]. Scores range from 5 to 35, with higher scores representing greater life satisfaction. This measure was selected because life satisfaction was seen as a related construct to quality of life.
Rosenberg's Self-Esteem Scale (RSES)
The RSES is a 10-item measure of global self-esteem [21]. Total scores range from 10 to 40, with higher scores representing greater self-esteem. This measure was selected because self-esteem was seen as a related construct to quality of life.
Marlowe-Crowne Social Desirability Scale Short Form X2 (MC X2)
The MC X2 [22] is a 10-item short form version of the Marlowe-Crowne Social Desirability Scale (MC SDS) [23]. Strahan and Gerbasi reported that it correlates .80 or higher with the MC SDS. The MC X2 provides an estimate of socially desirable responding as a potential source of measurement error. Total scores range from 0 to 10, with higher scores representing higher social desirability in responding. The MC X2 was selected because measures of pervasive characteristics such as social desirability are strongly recommended to assess discriminant validity [24,25].
Demographic Information
In examining criterion-related validity, the following demographic variables were used to create groups expected to differ in their quality of life: stability of housing, sex trade involvement, high-risk injection behaviours (i.e., lending or borrowing needles, daily use of heroin, cocaine, speed, or crack), involvement in a methadone maintenance program or drug treatment program, reporting hospitalization and emergency department attendance or overdose within the previous six months. Predictive criterion variables included: hospitalization and emergency treatment over the six-month period post-data collection. All variables were measured and coded dichotomously.
Procedures
Ethics approval for this study was obtained from the University of British Columbia and Providence Health Care Research Ethics Boards. Participants met one-on-one with one of three trained VIDUS staff members for a single session lasting approximately 25–30 minutes. Participants were identified only by their VIDUS study ID code on all research forms utilized for this project. All participants provided informed consent and then completed the study measures in the same order (IDUQOL, MC X2, SWLS, RSES). Demographic and predictive criterion data were obtained from VIDUS using the participants' VIDUS ID codes, a use that was disclosed to participants as part of their informed consent. Retest sessions for the sub-sample of 50 participants followed the same consent process, format, and tasks as the initial session.
Results
Essential unidimensionality and use of IDUQOL total score
To be able to use a summed total score on a measure such as the IDUQOL, it is important to demonstrate that the measure shows either strict or essential unidimensionality [26,27]. Strict unidimensionality denotes the presence of a single common factor whereas essential unidimensionality indicates the presence of a reasonably dominant common factor along with secondary minor dimensions [28,29].
An exploratory factor analysis using principal axis factoring in SPSS 12.0 was conducted on the 21 items of the IDUQOL to determine whether essential unidimensionality was present and supported the use of the IDUQOL total score. According to Gorsuch's guideline of 5 to 10 cases per item [30], the sample size for the present study (n = 241) was considered adequate for factor analysis of the 21-item IDUQOL. The data met the criteria for Bartlett's Test of Sphericity, χ2(210) = 1488.02, p < 0.001 and the Kaiser-Meyer-Olkin criteria for sampling adequacy, KMO = .88 [31]. The first factor had an eigenvalue of 6.40 and explained 30.5% of the variance in participants' responses. The ratio of the first to the second eigenvalue was 4.3 which exceeds the strict criterion of a ratio greater than 4.0 for evidence of unidimensionality [30,32,33]. These results, in addition to a visual examination of the scree plot (see Figure 1) indicated an essentially unidimensional factor structure for the IDUQOL, which supported the use of a total score [32-35]. Factor loadings ranged from .31 to .71 for all IDUQOL items on a single factor.
Figure 1 Scree Plot Showing Eigenvalues for Each Possible Factor of the IDUQOL.
Mean performance and reliability
Table 1 displays the inter-item correlation matrix for the IDUQOL. The mean inter-item correlation was .26, which Clark and Watson [36] describe as acceptable. Table 2 shows the mean performance and internal consistency of scores obtained by the sample on the IDUQOL, SWLS, RSES, and MC X2. Given the focus in the present study on the IDUQOL, gender differences on scores from this measure were also examined. No statistically significant differences in performance on the IDUQOL were found between men (M = 4.25, SD = 0.96) and women (M = 4.10, SD = 1.02), t (239) = 1.14, p = .20, and the effect size (d = 0.15) is considered small according to Cohen [37].
Table 1 Inter-item Correlations on the IDUQOL
Items BU DR DT ED FA FG FR HR HE HC HO IN LA MO NS PA RC SX SP TR
BU 1.00
DR 0.27 1.00
DT 0.14 0.37 1.00
ED 0.36 0.13 0.14 1.00
FA 0.35 0.12 0.15 0.13 1.00
FG 0.54 0.43 0.26 0.25 0.34 1.00
FR 0.31 0.22 0.20 0.28 0.35 0.52 1.00
HR 0.21 0.21 0.22 0.17 0.10 0.15 0.15 1.00
HE 0.29 0.38 0.25 0.13 0.25 0.43 0.29 0.13 1.00
HC 0.17 0.20 0.40 0.19 0.23 0.29 0.27 0.18 0.35 1.00
HO 0.23 0.30 0.21 0.16 0.30 0.33 0.40 0.10 0.40 0.37 1.00
IN 0.39 0.34 0.31 0.25 0.28 0.48 0.37 0.22 0.31 0.40 0.31 1.00
LA 0.45 0.35 0.23 0.28 0.22 0.38 0.37 0.22 0.16 0.30 0.31 0.31 1.00
MO 0.35 0.26 0.16 0.30 0.20 0.34 0.33 0.14 0.28 0.31 0.28 0.31 0.41 1.00
NS 0.31 0.18 0.24 0.19 0.22 0.32 0.42 0.26 0.29 0.36 0.36 0.37 0.34 0.34 1.00
PA 0.22 0.14 0.18 0.12 0.26 0.29 0.27 0.04 0.18 0.14 0.26 0.24 0.15 0.11 0.21 1.00
RC 0.18 0.09 0.18 0.24 0.17 0.20 0.29 0.25 0.12 0.27 0.13 0.26 0.25 0.26 0.24 0.11 1.00
SX 0.24 0.18 0.12 0.14 0.36 0.29 0.33 0.06 0.26 0.19 0.25 0.24 0.24 0.23 0.22 0.60 0.20 1.00
SP 0.27 0.25 0.28 0.25 0.29 0.45 0.37 0.17 0.21 0.16 0.18 0.29 0.30 0.24 0.17 0.18 0.11 0.20 1.00
TR 0.29 0.18 0.19 0.34 0.19 0.27 0.34 0.14 0.21 0.22 0.34 0.30 0.34 0.36 0.25 0.10 0.28 0.19 0.21 1.00
TO 0.39 0.29 0.29 0.24 0.25 0.46 0.50 0.13 0.39 0.37 0.34 0.43 0.35 0.31 0.40 0.17 0.24 0.18 0.31 0.30
Items: BU = Being Useful; DR = Drugs; DT = Drug Treatment; ED = Education; FA = Family; FG = Feeling Good; FR = Friends; HR = Harm Reduction; HE = Health; HC = Health Care; HO = Housing; IN = Independence; LA = Leisure Activities; MO = Money; NS = Neighbourhood Safety; PA = Partner(s); RC = Resources in the Community; SX = Sex; SP = Spirituality; TR = Transportation; TO = Treatment by Others.
Table 2 Mean Performance and Reliability on the IDUQOL, MC X2, SWLS, and RSES
Possible Score Range Actual Score Range Mean (Standard Deviation) Internal Consistency
IDUQOL 0 – 7 1.9 – 6.7 4.19 (0.98) .88
SWLS 5 – 35 5 – 32 14.44 (7.17) .85
RSES 10 – 40 11 – 40 27.39 (4.96) .82
MC X2 0 – 10 0 – 10 4.53 (2.10) .62
IDUQOL = Injection Drug User Quality of Life Scale, SWLS = Satisfaction with Life Scale, RSES = Rosenberg Self Esteem Scale, MC X2 = Marlowe-Crowne Social Desirability Short Form X2. Internal consistency reliability estimates were obtained using Cronbach's coefficient alpha.
In addition to an internal consistency reliability estimate, the one-week test-retest reliability estimate for the IDUQOL scores was also computed. Based on the sub-sample of 50 participants who completed the measure twice, the test-retest reliability estimate was .78, with correlations for each domain across the two sessions ranging from .32 to .67. Table 3 shows the test-retest correlations for all domains.
Table 3 One Week Test-Retest Reliability Estimates for the IDUQOL Domain and Total Scores
IDUQOL Domain Reliability Estimate
Being Useful .60**
Drugs .59**
Drug Treatment .32*
Education .44**
Family .43**
Feeling Good .67**
Friends .66**
Harm Reduction .47**
Health .44**
Health Care .44**
Housing .63**
Independence .57**
Leisure Activities .62**
Money .55**
Neighbourhood Safety .65**
Partner(s) .64**
Community Resources .34*
Sex .52**
Spirituality .57**
Transportation .59**
Treatment by Others .49**
IDUQOL Total Score .78**
* p <.05, ** p <.01; N = 50
Criterion-related validity evidence
Table 4 shows the correlations of the IDUQOL total scores with the dichotomously scored criterion variables. Of the statistically significant correlations, all were in the expected direction. That is, lower IDUQOL scores were related to unstable housing, sex trade involvement, borrowing and lending needles, daily use of heroin and speed, and overdose in the past six months. The IDUQOL scores did not correlate significantly with daily use of cocaine or crack, methadone or drug treatment, emergency treatment, or hospitalization within the six months prior to, or following, the initial test session.
Table 4 Correlations of IDUQOL Total Scores with Criterion Measures
Criterion Variable IDUQOL Total Score
Housing (stable = 0/ unstable = 1) -.16*
Engaged in sex trade (no = 0/yes = 1) -.17**
Currently borrowing needles (no = 0/yes = 1) -.19**
Currently lending needles (no = 0/yes = 1) -.25**
At least once daily use of heroin (no = 0/yes = 1) -.26**
At least once daily use of cocaine (no = 0/yes = 1) -.11
At least once daily use of speed (no = 0/yes = 1) -.14*
At least once daily use of crack (no = 0/yes = 1) -.12
Currently on methadone treatment (no = 0/yes = 1) .07
Drug treatment program in last 6 months (no = 0/yes = 1) .01
Overdose in last 6 months (no = 0/yes = 1) -.14*
Visited ER in last 6 months (no = 0/yes = 1) -.06
Hospitalized in last 6 months (no = 0/yes = 1) -.06
Visited ER in subsequent 6 months (no = 0/yes = 1) -.05
Hospitalized in subsequent 6 months (no = 0/yes = 1) .01
*p < .05, **p < .01; N = 241
IDUQOL scores are based on a wide range of domains that encompass social, physical and emotional realms, and therefore, as a total score, might not correlate significantly with specific criterion variables. To explore this possibility, analyses were carried out at the domain level, matching available criterion variables with relevant IDUQOL domains. For example, the criterion variables of engaged in sex trade and Rosenberg Self-Esteem Scale scores were correlated with the Feeling Good about Yourself IDUQOL domain score. Table 5 shows the correlations of selected IDUQOL domain scores and corresponding criterion variables.
Table 5 Correlations of Selected IDUQOL Domain Scores with Selected Criterion Variables
Criterion Variable IDUQOL Domains
Drugs
At least once daily use of heroin (no = 0/yes = 1) -.13
At least once daily use of cocaine (no = 0/yes = 1) -.07
At least once daily use of speed (no = 0/yes = 1) -.12
At least once daily use of crack (no = 0/yes = 1) -.07
Drug Treatment
Currently on methadone treatment (no = 0/yes = 1) .19**
Drug treatment program in last 6 months (no = 0/yes = 1) .21**
Feeling Good About Yourself
Rosenberg Self Esteem Scale .58**
Engaged in sex trade (no = 0/yes = 1) -.20**
Health
Currently on methadone treatment (no = 0/yes = 1) .06
Drug treatment program in last 6 months (no = 0/yes = 1) .01
Visited ER in last 6 months (no = 0/yes = 1) -.14*
Hospitalized in last 6 months (no = 0/yes = 1) -.16*
Health Care
Visited ER in last 6 months (no = 0/yes = 1) .03
Hospitalized in last 6 months (no = 0/yes = 1) .003
Housing
Housing (stable = 0/ unstable = 1) -.30**
How Others Treat You
Engaged in sex trade (no = 0/yes = 1) -.15*
*p < .05, **p < .01; N = 241
Convergent and discriminant validity evidence
Table 6 shows the correlations of IDUQOL total scores with the SWLS, RSES, and MC X2. The convergent measures (SWLS, RSES) showed moderately high correlations with the IDUQOL as would be expected between constructs that are related but not the same. The correlation between the IDUQOL and the discriminant measure (MC X2) was in the low to moderate range and thus acceptable [38]. As expected, the convergent measures were both more highly correlated with the IDUQOL total score than was the discriminant measure. Correlations were also conducted between the MC X2 and both the SWLS (r = .35) and RSES (r = .41). Because the relationship between the IDUQOL and the convergent measures could be due to the common influence of social desirability bias, partial correlations between the IDUQOL total scores and the SWLS and RSES, controlling for MC X2 scores, were conducted. These are reported in Table 6.
Table 6 Correlations and Partial Correlations of IDUQOL Total Scores With Convergent and Discriminant Measures
Correlations with IDUQOL Total Scorea Partial Correlations with IDUQOL Total Scoreb
Convergent Measures
Satisfaction With Life Scale .59** .54**
Rosenberg Self Esteem Scale .54** .47**
Discriminant Measure
Marlowe-Crowne Social Desirability Scale (MC X2) .35**
**p < .01; aN = 241; bPartial correlations control for MC X2 scores, N = 238
Discussion
The IDUQOL was developed to be a more appropriate and sensitive measure of quality of life for IDUs within their unique context of social, psychological, physical, occupational, and geographical factors. This study was designed to examine the construct validity of inferences made from the IDUQOL by exploring the factor structure, reliability, criterion-related validity evidence, and convergent and discriminant validity evidence. The exploratory factor analysis using principal axis factoring indicates the presence of essential unidimensionality, which, in turn, supports the use of a total score for the IDUQOL. Internal consistency and one week test-retest reliability estimates for the IDUQOL total score were satisfactory.
Criterion-related validity evidence for inferences made from IDUQOL total scores is weak. That is, although lower IDUQOL total scores were statistically significantly related to unstable housing, involvement in the sex trade, borrowing and lending needles, daily use of heroin and speed, and overdose in the previous six months, the correlations were low (r = -.14 to -.26). Moreover, IDUQOL total scores did not correlate significantly with daily use of cocaine or crack, methadone or drug treatment, emergency treatment within the previous six months, or hospitalization within the following six months (r = -.12 to .07). These results may not be too surprising, however, given that the IDUQOL measures numerous life domains.
When specific criterion variables were correlated with individual IDUQOL domains, some showed considerably stronger correlations (e.g., Rosenberg Self-Esteem Scale correlated .58 with the Feeling Good about Yourself domain; instability of housing correlated -.30 with the Housing domain; drug treatment program and methadone treatment correlated .21 and .19, respectively, with the Drug Treatment domain). The fact that the correlations between other criterion variables and specific domains did not change appreciably or even declined (e.g., daily use of specific drugs with the Drugs domain) suggests that there may be some lack of consistency in how participants interpreted the IDUQOL domains. For example, when rating their level of satisfaction with the Drugs domain, it is not clear whether individual participants may have indicated dissatisfaction because of a lack of availability of drugs or because of the impact of drugs in their lives. As a result, this lack of clarity may produce low or near-zero correlations between criterion variables and some IDUQOL domain ratings. In other cases, correlations may be low because of low variability (e.g., mortality) or reduced information (e.g., dichotomous (yes/no) rather than continuous (actual number) measurement of overdoses in previous six months) in the criterion variables. In future criterion-related validity research involving the IDUQOL, some criterion variables may need to be measured differently to improve the variability in scores.
These results suggest that improvements can be made to how (a) some IDUQOL domains are described, and (b) satisfaction is measured that would strengthen the utility of this measure. Individual qualitative interviews with IDUs to explore how individuals are interpreting the IDUQOL domains and assigning satisfaction ratings would provide important guidance on the types of modifications to be made. More importantly, further consideration needs to be given to how the IDUQOL can be used effectively as an outcome measure in intervention studies in which programs addressing specific aspects of quality of life (e.g., housing, health) are evaluated.
Convergent and discriminant validity evidence for the IDUQOL was strong. Convergent measures (SWLS, RSES) correlated more highly with the IDUQOL total scores than was the discriminant measure (MC X2). The moderate (r = .54 to .59) correlations between the IDUQOL total scores and the measures of related, but not identical, constructs of life satisfaction and self-esteem are to be expected. The finding of a significant but low moderate correlation of .35 between the IDUQOL total scores and the MC X2 provides evidence to support discriminant validity but also suggests social desirability plays some role in participants' responses. A similar relationship was found between the MC X2 and both the SWLS and RSES. Because the relationship between the IDUQOL and the convergent measures could be due to the common influence of social desirability bias, partial correlations between the IDUQOL total scores and the SWLS and RSES, controlling for MC X2 scores, were examined. The results showed that, although the magnitude of these correlations declined slightly, the relationships between the IDUQOL and both the SWLS and RSES were not due to social desirability bias.
Conclusion
The findings from this study provide preliminary evidence to support the meaningfulness, usefulness, and appropriateness of inferences made from IDUQOL total scores. Factor analysis supports the use of a total score. Both internal consistency (Cronbach alpha = .88) and one-week test-retest reliability (r = .78) for IDUQOL total scores are good. Convergent and discriminant validity evidence supports the interpretation of IDUQOL total scores as measuring a construct consistent with quality of life and yet distinctive from life satisfaction, self-esteem, and social desirability bias. The criterion-related validity evidence is weak, but also suggests that the utility of the IDUQOL could be further improved with greater attention to how some IDUQOL domains are described, how satisfaction is measured, and how the IDUQOL and its domains may be applied in both the development and evaluation of various interventions (e.g., drug treatment programs, health and clinical interventions, and social programs).
List of abbreviations
IDUQOL injection drug user quality of life scale
QoL quality of life
IDUs injection drug users
VIDUS Vancouver Injection Drug User Study
HIV Human immunodeficiency virus
SWLS Satisfaction with Life Scale
RSES Rosenberg's Self-Esteem Scale
MC SDS Marlowe-Crowne Social Desirability Scale
MC X2 Marlowe-Crowne Social Desirability Scale Short Form X2
Authors' contributions
AH obtained funding, designed the study, directed the statistical analyses, prepared the initial draft of the manuscript and conducted revisions. LR assisted in preparing the data, performed statistical analyses and assisted with revisions. AP conceived of the study, obtained funding, coordinated data collection, and conducted revisions of the manuscript. All authors read and approved the final manuscript.
Acknowledgements
This research was supported by an operating grant from the Canadian Institutes of Health Research (CIHR) to Dr. Anita Palepu and Dr. Anita Hubley. Additional support was provided through a Canadian Institutes for Health Research New Investigator Award and a Michael Smith Foundation for Health Research Senior Scholar Award to Dr. Anita Palepu. The authors would like to thank Kathy Li, Nancy Laliberte, Dave Isham, and Robin Brooks and the participants at the Vancouver Injection Drug User Study for their assistance in collecting and preparing the data for this research.
==== Refs
Darke S Hall W Wodak A Heather N Ward J Development and validation of a multidimensional instrument for assessing outcome of treatment among opiate users: Opiate Treatment Index British Journal of Addiction 1992 87 733 742 1591524
Torrens M San L Martinez A Castillo C Domingo-Salvany A Alonso J Use of the Nottingham Health Profile for measuring health status of patients in methadone maintenance treatment Addiction 1997 92 707 716 9246798 10.1046/j.1360-0443.1997.9267077.x
Puigdollers E Domingo-Salvany A Brugal MT Torrens M Alvaros J Castillo C Magri N Martin S Vazquez JM Characteristics of heroin addicts entering methadone maintenance treatment: Quality of life and gender Substance Use & Misuse 2004 39 1353 1368 15462234 10.1081/JA-120039392
Ryan CF White JM Health status at entry to methadone maintenance treatment using the SF-36 health survey questionnaire Addiction 1996 91 39 45 8822013 10.1046/j.1360-0443.1996.911397.x
Stein MD Mulvey KP Plough A Samet JH The functioning and well being of persons who seek treatment for drug and alcohol use Journal of Substance Abuse 1998 10 75 84 9720008 10.1016/S0899-3289(99)80142-4
Carretero MD Burgess AP Soler P Soler M Catalan J Reliability and validity of an HIV-specific health-related quality-of-life measure for use with injecting drug users AIDS 1996 10 1699 1705 8970691
Dalgard O Egeland A Skaug K Vilimas K Steen T Health-related quality of life in active injecting drug users with and without chronic hepatitis C virus infection Hepatology 2004 39 74 80 14752825 10.1002/hep.20014
Falck RS Wang J Carlson RG Siegal HA Crack-cocaine use and mental health status as defined by the SF-36 Addictive Behaviors 2000 25 579 584 10972449 10.1016/S0306-4603(99)00040-4
Smith KW Avis NE Assmann SF Distinguishing between quality of life and health status in quality of life research: A meta-analysis Quality of Life Research 1999 8 447 459 10474286 10.1023/A:1008928518577
Metzger DS O'Brien CP Spilker B Substance Abuse: The Challenge of Assessment Quality of Life Assessments in Clinical Trials 1990 New York, Raven Press, Ltd. 237 246
Koch T Life quality vs the 'quality of life': Assumptions and underlying prospective quality of life instruments in health care planning Social Science and Medicine 2000 51 419 427 10855928 10.1016/S0277-9536(99)00474-8
Globe DR Hays RD Cunningham WE Associations of clinical parameters with health-related quality of life in hospitalized persons with HIV disease AIDS Care 1999 17 71 86 10434984 10.1080/09540129948216
Brogly S Mercier C Bruneau J Palepu A Franco E Towards more effective public health programming for injection drug users: Development and evaluation of the Injection Drug User Quality of Life Scale. Substance Use & Misuse 2003 38 965 992 12801151 10.1081/JA-120017619
APA AERA NCME Standards for educational and psychological testing 1985 Washington, DC, American Psychological Association, American Educational Research Association, National Council on Measurement in Education
Hubley AM Zumbo BD A dialectic on validity: Where we have been and where we are going J Gen Psychol 1996 123 207 215
Messick S Validity of test interpretation and use. Research Report No. 90-11 1990 Educational Testing Service
Strathdee SA Patrick DM Currie SL Cornelisse PG Rekart ML Montaner JS Schechter MT O'Shaughnessy MV Needle exchange is not enough: lessons from the Vancouver injecting drug use study AIDS 1997 11 F59 65 9223727 10.1097/00002030-199708000-00001
Russell LB Hubley AM Importance ratings and weightings: Old concerns and new perspectives International Journal of Testing 2005 5 105 131 10.1207/s15327574ijt0502_2
Russell LB Hubley AM Palepu A Zumbo BD Does weighting capture what's important? Revisiting subjective importance weighting with a quality of life measure Social Indicators Research in press
Diener E Emmons RA Larsen RJ Griffin S The Satisfaction With Life Scale Journal of Personality Assessment 1985 49 71 74 16367493
Rosenberg M Society and the adolescent self-image 1965 Princeton, NJ, Princeton University Press
Strahan R Gerbasi KC Short, homonegeous versions of the Marlow-Crowne Social Desirability Scale Journal of Clinical Psychology 1972 28 191 193
Crowne DP Marlowe D A new scale of social desirability independent of psychopathology Journal of Consulting Psychology 1960 24 349 354 13813058
Campbell DT Recommendations for APA test standards regarding construct, trait and discriminant validity American Psychologist 1960 15 546 553
Foster SL Cone JD Validity issues in clinical assessment Psychological Assessment 1995 7 248 260 10.1037//1040-3590.7.3.248
Nandakumar R Ackerman T Kaplan D Test modeling The SAGE Handbook of Quantitative Methodology for the Social Sciences 2004 Thousand Oaks, CA, SAGE Publications 93 105
Pedhazur EJ Schmelkin LP Measurement and scientific inquiry Measurement, design, and analysis: An integrated approach 1991 Hillsdale, NJ, Lawrence Erlbaum Assoc. Pub.
Stout W A nonparametric approach for assessing latent trait unidimensionality Psychometrika 1987 52 589 617
Stout WF A new item response theory modeling approach with applications to unidimensionality assessment and ability estimation Psychometrika 1990 55 293 325
Gorsuch RL Factor analysis 1983 2nd Hillsdale, NJ, Lawrence Erlbaum
Pett MA Lackey NR Sullivan JJ Making sense of factor analysis: The use of factor analysis for instrument development in health care research 2003 Thousand Oaks, CA, SAGE Publications
Hattie J Methodology review: Assessing unidimensionality of tests and items Applied Psychological Measurement 1984 20 1 14
Hattie J An empirical study of the various indices for determining unidimensionality Multivariate Behavioral Research 1985 19 49 78
Fabrigar LR Wegener DT MacCallum RC Strahan EJ Evaluating the use of exploratory factor analysis in psychological research Psychological Methods 1999 4 272 299 10.1037//1082-989X.4.3.272
Russell DW In search of underlying dimensions: The use (and abuse) of factor analysis in Personality and Social Psychology Bulletin Pers Soc Psychol Bull 2002 28 1629 1646 10.1177/014616702237645
Clark LA Watson D Constructing validity: Basic issues in objective scale development Psychological Assessment 1995 7 309 319 10.1037//1040-3590.7.3.309
Cohen J A power primer Psychological Bulletin 1992 112 155 159 10.1037//0033-2909.112.1.155
Netemeyer RG Bearden WO Sharma S Scaling procedures: issues and applications 2003 Thousand Oaks, CA, SAGE Press
|
16029504
|
PMC1200562
|
CC BY
|
2021-01-04 16:38:13
|
no
|
Health Qual Life Outcomes. 2005 Jul 19; 3:43
|
utf-8
|
Health Qual Life Outcomes
| 2,005 |
10.1186/1477-7525-3-43
|
oa_comm
|
==== Front
Kinetoplastid Biol DisKinetoplastid Biology and Disease1475-9292BioMed Central London 1475-9292-4-51601880210.1186/1475-9292-4-5Original ResearchDetection of trypanosomes in small ruminants and pigs in western Kenya: important reservoirs in the epidemiology of sleeping sickness? Ng'ayo Musa O [email protected] Zablon K [email protected] Eucharia U [email protected] Geoffrey M [email protected] Ellie O [email protected] Daniel K [email protected] Department of Biochemistry and Biotechnology, Kenyatta University, P.O. Box 43844, 00100 – Nairobi, Kenya2 Molecular Biology and Biochemistry Department, International Centre of Insect Physiology and Ecology, Duduville, Kasarani, P.O. Box 30772, 00100 – Nairobi, Kenya3 Trypanosomiasis Research Centre, Kenya Agricultural Research Institute, P.O. Box 362, Kikuyu, Kenya4 Western Australia Biomedical Research Institute, Division of Health Sciences, School of Veterinary and Biomedical Sciences, Murdoch University, Murdoch, WA 6150, Australia2005 14 7 2005 4 5 5 28 9 2004 14 7 2005 Copyright © 2005 Ng'ayo et al; licensee BioMed Central Ltd.2005Ng'ayo et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Trypanosomosis is a major impediment to livestock farming in sub-Saharan Africa and limits the full potential of agricultural development in the 36 countries where it is endemic. In man, sleeping sickness is fatal if untreated and causes severe morbidity. This study was undertaken in western Kenya, an area that is endemic for both human and livestock trypanosomosis. While trypanosomosis in livestock is present at high levels of endemicity, sleeping sickness occurs at low levels over long periods, interspersed with epidemics, underscoring the complexity of the disease epidemiology. In this study, we sought to investigate the prevalence of trypanosomes in small ruminants and pigs, and the potential of these livestock as reservoirs of potentially human-infective trypanosomes. The study was undertaken in 5 villages, to address two key questions: i) are small ruminants and pigs important in the transmission dynamics of trypanosomosis? and ii), do they harbour potentially human infective trypanosomes? Answers to these questions are important in developing strategies for the control of both livestock and human trypanosomosis.
Results
Eighty-six animals, representing 21.3% of the 402 sampled in the 5 villages, were detected as positive by PCR using a panel of primers that identify trypanosomes to the level of the species and sub-species. These were categorised as 23 (5.7%) infections of T. vivax, 22 (5.5%) of T. simiae, 21 (5.2%) of the T. congolense clade and 20 (5.0%) of T. brucei ssp. The sheep was more susceptible to trypanosome infection as compared to goats and pigs. The 20 T. brucei positive samples were evaluated by PCR for the presence of the Serum Resistance Associated (SRA) gene, which has been linked to human infectivity in T. b. rhodesiense. Three samples (one pig, one sheep and one goat) were found to have the SRA gene. These results suggest that sheep, goats and pigs, which are kept alongside cattle, may harbour human-infective trypanosomes.
Conclusion
We conclude that all livestock kept in this T. b. rhodesiense endemic area acquire natural infections of trypanosomes, and are therefore important in the transmission cycle. Sheep, goats and pigs harbour trypanosomes that are potentially infective to man. Hence, the control of trypanosomosis in these livestock is essential to the success of any strategy to control the disease in man and livestock.
Trypanosoma brucei rhodesienseHuman African Trypanosomosis, sleeping sickness, epidemiologySerum Resistance Associated (SRA) genesmall ruminantspigsreservoirs.
==== Body
Background
African trypanosomosis has profoundly affected settlement and economic development in much of the African continent, especially south of the Sahara desert where it is transmitted mainly by tsetse flies. Overlaying a map of the distribution of livestock and that of tsetse infestation highlights the extent to which tsetse and trypanosomosis impede livestock development [1]. Additionally, sleeping sickness or Human African Trypanosomosis (HAT) is responsible for considerable morbidity and mortality in many countries of sub-Saharan Africa. Sleeping sickness is caused by two protozoan parasites of genus Trypanosoma: Trypanosoma brucei rhodesiense and T. b. gambiense. The latter is characterized by a chronic disease in man, often lasting several years, and is distributed mainly in western and central Africa. T. b. rhodesiense is more acute, with clinical signs apparent within weeks of infection and is confined to eastern and southern Africa. For a long time, wildlife were considered the most important, perhaps the only reservoir for both sleeping sickness and trypanosomosis in livestock. However, this changed when it was demonstrated that cattle were reservoirs of T. b. rhodesiense in Kenya [2]. With respect to T. b. gambiense on the other hand, it has been shown that pigs are reservoirs [3], although a direct man-fly-man transmission cycle is considered the more important [4]. The acute nature of sleeping sickness due to T. b. rhodesiense appears to preclude this, making the presence of a non-human reservoir obligatory in maintaining an endemic focus. Recent cases from areas where wildlife are protected, such as Serengeti National Park in Tanzania, indicate that wildlife continue to play a role in transmission of sleeping sickness [5].
Unlike cattle, sheep and goats are kept in a very broad range of agro-ecological zones [1], where they contribute considerably to the rural economies as a source of meat, milk, manure and readily disposable income [6]. Early reports suggested that trypanosomosis was not an important disease in small ruminants [7]. However, more recent studies have clearly demonstrated that sheep and goats acquire natural infections [8-12], and suffer economic loss [13,14]. Trypanosomes have also been shown to infect pigs, with reports of natural infections in different regions of Africa [9,11,15,16].
This study was undertaken in western Kenya, an area that is endemic for Human African Trypanosomosis (HAT) or sleeping sickness, and several species of trypanosomes of veterinary importance. We sought to investigate the role of small ruminants and pigs in the transmission dynamics of trypanosomosis. In this area, pigs are increasingly kept as free-range livestock, presumably thriving as a result of the varied nature of their diet. Our efforts to identify T. b. rhodesiense have benefited from evidence that the Serum Resistance Associated gene (SRA) is ubiquitous among all T. b. rhodesiense isolates tested so far that have also been characterized by other biochemical criteria [17-19].
Results
Detection of trypanosomes
The buffy coat and haematocrit centrifuge technique (HCT) [20] examination of blood from the 402 animals detected 5 trypanosome infections as shown in Table 1. The infections detected by microscopy were one T. congolense, one T. vivax and three T. brucei; no mixed infections were detected using microscopy alone. All 402 animals sampled were subjected to PCR analysis for detection of trypanosomes, based on primers specific for different species and subspecies (Table 2). Eighty-six (18.9%) animals out of the 402 were found infected by PCR. Thirty-one (27.7%) out of 112 animals sampled from Obuchun were infected by various species of trypanosomes. The infection rate was 28.05% (23/82) in Amoni, 22.1% (15/68) in Ongariama, 19.2% (10/52) in Amase and 7.9% (7/88) in Rukada. Twenty-four of the 95 sheep sampled (25.3%) were infected, while 52/255 goats (20.0%) and 10/52 pigs (19.2%) were infected. There was no significant difference in infection between animals (p > 0.05).
Table 1 Identification of trypanosome species using PCR.
Amoni
Goats 42 2 3 1 0 2 8 19.04
Sheep 37 0 6 5 0 4 15 40.54
Pigs 3 0 0 0 0 0 0 0
Amase
Goats 35 0 0 0 4 1 5 14.29
Sheep 8 0 0 0 0 1 1 12.5
Pigs 9 1 0 0 1 2 4 44.44
Ongariama
Goats 55 0 0 7 5 1 13 23.64
Sheep 7 0 0 0 0 0 0 0
Pigs 6 0 0 1 1 0 2 33.33
Obuchun
Goats 86 4 0 8 8 3 23 26.74
Sheep 13 0 1 0 1 2 4 30.77
Pigs 13 0 0 0 3 1 4 30.77
Rukada
Goats 37 1 1 0 0 1 3 8.11
Sheep 30 1 1 0 0 2 4 13.33
Pigs 21 0 0 0 0 0 0 0
TOTAL 402 9 12 22 23 20 86 21.39%
Table 2 Sequences of primers used for PCR and expected product size
Trypanosome species Primer sequence 5'-3' Size (bp) Reference
T. Congolense savannah For: CGAGAACGGGCACTTTGCGA 316 [28]
Rev: GGACAAACAAATCCCGCACA
T. congolense Kilifi For: GTGACCAAATTTGAAGTGAT 294 [28]
Rev: ACTCAAAATCGTGCACCTCG
T. vivax For: CCCGGCAGGTTGGCCGCCATC 399 [29]
Rev: TCGCTACCACAGTCGCAATCGCAATCGTCGTCTCAAGG
T. simiae For: CCGGTCAAAAACGCATT 437 [28]
Rev: AGTCGCCCGGAGTCGAT
T. brucei For: GAATATTAAACAATGCGCAG 164 [28]
Rev: CCATTTATTAGCTTTGTTGC
SRA – T. b. rhodesiense SRA-A: 5'GACAACAAGTACCTTGGCGC 460 [18]
SRA-E: 5'TACTGTTGTTGTACCGCCGC)
Prevalence of trypanosome species
The prevalence of Trypanosoma vivax as detected by PCR was 23 (5.7%), T. simiae 22 (5.5%), T. congolense savannah 9 (2.2%), T. congolense Kilifi 12 (3.0%) and T. brucei ssp 20 (5.0%). A number of mixed infections were reported in this study; T. congolense Savannah and T. brucei were encountered twice, and there was one mixed infection of T. congolense savannah and T. simiae. Sheep carried significantly more infections of T. congolense Kilifi than goats (χ2 = 9.8, df = 1, p < 0.05) and similarly with T. brucei ssp (χ2 = 6.6, df = 1, p < 0.05). There were more T. vivax infections in goats and pigs than in sheep respectively, and this difference was significant (p < 0.05).
Presence of Serum Resistance Associated (SRA) Gene
Twenty samples in which T. brucei spp. was detected by PCR were further analysed for the presence of the SRA. This analysis resulted in 3 positive samples for the SRA gene (from a goat, pig and sheep) (Figure 1), demonstrating the presence of this gene associated with human infectivity in trypanosomes infecting livestock other than cattle.
Figure 1 Map of study area showing sampling locations. Sampling was undertaken in 5 villages (4 in Teso and 1 in Busia Districts), whose GPS locations are shown in this map.
Discussion
Trypanosomes in small ruminants
Few records exist on the effect of trypanosomosis in small ruminants in East Africa. However, the work of Griffin and Allonby [8,13], who used microscopic examination of blood smears to demonstrate that the disease was an important natural infection, is a notable exception. These authors also reported that small ruminants survived better than cattle under medium tsetse challenge. More recent studies undertaken in an area of high tsetse fly challenge have shown that small ruminants succumb to trypanosomosis and that heavy economic loss is occasioned [12,14]. Although the microscopic techniques used in these studies have limitations of sensitivity [20], they provided an important basis for investigations into the importance of trypanosomosis in small ruminants. In the present study, using PCR raised the number of infections detected from 5 to 86, considerably improving our understanding of the prevalence of trypanosomosis in small ruminants and pigs.
A greater percentage of sheep were infected compared to goats and pigs, findings that are consistent with other studies that have shown sheep to be more frequently infected than goats under natural conditions [9,11,12], [21-24]. All these findings, drawn from experiments on different breeds suggest that goats are more refractory to trypanosome infections than sheep. The presence of many small ruminants with T. simiae (5 sheep and 16 goats) shows that this trypanosome has the potential to impact negatively on the growing importance of pig farming in the study area. Our data also show that sheep were more frequently infected with T. congolense Kilifi and T. brucei, while pigs and goats were more frequently infected with T. vivax.
All the 5 villages, except Ongariama, had infections of T. congolense, arguably the most important pathogens of cattle in Africa. Also, with the exception of Rukada and Amoni, T. vivax infections were identified. These two species constitute the most important trypanosomes of livestock in Africa. Considering the greater trypanotolerance of small ruminants, they may be acting as reservoirs of these pathogens, therefore limiting the economic impact of the cattle industry.
Trypanosome infections in pigs
The presence of trypanosomes in pigs has been documented from different parts of Africa [9,11,15,16]. In the present study, out of the 10 pigs infected, 3 carried T. brucei ssp., 5 had T. vivax, and there was one infection each of T. congolense savannah and T. simiae. Of important relevance to sleeping sickness epidemiology, the SRA gene was detected in one pig carrying T. brucei. These data show that pigs play a role in the epidemiology of both human and animal trypanosomosis.
Reservoirs of sleeping sickness
Small ruminants and pigs constitute a key component of livestock kept in the study area. It is now well recognized that these livestock naturally acquire trypanosome infections. On the basis of DNA amplification of the SRA gene, this study shows that sheep, goats and pigs may harbour T. b. rhodesiense. It is in this context that investigations were undertaken in these livestock as potential reservoirs of T. b. rhodesiense. Reports implicating cattle date back to the infection of human subjects with pathogens isolated from cattle [2]. In the intervening period, descriptions of human infective trypanosomes in cattle have been made on the basis of multilocus DNA fingerprinting [15], and more recently the presence of the SRA gene as a marker [17,18]. The epidemiology of HAT in western Kenya is very complex, and key factors can vary within very short geographical distances. The existence of two species of tsetse flies, G. pallidipes and G. fuscipes fuscipes, which have different habitats, and feeding preferences, adds to this complexity. G. fuscipes fuscipes feeds primarily on the Nile monitor lizard, Varanus niloticus [25]. This probably prompted experiments that led to the isolation of T. brucei from one monitor lizard [26], but the ability of these cold-blooded reptiles to support trypanosomes, other than Stercorarian forms such as T. grayi remains unconfirmed.
The ability to identify T. b. rhodesiense in reservoirs and tsetse provides a means for estimating disease risk, even in the absence of current human infections. Our study also identified one pig that was infected with trypanosomes bearing the SRA gene. It is estimated that there are more than 10,000 pigs reared in a free-range management system in the area (Dr. Samuel Maina, personal communication). Their large number, and the evidence that they can act as reservoirs underscores the importance of this study on the health and economic development of rural Africa, even on a small geographical scale.
Conclusion
To date, no significant efforts are made, either by the relevant government agents, or farmers to control trypanosomosis in small ruminants and pigs. This study contributes to the understanding of the epidemiology of sleeping sickness and livestock trypanosomosis in this area, with implications for the entire lake Victoria basin. Our conclusion is that small ruminants and pigs may constitute an enduring reservoir of trypanosomes that infect both humans and their livestock. Concerted efforts to control trypanosomes in these livestock can contribute significantly to a multi-disciplinary approach to trypanosomosis control in western Kenya, and other areas where the disease is endemic, and have an impact on the well being of the inhabitants of this region.
Methods
Study area
The study area included two districts in western Kenya, Busia and Teso. Sampling was carried out in five villages: Rukada in Busia district; Amoni, Amase, Obuchun and Ongariama in Teso district (Figure 2). These villages were selected on the basis of having reported cases of sleeping sickness within a period of about 10 years prior to the study. Samples were collected in August 2001 over a period of two weeks. The two districts lie between latitude 0° 1'South and 0° 46' North; and longitude 33° 54' West and 34° 26' East (Figure 2). The altitude varies from 1130 m to 1375 m. The mean annual rainfall is 1500 mm divided between two seasons, while annual maximum temperature ranges from 26°C to 30°C and minimum temperatures vary from 14°C to 18°C. Busia District borders Lake Victoria and the entire study area is traversed by two tributaries (rivers Sio and Malaba) and several streams that feed into them. It is a mixed farming area, with a variety of livestock species, mainly cattle, sheep and goats. Over the past two decades, free-range pig rearing has increased steadily. Two species of tsetse flies (Diptera: Glossinidae) infest the area. G. fuscipes fuscipes, which is a riverine species, associated with rivers/streams and the shores of Lake Victoria, while G. pallidipes is more widely distributed and associated with shrubs, thickets and open grassland. The distribution of G. pallidipes is discontinuous and occurs in several small foci of infestation (Kenya Trypanosomiasis Research Institute, unpublished). In areas infested with G. pallidipes, bushes of Lantana camara, a suitable habitat for resting and larviposition are widespread.
Figure 2 Detection of SRA gene. Ethidium Bromide stained 1.5% agarose gel showing PCR identification of samples containing putative T. b. rhodesiense using SRA A and E primers. M, 100 bp DNA marker; NC, Negative Control, +, positive control. Lanes 1 (pig), 9 (sheep) and 14 (goat) show the expected fragment of 460 bp, which indicates the detection of the SRA gene in these samples.
Animal sampling
Blood samples were collected from 402 animals (255 goats, 95 sheep and 52 pigs) that were brought to the sampling site by local farmers. The animals were bled from the ear vein into two capillary tubes. One capillary tube was used to estimate the percentage Packed Cell Volume (%PCV) and microscopic examination for trypanosomes, and also to make an initial diagnosis on the basis of morphology and movement [20,27]. Animals were also bled from the jugular vein into 2 ml vials containing EDTA, and stored in liquid nitrogen for further analysis.
Template preparation and PCR cycling
For each sample, 500 μl of cryopreserved blood was thawed into a single 1.5 ml microcentrifuge containing 500 μl of Saponin lysis buffer (0.15% w/v Saponin, 0.2% w/v NaCl and 1 mM EDTA) and mixed by vortexing. The mixture was then centrifuged at 11,000 g for 10 min in a microcentrifuge, followed by four washes in the same buffer. The resulting pellet was then resuspended in 100 μl of PCR buffer (50 mM KCl, 1.5 mM MgCl2 10 mM Tris-Cl, pH 8.3) and incubated at 95°C for 20 minutes, cooled and stored at -20°C for.
Standard PCR cycling was carried out in 25 μl reaction mixtures containing; 75 mM Tris-HCl, pH 8.8, 20 mM (NH4)2SO4, 0.1 % Tween 20), 200 μM of each of the four deoxynucleoside triphosphates (dNTPs), primers at 1 μM, 1 μl DNA template (except for screening the presence of SRA gene, where 4 μl of template was used), and 1 unit of Taq DNA polymerase (Fermentas MBI, Lithuania). The PCR cycling involved an initial denaturation step at 94°C for 3 minutes, followed by 30 cycles of denaturing at 94°C for 30 s, annealing at 60°C for 45 s and extension at 72°C for 30 s, with a final elongation step at 72°C for 5 minutes. Five microlitres of each PCR product was mixed with standard loading dye (Fermentas MBI, Lithuania) and electrophoresed in 1.5% agarose, with ethidium staining (5 μg/ml) and photographed under ultraviolet illumination. A positive control (with reference DNA) and a negative control (without DNA) were included with each set of reactions. PCR primers and their source references are given in Table 2.
Statistical analysis
Statistical analysis was performed using SPSS v11. The prevalence of trypanosome species and infection between animals were determined using chi-square test and odds ratios and their 95% confidence intervals. The level of significant difference was set at 0.05.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MNO undertook this study as part of his training for the Master of Science degree at Kenyatta University, where he was supervised by EUK and GMM. Field and laboratory activities were jointly supervised by ZKN (at KETRI), DKM and EOO (at ICIPE). All authors read and approved the final manuscript.
Acknowledgements
The authors gratefully acknowledge funding from the International Foundation for Science, IFS and the International Centre for Scientific Culture (ICSC) – World Laboratory. This study was carried out at KETRI and ICIPE, and we gratefully acknowledge the support from the Directors of the two institutions. This study benefited from the excellent fieldwork with support from staff at the KETRI Alupe station, which we hereby acknowledge. We are grateful to Dr. George Matete, head of Alupe station at the time of the study for coordinating the sampling.
==== Refs
Okech G Masinde A Stevenson P Ndung'u JM Njogu AR The role of isometamidium chloride in chemoprophylaxis against trypanosomosis of small ruminants in Kenya Proceedings of the 24th meeting of the International Scientific Council for Trypanosomiasis Research and Control (ISCTRC), Maputo, Mozambique) 1999 Nairobi, OAU/STRC 390 397
Onyango RJ Van Hoeve K De Raadt P The epidemiology of Trypanosoma rhodesiense sleeping sickness in Alego location central Nyanza, Kenya. I. Evidence that cattle may act as a reservoir host of trypanosomes infectious to man Trans R Soc Trop Med Hyg 1966 60 175 182 5950928 10.1016/0035-9203(66)90024-1
Mehlitz D Zillmann U Scott CM Godfrey DG Epidemiological studies on the animal reservoir of Gambian sleeping sickness Trypanozoon 1982 33 113 118
Gibson WC Will the real Trypanosoma brucei gambiense please stand up? Parasitol Today 1986 2 255 257 15462856 10.1016/0169-4758(86)90011-6
Jelinek T Bisoffi Z Bonazzi L van Thiel P Bronner U de Frey A Gundersen SG McWhinney P Ripamonti D Cluster of African trypanosomiasis in travellers to Tanzanian national parks Emerging Infect Dis 2002 8 643 635 12023926
Luckins AG Trypanosomiasis in small ruminants – a major constraint to livestock production Brit Vet J 1992 148 471 472 1467916
Stephen LE Clinical manifestations of the trypanosomiasis in livestock and other domestic animals The African trypanosomiases (George Allen and UNWIM/Ministry of overseas Development, London) 1970 774 794
Griffin L Allonby EW Disease syndromes in sheep and goats at a range research station in Kenya J Comp Path 1979 89 457 464 541451 10.1016/0021-9975(79)90037-9
Katunguka-Rwakishaya E The prevalence of trypanosomosis in small ruminants and pigs in a sleeping sickness endemic area of Buikwe country Mukono district, Uganda Rev Elev Med Vet Trop 1996 49 56 58
Snow WF Wacher TJ Rawlings P Observations on the prevalence of trypanosomiasis in small ruminants, equines and cattle, in relation to tsetse challenge, in The Gambia Vet Parasitol 1996 66 1 11 8988551 10.1016/S0304-4017(96)01003-5
Makumyaviri A Mehlitz D Kageruka P Kazyumba GL Molisho D Animal reservoir hosts of T. b. gambiense in Zaire: trypanosome infections in two foci in Bas-Zaire Trop Med Parasitol 1989 40 258 62 2617030
Masiga DK. Okech G. Irungu P Ouma J Wekesa S Ouma B Guya SO Ndung'u JM Growth and mortality in sheep and goats under high tsetse challenge in Kenya Trop Anim Health Pro 2002 34 489 501 10.1023/A:1021241220575
Griffin L Allonby EW The economic effects of trypanosomiasis in sheep and goats of a range research station in Kenya Trop Anim Health Pro 1979 11 127 132
Irungu P Nyamwaro SO Masiga DK Financial implication of rearing sheep and goats under natural trypanosomosis challenge at Galana ranch Kenya Trop Anim Health Pro 2002 34 503 513 10.1023/A:1021293204645
Hide G Welburn SC Tait A Maudlin I Epidemiological relationship of Trypanosoma brucei stock from Southeast Uganda: Evidence for different population structure in human-infective and non-human infective Parasitol 1994 109 95 111
Asoganyi T Suh S Tetuh MD Prevalence of domestic animal trypanosomiasis in the Fontem sleeping sickness focus, Cameroon Rev Elev Med Vet Trop 1990 43 69 74
Welburn SC Picozzi K Fevre EM Coleman PG Odiit M Carrington M Maudlin I Identification of human-infective trypanosomes in animal reservoir of sleeping sickness in Uganda by means of serum-resistance-associated (SRA) gene Lancet 2001 358 2017 2019 11755607 10.1016/S0140-6736(01)07096-9
Gibson W Backhous T Griffiths A The human serum resistance associated gene is ubiquitous and conserved in T. b. rhodesiense throughout East Africa Infect Genet Evol 2002 1 207 214 12798017 10.1016/S1567-1348(02)00028-X
Radwanska M Chamekh M Vanhamme L Claes F Magez S Magnus E De Baetselier P Buscher P Pays E The serum resistance-associated gene as a diagnostic tool for the detection of Trypanosoma brucei rhodesiense Am J Trop Med Hyg 2002 67 684 690 12518862
Murray M Murray PK Mclntyre WM An improved parasitological technique for the diagnosis of African trypanosomiasis Trans R Soc Trop Med Hyg 1977 71 325 326 563634 10.1016/0035-9203(77)90110-9
Bealby KA Connor RJ Rowlands GJ Trypanosomosis in Goats in Zambia ILRI (International Livestock Research Institute), Nairobi, Kenya 1996 88
Kalu AU Uzoukwu M Ikeme M Prevalence of tsetse fly and ruminant trypanosomosis in Katsina-Ala Local Government Area, Nigeria Roum Arch Microbiol Immunol 1996 55 341 52 9558969
Osear S Goossens B Kora S Gaye M Darboe L Health and productivity of traditionally managed Djallonke sheep and West African dwarf goats under high and moderate trypanosomosis risk Vet Parasitol 1999 82 101 109 10321582 10.1016/S0304-4017(99)00011-4
Kayang BB Bosompem KM Assoku RK Awumbla B Detection of T. brucei, T. congolense and T. vivax infections in cattle, sheep and goats using latex agglutination Int J Parasiot 1997 27 83 7 9076533 10.1016/S0020-7519(96)00160-9
Mohamed-Ahmad MM Odulaja A Diet activity patterns and host preferences of Glossina fuscipes fuscipes (Diptera: Glossinidae) along the shores of Lake Victoria, Kenya Bull Ent Res 1997 89 179 186
Njagu Z Mihok S Kokwaro E Verloo D Isolation of Trypanosoma brucei from the monitor lizard (Varanus niloticus) in an endemic focus of Rhodesian sleeping sickness in Kenya Acta Trop 1999 72 137 148 10206114 10.1016/S0001-706X(98)00092-8
Hoare CA 1972 trypanosomes of mammals. Blackwell scientific publications. Oxford and Edinburgh 483 484
Masiga DK Smyth AJ Bromidge TJ Hayes P Gibson WC Sensitive detection of trypanosomes in tsetse flies by DNA amplification Int J Parasitol 1992 22 909 918 1459784 10.1016/0020-7519(92)90047-O
Morlais I Ravel S Grebaut P Dumas V Cuny G New molecular marker for Trypanosoma (Duttonella) vivax identification Acta Trop 2001 80 207 213 11700177 10.1016/S0001-706X(01)00160-7
|
16018802
|
PMC1200563
|
CC BY
|
2021-01-04 16:38:32
|
no
|
Kinetoplastid Biol Dis. 2005 Jul 14; 4:5
|
utf-8
|
Kinetoplastid Biol Dis
| 2,005 |
10.1186/1475-9292-4-5
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-641598515910.1186/1465-9921-6-64ResearchMathematical modelling to centre low tidal volumes following acute lung injury: A study with biologically variable ventilation Graham M Ruth [email protected] Craig J [email protected] John F [email protected] Linda G [email protected] Bruce M [email protected] W Alan C [email protected] Department of Anesthesia, University of Manitoba, Winnipeg, Manitoba, Canada2 Institute of Industrial Mathematical Sciences, University of Manitoba, Winnipeg, Manitoba, Canada3 Department of Pathology and Laboratory Medicine, James Hogg iCAPTURE Centre for Cardiovascular and Pulmonary Research, University of British Columbia, Vancouver, British Columbia, Canada2005 28 6 2005 6 1 64 64 1 4 2005 28 6 2005 Copyright © 2005 Graham et al; licensee BioMed Central Ltd.2005Graham et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
With biologically variable ventilation [BVV – using a computer-controller to add breath-to-breath variability to respiratory frequency (f) and tidal volume (VT)] gas exchange and respiratory mechanics were compared using the ARDSNet low VT algorithm (Control) versus an approach using mathematical modelling to individually optimise VT at the point of maximal compliance change on the convex portion of the inspiratory pressure-volume (P-V) curve (Experimental).
Methods
Pigs (n = 22) received pentothal/midazolam anaesthesia, oleic acid lung injury, then inspiratory P-V curve fitting to the four-parameter logistic Venegas equation F(P) = a + b[1 + e-(P-c)/d]-1 where: a = volume at lower asymptote, b = the vital capacity or the total change in volume between the lower and upper asymptotes, c = pressure at the inflection point and d = index related to linear compliance. Both groups received BVV with gas exchange and respiratory mechanics measured hourly for 5 hrs. Postmortem bronchoalveolar fluid was analysed for interleukin-8 (IL-8).
Results
All P-V curves fit the Venegas equation (R2 > 0.995). Control VT averaged 7.4 ± 0.4 mL/kg as compared to Experimental 9.5 ± 1.6 mL/kg (range 6.6 – 10.8 mL/kg; p < 0.05). Variable VTs were within the convex portion of the P-V curve. In such circumstances, Jensen's inequality states "if F(P) is a convex function defined on an interval (r, s), and if P is a random variable taking values in (r, s), then the average or expected value (E) of F(P); E(F(P)) > F(E(P))." In both groups the inequality applied, since F(P) defines volume in the Venegas equation and (P) pressure and the range of VTs varied within the convex interval for individual P-V curves. Over 5 hrs, there were no significant differences between groups in minute ventilation, airway pressure, blood gases, haemodynamics, respiratory compliance or IL-8 concentrations.
Conclusion
No difference between groups is a consequence of BVV occurring on the convex interval for individualised Venegas P-V curves in all experiments irrespective of group. Jensen's inequality provides theoretical proof of why a variable ventilatory approach is advantageous under these circumstances. When using BVV, with VT centred by Venegas P-V curve analysis at the point of maximal compliance change, some leeway in low VT settings beyond ARDSNet protocols may be possible in acute lung injury. This study also shows that in this model, the standard ARDSNet algorithm assures ventilation occurs on the convex portion of the P-V curve.
==== Body
Background
Mathematical modelling has contributed to our understanding of lung mechanics and helped direct therapy in patients with acute respiratory distress syndrome (ARDS). Hickling [1,2] generated sigmoidal pressure-volume (P-V) curves based on a model where airway opening could occur over the entire inflation limb. Venegas and colleagues [3,4] devised a four-parameter logistic model to fit P-V inflation curves in patients with ARDS. In most instances, the Venegas equation fits static P-V data with great precision. Such modelling indicates that ventilation is limited to the convex portion of the static inflation curve when the low tidal volume (VT) ARDSNet algorithm is utilised [5].
We have recently shown mathematically that if ventilation is occurring on the convex portion of the P-V curve, there is an advantage to adding noise to the end-inspiratory pressure signal [6,7]. Using a newer mode of mechanical ventilation – termed biologically variable ventilation (BVV) – noise is added to the end-inspiratory pressure signal [8]. As configured this noise can be shown to have fractal or 1/f characteristics [9]. This computer-controlled ventilator simulates breath-to-breath variation in respiratory frequency (f) and VT that characterises normal spontaneous ventilation. The added noise results in greater mean VT over time at the same mean driving pressure. This, perhaps counter-intuitive finding, can be deduced by applying Jensen's inequality – a simple probabilistic proof [7,10]. Jensen's inequality states that the average or expected value of a convex function over a random variable is greater than the value of that function at the average of the random variable. In mathematical terms in the notation of the Venegas equation, "if F(P) = V is a convex function defined on an interval (r, s), and if pressure (P) is a random variable taking values in (r, s), then the expected value (E) at F(P); E(F(P)) > F at the expected value of P; F(E(P))." Such conditions are met with BVV since noisy ventilation provides a series of individualised observations of pressure (P), that are transformed to volume F(P) as determined by Venegas curve fitting.
Jensen's inequality, thus, provides us with an important tool to determine if noise will be beneficial or not. Indirectly it also indicates where the noise will be most beneficial – when ventilation is centred at the point where the convexity is most pronounced – the point where the second derivative of the convex interval of the function is maximised. For the Venegas equation this occurs at the point of maximal compliance change: when P = c - 1.317d or when V = a + 0.211b [3,6].
Based on the above information, we designed an experiment to compare the presumed "mathematically optimised" point about which to centre noise (Experimental) to an approach using the ARDSNet algorithm (Control). We presumed that the ARDSNet algorithm would also result in ventilation on the convex portion of the P-V curve, but advanced the hypothesis that by mathematical modelling individual P-V curves, we could find an optimised strategy for BVV that would result in discernable improvements over an approach using the ARDSNet algorithm alone with BVV. A porcine model of lung injury with oleic acid was studied. We compared gas exchange, respiratory mechanics and a single marker of inflammation over 5 hrs for the two approaches.
Methods
Experimental preparation
Twenty-eight pigs were studied following the Canadian Council on Animal Care Guidelines. The experimental preparation has been described previously [11]. Briefly, animals were ventilated initially with an Esprit® ventilator (Respironics Inc., Carlsbad, CA) using VT = 12 mL/kg, f = 20 bpm, FIO2 0.5 and PEEP 4 cm H2O during surgical placement of monitoring cannulae. Anaesthesia was maintained with an intravenous loading dose and continuous infusion of sodium thiopental/midazolam at 16/0.1 mg/kg/hr and paralysis with doxacurium infusion (1.5 – 2 mg/kg/hr).
Oleic Acid Lung Injury
Baseline measurements were obtained and an infusion of oleic acid (BDH, Toronto, ON) started at 0.2 mL/kg/hr through a catheter, placed in the inferior vena cava, above the level of the diaphragm. The oleic acid infusion was continued until PaO2 decreased to <80 mm Hg for two consecutive measurements, 5 min apart (20 – 45 min of infusion), and the volume noted. An additional 4 cm H2O PEEP was then added (to a total of 8 cm H2O) and arterial blood gases were obtained at 10 and 15 min. The criterion for study inclusion was a PaO2 > 80 mmHg and <200 mmHg on 8 cm H2O of PEEP. This was not considered to represent steady-state but was used as an index of adequate lung injury (PaO2 < 80 mm Hg) as well as evidence for lung recruitability (80 < PaO2 < 200 mm Hg). A continuous infusion of dopamine (5 – 10 μg/kg/min) was started with oleic acid infusion to maintain mean arterial pressure > 50 mmHg.
Static Pressure-Volume Curves
Pressure-volume curves were generated for each animal after established lung injury. FIO2 was increased to 1.0, PEEP decreased to 0 cm H2O, and f was decreased to 10 bpm with an inspiratory hold of 2 sec, at a square-wave flow rate of 30 L/min. A sequence of VTs from 50 to 1200 mL was delivered and plateau pressure measured 1 sec after end inspiration. Preliminary trials yielded similar curves with VT delivered in either random or ascending sequences, so the latter was used. The resulting P-V curve was analysed using a non-linear regression curve-fitting program (NCSS 97) that performs a series of iterations. We used the four-parameter logistic equation derived by Venegas et al. [3] to curve fit:
F(P) = a + b[1 + e-(P-c)/d]-1
where: a = volume at the lower asymptote, b = the vital capacity or the total change in volume between the lower and upper asymptotes, c = pressure at the true inflection point and d = an index of the linear compliance for the curve. The maximal rate of change in compliance is the point where the second derivative is maximal: found at the point P = c - 1.317d or V = a + 0.211b.
Ventilation Protocol
The animals were then randomised to BVV centred at VT of 7 mL/kg, (Control) or at the VT corresponding to the maximal rate of change in compliance from the P-V curve (Experimental) for 5 hrs. We used the computer-controller and software to generate the variable ventilatory pattern as previously described [11]. A representative variability file is shown in Figure 1. FIO2 was set at 0.5 with PEEP 8 cm H2O. Respiratory frequency was initially set at 25 bpm. We followed the ARDSNet algorithm for pH control – the base VT of 7 mL/kg in the Control group is greater than the 6 mL/kg seen in human studies due, in part, to dead space associated with in-line breathing circuit measurement devices. If pH fell below 7.2, f was incrementally increased by 5 bpm up to a maximum of 35 bpm. If respiratory acidosis persisted, VT could be adjusted in increments of 0.5 mL/kg to a maximum of 8 mL/kg. Haemodynamics, airway pressures, arterial and venous blood gases and static compliance were determined at baseline, after oleic acid, after generation of the P-V curve and then hourly for 5 hrs.
Figure 1 Delivery of Variable Tidal Volume. The complete data set of delivered tidal volume (VT) using BVV in one animal. There were 376 breaths in the file. Mean VT was set at 180 mL in this example.
Bronchoalveolar Fluid Cytokines and Wet/Dry Weight Ratios
Bronchoalveolar fluid aspirates were obtained immediately post-mortem. These samples were frozen and kept at -80°C until analysis. Analyses were made in duplicate to determine the concentrations of IL-8 by sandwich ELISA. A species-specific assay was used (IL-8, Medicorp KSC0082, detection limit 10 pg/mL). ELISA plates were incubated at 4°C overnight with 50 μL per well with 1 mg/mL of anti-IL-8. Plates were washed 4 times and nonspecific binding was blocked with 200 μL of phosphate-buffered saline (PBS) with 2% bovine serum albumin (BSA) per well for 90 min. Diluted cell-free supernatants (50 μL) were added and incubated for 3 hr. A volume of 50 μL (1 mg/mL) of biotinylated antibody was added and incubated for 60 min. Subsequently, avidin peroxidase conjugate was added (Bio-Rad Laboratories) followed by chromogen substrate (ortho-phenylenediamine [OPD], Dako). Plates were read at 490 nm using an ELISA reader (Rainbow Reader, SLT Lab Instruments). The analysis of aspirates was done in a blinded fashion at the James Hogg iCAPTURE Centre for Cardiovascular and Pulmonary Research, University of British Columbia.
Statistical Analysis
Data were analysed by repeated measures analysis of variance (ANOVA) as previously described. The group × time interactions were considered significant when p < 0.05. Least squares means test matrices were generated for post-hoc comparisons and Bonferroni's correction applied for multiple comparisons within groups. Single between group comparisons were by unpaired t-test; p < 0.05 considered significant.
Results
Four pigs died after generation of the P-V curve or within one hr of initiation of mechanical ventilation due to profound hypoxaemia and were excluded from analysis. Two pigs were excluded prior to randomisation for failure to meet blood gas criteria leaving 22 animals that completed the protocol, 11 in each group. There were no differences in body weight, volume of oleic acid infused, or dopamine dose administered between groups.
Venegas Equation Curve Fitting
All curves fit the Venegas equation with R2 > 0.995. The derived Venegas parameters for all animals are shown in Table 1. In the Control group, average VT was 7.4 ± 0.4 mL/kg. This was higher than the 7 mL/kg target due to increased VT in 6 of 11 animals to control pH following oleic acid injury as per the ARDSNet algorithm. In the Experimental group average VT, optimised to the point of maximal compliance change, was 9.5 ± 1.6 mL/kg, significantly higher than in the Control group (p < 0.05). In the Experimental group, the average 95% margin of error for the target VT was 1.06 mL/kg. Of the 11 animals in this group, 10 had VT values higher than the ARDSNet protocol, and 9 of the associated 95% confidence intervals did not contain 7.0 mL/kg. One animal had a target VT below 7.0 mL/kg, although the associated confidence interval contained 7.0 mL/kg. Combining all 22 animals as one group, values for VT calculated from the Venegas equation at P = c - 1.317d or V = a + 0.211b varied from a low of 3.3 mL/kg to a high of 13.5 mL/kg (Figure 2). This targeted VT was greater than 7 mL/kg in 18/22 animals. In the Experimental group alone, the range of individualized VTs was 6.6 -10.8 mL/kg. The corresponding Paw at the point c - 1.317d for each animal is also shown in Table 1. The range of Paw at this point showed substantial inter-animal variability ranging from 8.4 to 20.8 cm H2O, with a mean of 15.9 ± 3.4 cm H2O overall.
Table 1 Venegas Parameters
CONTROL GROUP EXPERIMENTAL GROUP
Venegas Parameters Venegas Parameters
a b c d V at c-1.317d (mL/kg) VT delivered (mL/kg) P at c-1.317d (cmH2O) a b c d V at c-1.317d (mLl/kg) VT delivered (mL/kg) P at c-1.317d (cmH2O)
1 -38.6 1165 22.9 6.8 8 7.5 14 1 -223.3 2332.3 35.3 15.8 10.2 10.1 14.5
2 -213.6 1562.9 28.5 15.3 4.5 8 8.4 2 -97.9 1652 30.7 11.9 10.5 9.7 15
3 -41.8 1472.3 26.2 9 11.7 6.6 14.5 3 -151.8 1461.8 25.4 11.3 7.1 7.1 10.5
4 -98.7 1600 25.8 7.9 9.6 7.1 15.4 4 -116.5 1645.5 26.4 9.8 10 10.9 13.5
5 -25.1 1292.2 29.2 7.9 9.2 7.2 18.7 5 -46.5 1303.4 24.1 6.4 8.2 7.9 15.7
6 -168 1152.3 27.9 12.1 3.3 7.7 12.1 6 -72.1 1400.4 31.5 9.4 10.2 11.3 19.1
7 -54 1232.8 31.9 8.4 8.6 7.7 20.8 7 -0.38 1159.6 31.4 8.1 9.4 9.8 20.7
8 -223.9 2295.2 34.5 15.4 10.4 7.5 14.3 8 -51 1573.6 25.8 8.7 10.8 11 14.3
9 10.8 1087.4 27.4 6.2 10.9 7.2 19.2 9 -23.3 1319.6 24.5 6.2 10.2 10.5 16.3
10 -64.5 1837.8 34.4 10.9 13.5 7.2 20 10 -16.6 1163.4 29.5 9.4 8.9 9.4 17.1
11 -46.7 1563.2 25.8 8.7 10.1 7.5 14.4 11 -62.2 1015.2 31.1 7.8 6.1 6.6 20.9
MEAN -87.6 1478.3 28.6 9.9 9.1 7.4 15.6 MEAN -78.3 1457.0 28.7 9.5 9.2 9.5 16.1
SD 79.0 357.5 3.7 3.2 3.0 0.4 3.7 SD 65.7 356.2 3.6 2.7 1.5 1.6 3.2
Figure 2 Mathematically Calculated versus Algorithm Low Tidal Volume. VT calculated from the point of maximal compliance change (c - 1.317d) on the P-V curve for all animals (left hand points) as compared to Control group (ARDSNet algorithm) low VT (right hand points). Mean VT of each group given by large open square, connected by the dotted line.
A representative P-V curve is shown in Figure 3, corresponding to animal 9 in the Experimental group. For this animal, the estimated parameters in the Venegas model were given by a = -23.3 mL, b = 1319.6 mL, c = 24.5 cm H2O, and d = 6.2 cm H2O. The data were collected over a wide range of pressures and volumes, and the model provided a good fit to the data (R2 = 0.996). The mean volume at the point of maximal compliance change was estimated to be VT = (a + 0.211b)/w = 10.2 mL/kg, (weight = 25 kg). The associated 95% margin of error was 0.54 mL/kg, derived from the estimated variance-covariance matrix for the calculated parameters. Thus the 95% confidence interval for this volume was 9.7 to 10.8 mL/kg, which, in this example, does not encompass the 7.0 mL/kg VT of the Control group.
Figure 3 Representative Pressure-Volume Curve Fit to the Venegas Equation. Representative P-V curve generated at zero end expiratory pressure in a single animal in the Experimental group. Dots are individual data points. The line represents the Venegas equation derived P-V curve. The Venegas parameters a, c and b are labelled, as well as the volume at the point of maximal compliance change (P = c - 1.317d) and the volume equivalent to 7 mL/kg in this animal. See text for further explanation.
Variability Frequency Distribution
The variability file for BVV introduced a coefficient of variation in VT of 15%. This translated into overall fluctuations in VT between 4.1 – 11.1 mL/kg in the Control group and 3.4 – 16.8 mL/kg in the Experimental group. Figure 4 shows the frequency distribution of VTs for each group calculated in bins of 0.5 mL/kg. A broader distribution was present in the Experimental group due to the greater range of initial centring VTs (6.6 – 11.3 mL/kg vs. 6.6 – 8 mL/kg) but it is evident that the addition of a variable ventilation pattern to both groups results in substantial VT overlap between groups.
Figure 4 Frequency Distribution Curves for Tidal Volume for the Two Groups. Frequency distribution curves of VT for each group calculated in bins of 0.5 mL/kg. The VT bins are represented on the x-axis and the percentage of all VTs from each group in each bin is represented on the y-axis. Control = solid diamonds. Experimental = open squares.
Minute Ventilation and Airway Pressures
Equivalent minute ventilation was maintained in both groups at all times. This required a statistically significant increase in f to 30 ± 5 bpm in the Control group compared to 25 ± 6 bpm (p < 0.05) in the Experimental group from hr 1 to hr 5 (Figure 5). Peak and mean Paw increased to a similar extent after oleic acid infusion in both groups. Peak Paw was modestly, but not significantly, higher in the Experimental group compared to Control (25.0 and 25.9 cm H2O at 1 and 5 hrs respectively vs. 23.2 and 22.2 cm H2O at analogous time periods in Control) (Figure 6). Mean Paw was approximately 7.5 cm H2O at baseline and increased to 12.5 cm H2O after oleic acid, not different between groups at any time-period.
Figure 5 Ventilation Parameter for the Control and Experimental Groups. Mean values of VT, f, and minute ventilation (VE) for Control (diamond symbol) and Experimental groups (square symbol) at each time period. Bars represent standard deviation. * p < 0.05 between groups at specified time periods.
Figure 6 Mean Peak and Mean Airway Pressure for the Two Groups. Mean values for peak and mean Paw for Control (diamond) and Experimental (square) groups at each time period. Bars represent standard deviation. *p < 0.05 following oleic acid.
In each animal, plateau pressure was determined by clamping the expiratory line for 12 – 18 individual breaths over the five hr period. Using this technique, plateau pressures were well within the "safe" range of <30 cm H2O in both groups: 22.3 ± 3.7 vs. 24.5 ± 4.7 cm H2O in the Control and Experimental groups respectively. Even mean peak airway pressures were below this theoretical upper limit (Figure 6). As well, plateau pressures were below the calculated value for the true inflection point c (mean value 28.6 ± 3.7 cm H2O and 28.7 ± 3.6 cm H2O in the Control and Experimental groups respectively), except for a single measurement in each 376-breath cycle of the program file (see Figure 1 for demonstration of the largest VT).
Respiratory Gases and Mechanics
Data are shown in Table 2. Similar changes in arterial blood gases occurred with oleic acid administration in each group. PaO2 decreased from approximately 250 mmHg to approximately 100 mmHg with oleic acid infusion in both groups, then it increased over 5 hrs to a mean of greater than 150 mmHg with no differences between groups. Following oleic acid injury, the PaCO2 increased significantly in both groups and remained elevated with no between group differences. Mixed venous O2 (PvO2) decreased with oleic acid injury and remained depressed over the duration of the experiment in both groups. Arterial pH fell from 7.4 at control to 7.3 after P-V curve determination in both groups, associated with the respiratory acidosis. After 1 hr of ventilation, with adjustments in f and VT, PaCO2 showed evidence of normalisation and pH improved in both groups for the duration of the study. Respiratory system compliance (Crs) decreased significantly with oleic acid administration and decreased further after P-V curve analysis, with no difference between groups and no group × time interaction.
Table 2 Respiratory Gases and Mechanics
Pre OA Post OA Post PV Curve 1 hr 5 hr
PaO2
Control Group 265 ± 23 * 101 ± 15 113 ± 9 147 ± 46 * 176 ± 36 *
Experimental Group 267 ± 19 * 89 ± 16 100 ± 45 139 ± 59 * 163 ± 62 *
PaCO2
Control Group 38 ± 4 * 48 ± 5 56 ± 9 57 ± 7 51 ± 5
Experimental Group 38 ± 2 * 48 ± 7 53 ± 10 54 ± 21 49 ± 12
PvO2
Control Group 43 ± 4 * 35 ± 3 38 ± 2 37 ± 5 35 ± 4
Experimental Group 48 ± 5 * 38 ± 5 37 ± 5 39 ± 7 36 ± 6
pHa
Control Group 7.45 ± 0.04 * 7.34 ± 0.04 7.30 ± 0.07 7.29 ± 0.06 7.35 ± 0.04
Experimental Group 7.45 ± 0.03 * 7.33 ± 0.04 7.30 ± 0.08 7.30 ± 0.08 7.35 ± 0.07
Crs
Control Group 1.15 ± 0.17 * 0.66 ± 0.12 0.58 ± 0.12 0.55 ± 0.13 0.61 ± 0.11
Experimental Group 1.11 ± 0.17 * 0.70 ± 0.22 0.58 ± 0.17 0.60 ± 0.16 0.68 ± 0.18
PaO2 = Arterial Partial Pressure O2 (mmHg)
PaCO2 = Arterial Partial Pressure CO2 (mmHg)
PvO2 = Mixed Venous Partial Pressure O2 (mmHg)
pHa = Arterial pH
Crs = Compliance L/cmH2O
* p < .05 Within Groups vs Post PV Curve
Control Group n = 11 Mean ± SD
Experimental Group n = 11 Mean ± SD
Haemodynamic Data
Results are shown in Table 3. Mean arterial pressure (MAP) did not differ between groups at any time period. With P-V curve generation, mean pulmonary artery pressure (MPAP) increased over baseline but returned towards baseline values at end experiment in both groups. Pulmonary artery occlusion pressure (PAOP) did not differ between groups at any time-period. Pulmonary vascular resistance (PVR) essentially tripled with oleic acid injury and following generation of the P-V curve with some return towards normal over time in both groups. Cardiac output (CO) decreased significantly with oleic acid injury and remained depressed in both groups for the duration of the experiment.
Table 3 Haemodynamic Data
Pre OA Post OA Post PV Curve 1 hr 5 hr
MAP
Control Group 108 ± 17 101 ± 11 111 ± 19 114 ± 21 99 ± 29
Experimental Group 119 ± 15 108 ± 25 122 ± 27 122 ± 25 104 ± 24
MPAP
Control Group 24 ± 9 * 35 ± 6 36 ± 8 32 ± 5 28 ± 3 *
Experimental Group 23 ± 5 * 35 ± 6 34 ± 7 33 ± 7 29 ± 7 *
PAOP
Control Group 9.3 ± 1.5 10.5 ± 1.9 9.9 ± 1.1 9.9 ± 1.3 9.6 ± 1.5
Experimental Group 10.0 ± 1.7 10.6 ± 1.4 11.0 ± 1.5 10.8 ± 1.5 11.1 ± 1.5
PVR
Control Group 4.3 ± 2.6 * 11.1 ± 4.0 12.4 ± 6.0 12.3 ± 4.6 8.5 ± 2.8 *
Experimental Group 3.2 ± 1.5 * 8.6 ± 2.0 8.8 ± 3.0 9.8 ± 3.1 7.6 ± 3.0 *
CO
Control Group 3.5 ± 0.7 * 2.3 ± 0.5 2.3 ± 0.6 2.0 ± 0.5 2.2 ± 0.6
Experimental Group 4.2 ± 0.8 * 2.9 ± 0.5 2.8 ± 0.6 2.5 ± 0.5 2.4 ± 0.5
MAP = Mean Arterial Pressure (mmHg)
MPAP = Mean Pulmonary Arterial Pressure (mmHg)
PAOP = Pulmonary Artery Occlusion Pressure (mmHg)
PVR = Pulmonary Vascular Resistance (mmHg·L·min-1)
CO = Cardiac Output (L/min)
* p < .05 Within Groups vs Post PV Curve
Control Group n = 11 Mean ± SD
Experimental Group n = 11 Mean ± S.D.
Bronchoalveolar Fluid Inflammatory Mediators
The average concentration of IL-8 in tracheal aspirate was 5510 ± 2540 pg/mL in the Control group versus 6500 ± 2440 pg/mL in the Experimental group, not different between groups by unpaired t-test (t statistic = -0.866, p = 0.397).
Discussion
With BVV, when "mathematically optimising" VT to the point of maximal compliance change on the convex portion of the P-V curve, no statistical improvement over the ARDSNet algorithm for VT selection was seen for oxygenation, respiratory mechanics or inflammatory cytokines in this animal model of ARDS. With either approach, acceptable gas exchange was maintained over 5 hrs and no significant differences in ventilating pressures, respiratory mechanics, dead space, shunt fraction or IL-8 cytokine levels were seen. We did not include a group with low VT in control mode as a previous study demonstrated significant advantages with BVV, both using the ARDSNet protocol [12]. Four mechanisms have been invoked to account for these advantages: i) stochastic resonance (noise enhancement of an input signal) [13], ii) Jensen's inequality [6], iii) increased surfactant [14] and iv) enhanced respiratory sinus arrhythmia[9]. The goal of the present study was therefore, not to examine BVV mechanistically, but to determine if the benefits seen previously with BVV could be optimised based on fitting the Venegas equation to individual P-V curves and then determining the ideal point on the convex portion of the curve about which to ventilate.
Using the "mathematically optimised" approach yielded mean VTs that were higher than the ARDSNet algorithm (see Figure 2). The majority of calculated VTs were between 8 and 10 ml/kg, and both Control and Experimental VTs were within the convexity of their individualised P-V curves. By ventilating with BVV in both groups, which introduced a coefficient of variation in VT of 15%, a substantial overlap of delivered VT was seen – apparent in the frequency distributions of VT (see Figure 4). We could not know a priori what VT calculated at the point P = c - 1.317d or V = a + 0.211b would be relative to Control VT but their proximity, coupled with the extensive overlap due to the addition of BVV, contributed to the lack of discernable difference between groups.
The average "mathematically optimised" VT is marginally higher than that currently recommended by proponents of low VT ventilation. However, this VT is within the range chosen by most intensive care units managing ARDS patients worldwide [15] as well as the VT selected for the control arm of the three clinical trials showing no benefit from VT reduction (Control VT = 10.8, 10.3 and 10.2 mL/kg) [16-18]. The highest derived VT from Venegas curve fitting (13.5 ml/kg) approached what some clinicians may consider unsafe in ARDS patients using conventional ventilation [19]. However, use of higher VTs in combination with BVV did not significantly increase airway pressures or IL-8 concentrations compared to Control and minute ventilation could be maintained at lower f. Lower f may ameliorate gas trapping that has been demonstrated in ARDS patients ventilated at f greater than 30 bpm [20]. Although unable to demonstrate an advantage by "mathematically optimising" VT in the Experimental group, knowledge of the point of maximal compliance in combination with BVV may provide greater flexibility in choosing ventilator settings in individual patients, permitting a higher VT/lower f combination while maintaining acceptable airway pressures.
Noisy ventilation can demonstrate Jensen's inequality in two ways. Experiments have shown that with BVV, expected mean VT is greater at the same mean Paw [21] or, conversely, that expected Paw is lower at the same mean VT [12]. That is, the noise can be introduced in either pressure or volume. In the latter case, the concavity of the inverse function
is being exploited in the region of low volumes. Thus, a lower expected plateau pressure for a given mean VT with BVV is also a consequence of Jensen's inequality.
Current recommendations to limit ventilator induced lung injury (VILI) include a combination of low VT and adequate PEEP to maintain an open lung with plateau pressures below 30 cm H2O. The plateau pressures over the course of this experiment, determined by clamping the expiratory line in 12 – 18 individual breaths per experiment, were well within the "safe" range in both groups: 22.3 ± 3.7 vs. 24.5 ± 4.7 cm H2O in the Control and Experimental groups respectively. Plateau pressures were also below the calculated value for the true inflection point c on the Venegas curve (mean value 28.6 ± 3.7 cm H2O and 28.7 ± 3.6 cm H2O in the Control and Experimental groups respectively). Even mean peak airway pressures were below this theoretical upper limit (Figure 6). In each experiment, a single breath exceeded point c (the upper bound for the convexity of the P-V curve). The average VT of this single largest breath delivered once in the file of 376 breaths (Figure 1) for the Experimental group was 426 ± 79 mL associated with a peak Paw of 42 ± 9 cm H2O at PEEP 8 cm H2O.
Brower et al. [22] recently re-analysed the ARDSNet data by quartiles for plateau pressure and found that VT reduction was associated with reduced mortality in all patients, including those with plateau pressures less than 32 cm H2O. These authors contend that there is no "safe" level of plateau pressure in acute lung injury and the lowest VT and plateau pressure compatible with acceptable gas exchange should be a goal. If this is borne out, then application of a variable ventilator pattern, which has consistently resulted in improved gas exchange and respiratory mechanics at airway pressures that were equivalent to or lower than those obtained during conventional ventilation, in combination with the lowest VT, may be beneficial.
The risk of strict adherence to low VT with conventional ventilation is alveolar derecruitment and reduced PaO2 and SaO2. While low VT strategies have been embraced, recent work indicates that survival from ARDS may be complicated by neurocognitive decline correlated to hypoxaemic periods during mechanical ventilation [23,24]. Richard et al. [25] advocate recruitment manoeuvres or increasing PEEP as alternative strategies to counteract low VT derecruitment. But recruitment manoeuvres did not show a sustained benefit for gas exchange in the ARDSNet trial [26] and increasing PEEP alone may be problematic. We have recently demonstrated that BVV is superior to an established recruitment manoeuvre to improve oxygenation in this animal model [8]. Martynowicz et al. [27], using parenchymal markers in an oleic acid injury model have demonstrated that PEEP restores airspace volume only at pressures that result in a universal increase in parenchymal stress. Eisner et al. [28] showed that the risk of barotrauma increased 1.67 fold for each 5-cm increment in PEEP and Esteban et al. [15] determined that increasing PEEP was an independent factor associated with mortality during mechanical ventilation.
A recent multi-centre trial has demonstrated no difference in patient outcome for low (8.3 cm H2O) versus high (13.2 cm H2O) levels of PEEP [29,30]. Since PEEP greater than 8 cm H2O is not beneficial to outcome, application of a variable signal to the ventilatory pattern with BVV may provide an alternative approach, producing net recruitment of previously fluid filled or atelectatic units without the potentially harmful effects of either increasing PEEP or the use of prolonged recruitment manoeuvres that deliver high levels of distending stress. Additionally, variable ventilation promotes release of endogenous surfactant, offering another mechanism for improvement in alveolar stability [14].
Potential criticisms of the present study include the use of PEEP and the use of a static P-V curve analysis for a dynamic application. Eight cm H2O PEEP was applied to both groups and resulted in peak airway pressures that exceeded the point of maximal compliance change for greater than 90% of breaths even in the Control group. The major consequence of this degree of PEEP is an upward shift of the P-V curve derived at zero end expiratory pressure (ZEEP) such that total volume would be greater at any point below c. However, in the models proposed by Hickling, this level of PEEP is associated with maintained convexity at lower VT and as such, has no effect on the interpretation of the results with regards to Jensen's inequality. We chose this level of PEEP to more closely approximate clinical practice, to provide evidence of a consistent lesion between animals with oleic acid administration and to ensure acceptable levels of gas exchange with the lower VT strategy. Initial attempts to analyse the P-V relationship on PEEP gave unreliable values for the point P = c - 1.317d in our hands; a consequence of inadequate definition of the lower asymptote in these earlier experiments.
We recognize that application of VT values obtained under "quasi" static conditions may not be directly applicable to the dynamic breathing cycle due to a shift to the right that occurs with increasing inspiratory flows and the increase in volume that may occur at the beginning of the dynamic inspirations due to recruitment from tidal volume independent of PEEP [31]. As such, our Experimental VT settings should be considered a first attempt to utilise information obtained from P-V curves to individualise BVV settings and may have also contributed to the lack of difference seen. However, a rightward shift of the curve implies that Jensen's inequality may apply over an even broader VT range provided that a convex P-V relationship is maintained. Examination of dynamic P-V loops from Rimensberger et al. [32] indicate that convexity persists. In addition, further analysis of the Venegas equation reveals its derivative closely resembles a Gaussian distribution [3]. Thus the probability density function of this equation indicates the likelihood of alveolar recruitment at a given airway pressure and the equation itself could be deemed the cumulative distribution function for alveolar recruitment. Airway opening leading to alveolar recruitment is curvilinear [13]. Under such circumstances Jensen's inequality implies that a noisy mean driving pressure could augment recruitment.
In an oleic acid lung injury model, Wilson and colleagues [33] suggested that oedematous lung did not open and close, but that alveoli changed from fluid-filled units to air-filled units with the P-V curve strongly sigmoidal. Following lung injury, at 3-5 cm H2O PEEP, convex curvature was seen in P-V curves over the range of inflation pressures seen in our study. They examined P-V curves for regional lung volumes of 1-2 mL, using the parenchymal marker technique. Thus in a similar model to ours, convexity was demonstrated for small regional lung volumes, suggesting applicability of Jensen's inequality with an oleic acid lung injury model of ARDS in the presence of PEEP. These authors further developed a mathematical model based on their findings. Until airway pressure exceeded 8 cm H2O (the level of PEEP in our study), the duct to the fluid-filled alveolus was assumed blocked by a liquid bridge. Above this pressure, the P-V curve was convex as the air bubble penetrated the mouth of the alveolus for various degrees of fluid filling. When the alveolus remained fluid-filled, the lung compliance was low. An abrupt change occurred as the air bubble entered the alveolus. At this transition, compliance rapidly increased with no change in alveolar tissue volume. Based on this modeling, when the lung is oedematous, noise added to the mean airway pressure signal will increase the likelihood of inducing the abrupt change in compliance, seen with entry of air bubbles into alveoli, a situation that would result in improved gas exchange.
The lack of difference between groups suggests that determination of a "mathematically optimal" VT may be clinically irrelevant during BVV, provided that ventilation is occurring on the convex portion of the P-V curve. Adhering to the ARDSNet algorithm in this study assured that ventilation occurred on the convex portion of the curve in all animals. Knowing this may be advantageous due to the difficulties applying static curves to dynamic conditions listed above. Moreover, rigorous definition of the P-V relationship requires a finite time off the ventilator, can be difficult to analyse clinically, and imposes a potential risk for instability in gas exchange and mechanics during the manoeuvre. PaO2 and compliance decreased in 18 of 22 animals immediately following P-V curve determination in the present study, providing credence to the above concerns. However, the results of the present study suggest that when ventilating with BVV, knowledge of the point of maximal compliance change on the convex portion of the P-V curve relative to VT determined using the ARDSNet algorithm might permit adjustments in VT in selected patients without the risk of excessive airway pressures. Finally, Jensen's inequality can be generalised and as such defining a simpler equation for the convex interval of the P-V curve under low VT conditions is possible.
We chose to examine only intratracheal IL-8 levels as a marker of inflammatory changes in the present study as previous work in our laboratory did not demonstrate measurable effects on tumour necrosis factor α, IL-6 or IL-10 in this porcine oleic acid model [11]. The high level of tracheal fluid IL-8 measured in both groups is comparable to our previous results. Similar cytokine levels for the two ventilatory approaches suggest that the inflammatory injury was comparable.
Conclusion
In this porcine model of acute lung injury, "mathematically optimised" P-V curve fitting to calculate the mean VT/kg about which to centre variable ventilation yielded a broad range for this calculated volume. Although not clearly advantageous over the standard approach – low VT as determined by the ARDSNet algorithm – VT selected mathematically according to the point of maximal compliance change on the P-V curve, in combination with a variable pattern of ventilation, may permit some leeway in VT settings provided that airway pressures are maintained within acceptable limits. This study also indicates that the standard ARDSNet algorithm assures ventilation is occurring on the convex portion of the P-V curve with this model. Application of Jensen's inequality provides theoretical proof of why a noisy or variable ventilatory approach is advantageous under these circumstances.
Competing interests
Dr. Mutch is co-founder of Biovar Life Support Inc., which has developed the mechanical ventilator described in this paper. Worldwide exclusive rights to this ventilator have been licensed to Respironics Inc. To date no ventilators have been sold clinically. In the event of sales of this ventilator, Dr. Mutch and the University of Manitoba would stand to gain financially. None of the other authors have a financial interest in the ventilator.
Authors' contributions
M.R.G. supervised conduct of the experiments, helped analyse the data and helped write the paper. C.J.H. was responsible for conduct of the experiments as a fellow in the Anesthesia Laboratory. J.F.B. did the Jensen's inequality modelling and the statistical analysis related to the P-V curve fitting. L.G.G. helped with the experiments, data retrieval and collation and table and figure production. B.M.M. supervised the cytokine assays and their interpretation and helped write the paper. W.A.C.M. conceived the study, analysed and interpreted data and helped write the paper.
Acknowledgements
The authors thank Respironics Inc. for providing the Esprit ventilator for the study and the hardware and software development for BVV as well as financial support. The authors also thank Biovar Life Support Inc. for financial support. We thank Elizabeth Walker for carrying out the cytokine assays.
==== Refs
Hickling KG The pressure-volume curve is greatly modified by recruitment Am J Respir Crit Care Med 1998 158 194 202 9655729
Hickling KG Best compliance during a decremental, but not incremental, positive end-expiratory pressure trial is related to open-lung positive end-expiratory pressure: a mathematical model of acute respiratory distress syndrome lungs Am J Respir Crit Care Med 2001 163 69 78 11208628
Venegas JG Harris RS Simon BA A comprehensive equation for the pulmonary pressure-volume curve J Appl Physiol 1998 84 389 395 9451661
Harris RS Hess DR Venegas JG An objective analysis of the pressure-volume curve in the acute respiratory distress syndrome Am J Respir Crit Care Med 2000 161 432 439 10673182
The Acute Respiratory Distress Syndrome Network Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome N Engl J Med 2000 342 1301 1308 10793162 10.1056/NEJM200005043421801
Brewster JF Graham MR Mutch WAC Convexity, Jensen's inequality and benefits of noisy mechanical ventilation J R Soc Interface 2005 2 393 6 16849198 10.1098/rsif.2005.0043
Jensen JL Sur les functions convexes et les inégualités entre les valeurs moyennes Acta Math 1906 30 175 193
Funk DJ Graham MR Girling LG Thliveris JA McManus BM Walker EK Rector ES Hillier C Scott JE Mutch WAC A comparison of biologically variable ventilation to recruitment manoeuvres in a porcine model of acute lung injury Respir Res 2004 5 22 15563376 10.1186/1465-9921-5-22
Mutch WAC Graham MR Girling LG Brewster JF Fractal ventilation enhances respiratory sinus arrhythmia Respir Res 2005 6 41 15882460 10.1186/1465-9921-6-41
Ross S A First Course in Probability 2002 6 Upper Saddle River, NJ, USA:Prentice Hall
Mutch WAC Lefevre GR Health, 'small-worlds', fractals and complex networks: an emerging field Med Sci Monit 2003 9 MT19 MT23 12761464
Boker A Graham MR Walley KR McManus BM Girling LG Walker E Lefevre GR Mutch WAC Improved arterial oxygenation with biologically variable or fractal ventilation using low tidal volumes in a porcine model of acute respiratory distress syndrome Am J Respir Crit Care Med 2002 165 456 462 11850336
Suki B Alencar AM Sujeer MK Lutchen KR Collins JJ Andrade JR. JS Ingenito EP Zapperi S Stanley HE Life-support system benefits from noise Nature 1998 393 127 128 9603516 10.1038/30130
Arold SP Suki B Alencar AM Lutchen KR Ingenito EP Variable ventilation induces endogenous surfactant release in normal guinea pigs Am J Physiol Lung Cell Mol Physiol 2003 285 L370 L375 12851212
Esteban A Anzueto A Frutos F Alia I Brochard L Stewart TE Benito S Epstein SK Apezteguia C Nightingale P Arroliga AC Tobin MJ Characteristics and outcomes in adult patients receiving mechanical ventilation: a 28-day international study JAMA 2002 287 345 355 11790214 10.1001/jama.287.3.345
Stewart TE Meade MO Cook DJ Granton JT Hodder RV Lapinsky SE Mazer CD McLean RF Rogovein TS Schouten BD Todd TRJ Slutsky AS Evaluation of a ventilation strategy to prevent barotrauma in patients at high risk for acute respiratory distress syndrome N Engl J Med 1998 338 355 361 9449728 10.1056/NEJM199802053380603
Brochard L Roudot-Thoraval F Roupie E Delclaux C Chastre J Fernandez-Mondejar E Clementi E Mancebo J Factor P Matamis D Ranieri M Blanch L Rodi G Mentec H Dreyfuss D Ferrer M Brun-Buisson C Tobin M Lemaire F Tidal volume reduction for prevention of ventilator-induced lung injury in acute respiratory distress syndrome. The Multicenter Trail Group on Tidal Volume reduction in ARDS Am J Respir Crit Care Med 1998 158 1831 1838 9847275
Brower RG Shanholtz CB Fessler HE Shade DM White PJ Wiener CM Teeter JG Dodd-o JM Almog Y Piantadosi S Prospective, randomized, controlled clinical trial comparing traditional versus reduced tidal volume ventilation in acute respiratory distress syndrome patients Crit Care Med 1999 27 1492 1498 10470755 10.1097/00003246-199908000-00015
Tobin MJ Culmination of an era in research on the acute respiratory distress syndrome N Engl J Med 2000 342 1360 1361 10793169 10.1056/NEJM200005043421808
de Durante G Del Turco M Rustichini L Cosimini P Giunta F Hudson LD Slutsky AS Marco Ranieri V ARDSNet lower tidal volume ventilatory strategy may generate intrinsic positive end-expiratory pressure in patients with acute respiratory distress syndrome Am J Respir Crit Care Med 2002 165 1271 1274 11991877 10.1164/rccm.2105050
Mutch WAC Harms S Lefevre GR Graham MR Girling LG Kowalski SE Biologically variable ventilation increases arterial oxygenation over that seen with positive end-expiratory pressure alone in a porcine model of acute respiratory distress syndrome Crit Care Med 2000 28 2457 2464 10921579 10.1097/00003246-200007000-00045
Brower RG Krishnan JA Thompson BT Effects of tidal volume reduction (VT-R) in acute lung injury patients with inspiratory plateau pressures < 32 cmH2O before VT-R Am J Respir Crit Care Med (supplements) 2003 167 A616
Hopkins RO Weaver LK Collingridge D Parkinson RB Chan KJ Orme JFJ Two-year cognitive, emotional, and quality-of-life outcomes in acute respiratory distress syndrome Am J Respir Crit Care Med 2005 171 340 347 15542793 10.1164/rccm.200406-763OC
Hopkins RO Weaver LK Chan KJ Orme JFJ Quality of life, emotional, and cognitive function following acute respiratory distress syndrome J Int Neuropsychol Soc 2004 10 1005 1017 15803563 10.1017/S135561770410711X
Richard JC Maggiore SM Jonson B Mancebo J Lemaire F Brochard L Influence of tidal volume on alveolar recruitment. Respective role of PEEP and a recruitment maneuver Am J Respir Crit Care Med 2001 163 1609 1613 11401882
Villagra A Ochagavia A Vatua S Murias G Del Mar F Lopez AJ Fernandez R Blanch L Recruitment maneuvers during lung protective ventilation in acute respiratory distress syndrome Am J Respir Crit Care Med 2002 165 165 170 11790648
Martynowicz MA Minor TA Walters BJ Hubmayr RD Regional expansion of oleic acid-injured lungs Am J Respir Crit Care Med 1999 160 250 258 10390408
Eisner MD Thompson BT Schoenfeld D Anzueto A Matthay MA Airway pressures and early barotrauma in patients with acute lung injury and acute respiratory distress syndrome Am J Respir Crit Care Med 2002 165 978 982 11934725
Brower RG Lanken PN Macintyre N Matthay MA Morris A Ancukiewicz M Schoenfeld D Thompson BT Higher versus lower positive end-expiratory pressures in patients with the acute respiratory distress syndrome N Engl J Med 2004 351 327 336 15269312 10.1056/NEJMoa032193
Levy MM PEEP in ARDS – how much is enough? N Engl J Med 2004 351 389 391 15269320 10.1056/NEJMe048103
Adams AB Cakar N Marini JJ Static and dynamic pressure-volume curves reflect different aspects of respiratory system mechanics in experimental acute respiratory distress syndrome Respir Care 2001 46 686 693 11455939
Rimensberger PC Cox PN Frndova H Bryan AC The open lung during small tidal volume ventilation: concepts of recruitment and "optimal" positive end-expiratory pressure Crit Care Med 1999 27 1946 1952 10507623 10.1097/00003246-199909000-00038
Wilson TA Anafi RC Hubmayr RD Mechanics of edematous lungs J Appl Physiol 2001 90 2088 2093 11356770
|
15985159
|
PMC1200564
|
CC BY
|
2021-01-04 16:23:24
|
no
|
Respir Res. 2005 Jun 28; 6(1):64
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-64
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-1911604580310.1186/1471-2105-6-191Research ArticleEvaluation of normalization methods for cDNA microarray data by k-NN classification Wu Wei [email protected] Eric P [email protected] Connie [email protected] I Saira [email protected] Mina J [email protected] Life Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA2 Dorothy P. and Richard P. Simmons Center for Interstitial Lung Disease, Division of Pulmonary, Allergy and Critical Care Medicine, University of Pittsburgh Medical Center, Pittsburgh, PA 15213, USA3 Center for Automated Learning and Discovery and Language Technology Institute, School of Computer Science, Carnegie Mellon University, Pittsburgh, PA 15213, USA2005 26 7 2005 6 191 191 17 12 2004 26 7 2005 Copyright © 2005 Wu et al; licensee BioMed Central Ltd.2005Wu et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Non-biological factors give rise to unwanted variations in cDNA microarray data. There are many normalization methods designed to remove such variations. However, to date there have been few published systematic evaluations of these techniques for removing variations arising from dye biases in the context of downstream, higher-order analytical tasks such as classification.
Results
Ten location normalization methods that adjust spatial- and/or intensity-dependent dye biases, and three scale methods that adjust scale differences were applied, individually and in combination, to five distinct, published, cancer biology-related cDNA microarray data sets. Leave-one-out cross-validation (LOOCV) classification error was employed as the quantitative end-point for assessing the effectiveness of a normalization method. In particular, a known classifier, k-nearest neighbor (k-NN), was estimated from data normalized using a given technique, and the LOOCV error rate of the ensuing model was computed. We found that k-NN classifiers are sensitive to dye biases in the data. Using NONRM and GMEDIAN as baseline methods, our results show that single-bias-removal techniques which remove either spatial-dependent dye bias (referred later as spatial effect) or intensity-dependent dye bias (referred later as intensity effect) moderately reduce LOOCV classification errors; whereas double-bias-removal techniques which remove both spatial- and intensity effect reduce LOOCV classification errors even further. Of the 41 different strategies examined, three two-step processes, IGLOESS-SLFILTERW7, ISTSPLINE-SLLOESS and IGLOESS-SLLOESS, all of which removed intensity effect globally and spatial effect locally, appear to reduce LOOCV classification errors most consistently and effectively across all data sets. We also found that the investigated scale normalization methods do not reduce LOOCV classification error.
Conclusion
Using LOOCV error of k-NNs as the evaluation criterion, three double-bias-removal normalization strategies, IGLOESS-SLFILTERW7, ISTSPLINE-SLLOESS and IGLOESS-SLLOESS, outperform other strategies for removing spatial effect, intensity effect and scale differences from cDNA microarray data. The apparent sensitivity of k-NN LOOCV classification error to dye biases suggests that this criterion provides an informative measure for evaluating normalization methods. All the computational tools used in this study were implemented using the R language for statistical computing and graphics.
==== Body
Background
Molecular profiling technology allows for the simultaneous assaying of the abundance of tens of thousands of transcripts in a biological sample. Once these abundance values have been obtained for many samples, prevalent higher-order data analyses may include clustering, classification, feature selection, and network estimation. A variety of algorithms seeking to address these higher-order tasks have been investigated and applied, to interpret gene expression patterns and to generate biological predictions. However, the accuracy of these predictions may depend on the low-level transformations utilized to produce abundance values from raw measurements, i.e., data pre-processing may be a critical factor in determining the validity and success of downstream studies. Some key pre-processing steps for profiling data include image quantification and normalization. Several image analysis software (e.g., GenePix and SPOT) have been designed for image analysis of the spots on microarrays [1,2]. Background estimation has also been considered as an important issue in image quantification, however, evidence [2,3] showed that 'inappropriate' local background adjustment could add noise into the microarray data and thus be detrimental to the downstream studies. Background adjustment, therefore, is still an issue to be resolved. After image analysis, normalization usually needs to be performed. It is a procedure designed to minimize the unwanted variations in measurements arising from the technology, but to retain the intrinsic biological variations, and is also the focus of this work. In this study, we examined normalization in the context of a particular transcriptional profiling platform, cDNA microarrays [4-6], and the specific analytical task of classifying biological samples characterized by gene expression profiles.
In cDNA microarray-based investigations, RNA from two samples are reverse-transcribed and labeled with distinct (red and green) fluorescent dyes, then hybridized to a microarray spotted with DNA sequences ("probes"). An ensuing scanned image of the microarray is processed to yield an intensity measurement for each dye at every spot (Figure 1). If R and G are the spot-specific, quantitated, fluorescent intensities of the target and reference expression signals respectively, relative gene expression is defined as the log ratio M = log2(R / G), and average expression is the log intensity . Based on different biological assumptions and design principles, many normalization methods for cDNA microarray data have been proposed. Global normalization techniques adjust the center (e.g., mean or median) of the distribution of the log ratio M values on each microarray to a constant [1,7-9]. These methods, however, do not correct for any intensity- or spatial effect.
Figure 1 A scanned image of an illustrative cDNA microarray. The configuration (layout of spots) can be described via a previously defined notation encompassing four numbers (ngr, ngc, nsr, nsc) [12]. A print-tip (PT) group is a set of spots arranged in a grid with "nsr" rows and "nsc" columns. A microarray is a set of PT groups arranged in a pattern of "ngr" rows and "ngc" columns. The configuration of the microarray shown is (ngr = 2, ngc = 2, nsr = 24, nsc = 24), i.e., 2 × 2 PT groups each composed of 24 × 24 spots. The terms "local" and "global" level refer to the spots in a PT group and the entire microarray respectively.
A variety of techniques have been proposed to remove intensity effect. A non-linear approach employs robust locally weighted regression (lowess) [10] to smooth the dependence of log ratios on intensities [4,11,12]. The basic assumption of this approach is either that the majority of genes are not differentially expressed, or that genes are influenced by random effects (i.e., the numbers of up-regulated and down-regulated genes are similar) [4,11,12]. A 'qspline' method uses a target array to adjust R and G values so that their distribution is similar for all arrays [13], but the performance of this method may depend upon the choice of the baseline array [14]. A composite method employs both external control samples and total genes on a microarray to remove intensity effect [15]. To relax critical biological assumptions, 'housekeeping-gene'-related methods first identify non-differentially-expressed genes, and then use these genes for normalization [16-18]. Semi-linear models are designed to account for the effects of print-tips (PTs), signal intensity, and differences in gene expression levels jointly in a single model [19,20].
The removal of intensity effect at the PT level can partially remove spatial effect [4,11]. To remove spatial effect more completely, the dependence of M values on physical position can be smoothed using lowess [12], or can be eliminated using weighted mean [13] or median filter methods [17], both of which assume that differentially expressed genes are not co-localized in the neighboring spots. Since spatial- and intensity effect may be mutually dependent, a method that removes global spatial effect and global intensity effect in a single step has been proposed [21].
Whereas the above location normalization methods remove spatial- and intensity effect, scale normalization methods adjust differences in the scale of M values within and/or between microarrays. The assumption is that since the majority of genes are not differentially expressed, the scale of their M values should be constant. A robust estimate of the scale factor for scale normalization is median absolute deviation [15].
Normalization approaches seek to ensure that dye effect is removed, while biological variations are retained. Spatial- and intensity effect and scale effect arise from printing, hybridization, scanning, or other technical factors, and can mask the signals arising from genuine biological variations in gene expression. Visual aids used to assess the effectiveness of normalization methods [11,13,15,21] include scatter plots of log ratio (M) versus average log intensity (A) ("MA plots"). Spatial plots are a color-coded representation of each spot on a microarray that depicts M values, or a quality (e.g., shape, size) measure of some test statistic. These two types of diagnostic plots [4,21] suggest that raw M values are often biased estimates of relative expression and that the dye intensities per spot need to be adjusted. Quantitative criteria used to assess the robustness of normalization methods in removing dye effect include (i) rank variations of spot intensity in non-normalized versus normalized data [9,22], and (ii) correlation [16,21], variance [8,13], or error [18,22] of the normalized M values in replicated data.
To ensure that biological variations are retained after normalization, several functional criteria have been employed. Prevailing approaches determine the ability to predict a fixed number of differentially expressed genes in real or simulated data using quantitative measures based on t-statistics [4,11,13,21], adjusted p-values [11], and false-discovery rates [23]. However, there is uncertainty associated with these measures, and the true number of differentially expressed genes is unknown. Spike-in data have been used to assess normalization approaches for Affymetrix GeneChip data [14,24,25]. However, external control samples are not widely used for evaluation of normalization methods for cDNA microarrays.
In this paper, we evaluated normalization methods for cDNA microarray data using the k-NN LOOCV classification error (of biological samples characterized by the gene expression profiles), an alternative quantitative functional measure that is relatively unambiguous, objective and readily computed. We used k-NN classifiers because (i) their sensitivity enables us to discriminate between, and hence evaluate normalization techniques, (ii) they are readily available, (iii) they perform well in practice, and (iv) their non-parametric nature means that assumptions about how the data are distributed have little influence on classification performance. Since the primary aim of our evaluation of normalization methods was to assist practitioners in choosing effective data pre-processing schemes, we did not consider factors that may influence classification performance, such as feature selection and distance metrics. We investigated a wide spectrum of well-known and widely available normalization techniques: ten location normalization methods that adjust spatial effect and/or intensity effect (Table 1), and three scale methods that adjust scale differences (Table 3). We applied these methods, individually and in combination (41 strategies in all, Tables 1, 2, 3), to five diverse, published, cancer biology-related cDNA microarray data sets (Table 4), and we generated data sets with spatial effect, intensity effect and scale differences removed to varying degrees. Computing the LOOCV classification error of k-NNs estimated from these multi- and two-class data sets allowed us to investigate which and how much of the dye effect are removed by the 41 strategies.
Table 1 Single-bias-removal location normalization techniques used in this study. These strategies remove spatial- or intensity effect in a single step. The abbreviations are as follows, (for a given microarray), Ml: location-normalized log ratio; median(M): median value of non-normalized log ratios; lowess(rloci, cloci): lowess curve fitted as a function of the row location (rloci) and column location (cloci) of spots in PT group i; median(Mw): median value of non-normalized log ratios within the window size determined by w; lowess(A): lowess curve fitted to an MA plot of spots on a microarray; lowess(Ai): lowess curve fitted to an MA plot of spots in PT group i; spline(Aiset): spline curve fitted to an MA plot of spots in the invariant set, iset; Rl: location-normalized R value; qspline(Gi): qspline smoothing using geometric mean of the G channels of all arrays as a target array; Gl: location-normalized G value; qspline(Rt): qspline smoothing using geometric mean of the R channels of all arrays as a target array.
Name * Description: Effect/Level Bioconductor R package/function(parameters)
NONRM No normalization Ml = M marray/maNorm(norm="none")
GMEDIAN Global Ml = M - median(M) marray/maNorm (norm="median", subset = T)
SLLOESS Spatial/local lowess Ml = M - loess(rloci, cloci) marray/maNormMain (f.loc = list(maNorm2D(g="maPrintTip", subset = T, span = 0.4))
SLFILTERW3 Spatial/Local median filter
Ml = M - median(Mw), W = 3 × 3 tRMA/SpatiallyNormalise** (M, width = 3, height = 3)
SLFILTERW7 Spatial/Local median filter
Ml = M - median(Mw), W = 7 × 7 tRMA/SpatiallyNormalise** (M, width = 7, height = 7)
IGLOESS Intensity/Global lowess Ml = M - loess(A) marray/maNorm (norm="loess", subset = TRUE, span = 0.4)
ILLOESS Intensity/Local lowess Ml = M - loess(Ai) marray/maNorm (norm="printTipLoess", subset = T, span = 0.4)
ISTSPLINE Intensity/Global spline Ml = M - spline(Aiset) affy/normalize.invariantset**(prd.td = c(0.003, 0.007))
QSPLINEG Intensity/Global qspline
Rl = R - qspline(Gt), Gl = G - qspline(Gt), Ml = log(Rl / Gl) affy/Rl ← normalize.qspline(R, 2^rowMeans(log2(G), na.rm = T), na.rm = T, *default*)
Gl ← normalize.qspline(G, 2^rowMeans(log2(G), na.rm = T), na.rm = T, *default*)
QSPLINER Intensity/Global qspline
Rl = R - qspline(Rt), Gl = G - qspline(Rt), Ml = log(Rl / Gl) affy/ Rl ← normalize.qspline(R, 2^rowMeans(log2(R), na.rm = T), na.rm = T, *default*)
Gl ← normalize.qspline(G, 2^rowMeans(log2(R), na.rm = T), na.rm = T, *default*)
* We adopted the terminology given in the table to avoid confusion within this work. Elsewhere, these methods are known as: GMEDIAN, global or median [4]; SLLOESS, 2D spatial [12]; SLFILTERW3, spatial normalization using median filter of the block size 3 × 3 [17]; SLFILTERW7, spatial normalization using median filter of the block size 7 × 7 [17]; IGLOESS, global loess [4, 26]; ILLOESS, print-tip loess [4]; ISTSPLINE, invariant set normalization [38]; QSPLINER, qspline using geometric mean of the R channels of all arrays as the target array [13]; QSPLINEG, qspline using geometric mean of the G channels of all arrays as the target array [13].
** The SpatiallyNormalise function in the tRMA package was modified to remove scale normalization. The normalize.invariantset function in Affy package was modified so that the function could be applied on cDNA microarray data.
*default* The default parameters for QSPLINEG and QSPLINER are (fit.iters = 5, min.offset = 5, spline.method="natural", smooth = T, spar = 0, p.min = 0, p.max = 1.0, incl.ends = T, converge = F)
Table 2 Double-bias-removal location normalization techniques used in this study. These strategies remove both spatial- and intensity effect either in a single step (IGSGLOESS) or in two steps (the remaining thirteen approaches) by combining methods listed in Table 1.
Name Description: Method/Effect/Level
IGSGLOESS* Joint Intensity/Global & Spatial/Global Ml = M - lowess(A, rloc, cloc)
IGLOESS-SLLOESS Step 1: IGLOESS/Intensity/Global lowess
Step 2: SLLOESS/Spatial/Local lowess
ILLOESS-SLLOESS Step 1: ILLOESS/Intensity/Local lowess
Step 2: SLLOESS/Spatial/Local lowess
IGLOESS-SLFILTERW3 Step 1: IGLOESS/Intensity/Global lowess
Step 2: SLFILTERW3/Spatial/Local median filter
IGLOESS-SLFILTERW7 Step 1: IGLOESS/Intensity/Global lowess
Step 2: SLFILTERW7/Spatial/Local median filter
ISTSPLINE-SLLOESS Step 1: ISTSPLINE/Intensity/Global spline
Step 2: SLLOESS/Spatial/Local lowess
ISTSPLINE-SLFILTERW3 Step 1: ISTSPLINE/Intensity/ Global spline
Step 2: SLFILTERW3/Spatial/Local median filter
ISTSPLINE-SLFILTERW7 Step 1: ISTSPLINE/Intensity/Global spline
Step 2: SLFILTERW7/Spatial/Local median filter
QSPLINEG-SLLOESS Step 1: QSPLINEG/Intensity/Global qspline
Step 2: SLLOESS/Spatial/Local lowess
QSPLINEG-SLFILTERW3 Step 1: QSPLINEG/Intensity/Global qspline
Step 2: SLFILTERW3/Spatial/Local median filter
QSPLINEG-SLFILTERW7 Step 1: QSPLINEG/Intensity/Global qspline
Step 2: SLFILTERW7/Spatial/Local median filter
QSPLINER-SLLOESS Step 1: QSPLINER/Intensity/Global qspline
Step 2: SLLOESS/Spatial/Local lowess
QSPLINER-SLFILTERW3 Step 1: QSPLINER/Intensity/Global qspline
Step 2: SLFILTERW3/Spatial/Local median filter
QSPLINER-SLFILTERW7 Step 1: QSPLINER/Intensity/Global qspline
Step 2: SLFILTERW7/Spatial/Local median filter
* IGSGLOESS was implemented in the following package/function: MAANOVA R package/smooth (method="rlowess", f = 0.4, degree = 2). Elsewhere, IGSGLOESS is known as joint loess [21]. lowess(A, rloc, cloc): lowess curve fitted as a function of average log intensity (A), row location (rloc), and column location (cloc) of spots on a microarray.
Table 3 Extant scale normalization techniques used in this study. For a given microarray, if Ml is a location-normalized log ratio, then Ms is the scale-normalized log ratio, where Ms = Ml / s, and s is median absolute deviation from the median (MAD), a robust estimate of the scale of the data distribution. The remaining abbreviations are as follows, median(Ml): median value of Ml values of spots on all microarrays in a data set; : median value of Ml values of spots in PT group i on a microarray.
Name * Description Bioconductor R package/function (parameters)
WSCALE Within-microarray scale normalization
Ms = Ml / si marrayNorm/maNormScale (norm="printTipMAD", subset = T, geo = T, Mscale = T)
BSCALE Between-microarray scale normalization
s = median(Ml - median(Ml))
Ms = Ml / s marrayNorm/maNormScale (norm="globalMAD", subset = T), geo = T, Mscale = T)
WBSCALE Step 1: Within-microarray scale normalization
marrayNorm/maNormScale (norm="printTipMAD", subset = T, geo = T, Mscale = T)
Step 2: Between-microarray scale normalization
s = median(Ml - median(Ml))
marrayNorm/maNormScale (norm="globalMAD", subset = T, geo = T, Mscale = T)
* We adopted the terminology given in this table to avoid confusion within this work. Elsewhere, the methods are known as: WSCALE, within-print-tip-group scale normalization [4]; and BSCALE, between slide scale normalization [4, 15].
Table 4 The multi-class, cancer-biology related transcriptional profiling data sets analyzed in this work. For each of the five published studies, the fluorescent intensities, microarray images, and associated information were downloaded from the URLs indicated. The statistics refer to data sets produced after application of all pre-normalization data processing, location/scale normalization, and post-normalization data processing steps. The abbreviations are as follows, Microarrays: number of cDNA microarrays; Probes: number of probes; K: total number of categories to which a sample could be assigned; Samples and Class: number of samples in the specified pre-defined category; Configuration: configuration of a microarray using the convention described in Figure 1.
Data set name Description
LIVER CANCER [46] Microarrays: 181; Probes: 6,605; K = 2
Samples and Class: 76 normal; 105 tumor
Configuration: (ngr = 8, ngc = 4, nsr = 27, nsc = 28)
LYMPHOMA [47] Microarrays: 81; Probes: 6,850; K = 3
Samples and Class: 29 normal, 43 diffuse large B-cell lymphoma (DLBCL); 9 follicular lymphoma (FL)
Configuration: (ngr = 4, ngc = 4, nsr = 24, nsc = 24); (ngr = 8, ngc = 4, nsr = 24, nsc = 24)
RENAL CELL CARCINOMA [48] Microarrays: 38; Probes: 13,608; K = 4
Samples and Class: 3 normal; 26 clear cell carcinoma (CCC); 5 granular cell carcinoma (GCC);
4 papillary carcinoma (PC)
Configuration: (ngr = 8, ngc = 4, nsr = 27, nsc = 28)
GASTRIC CARCINOMA [49] Microarrays: 130; Probes: 15,541; K = 2
Samples and Class : 28 normal; 102 tumor
Configuration: (ngr = 12, ngc = 4, nsr = 30, nsc = 32); (ngr = 12, ngc = 4, nsr = 29, nsc = 32);
(ngr = 12, ngc = 4, nsr = 30, nsc = 30)
LUNG CANCER [50] Microarrays: 60; Probes: 20,601; K = 5
Samples and Class: 6 normal; 35 adenocarcinoma (AC); 11 squamous cell carcinoma (SCC);
4 large cell lung cancer (LCLC); 4 small cell lung cancer (SCLC)
Configuration: (ngr = 8, ngc = 4, nsr = 27, nsc = 28)
Results
Spatial- and intensity-dependent normalization
Diagnostic plots
We used diagnostic plots to examine the ability of different location normalization methods to remove spatial- and/or intensity effect (Tables 1 and 2). Figure 2 shows spatial plots for two specific LYMPHOMA microarrays normalized with four approaches designed to correct spatial effect (SLLOESS, SLFILTERW3, SLFILTERW7, IGSGLOESS). The non-normalized M values (NONRM) for microarray "5850" display global spatial effect (left-to-right, green-to-red pattern) whereas those for microarray "5938" exhibit local spatial effect (top-to-bottom, green-to-red pattern in each PT group). Removal of spatial effect should result in a "random" red and green pattern of M values. SLLOESS and SLFILTERW7 exhibit similar dye bias-removal abilities in that they both remove global spatial effect more effectively than local spatial effect. SLFILTERW3 removes both global and local dye effect effectively, largely because it uses a median filter of a small window size (3 × 3 spots) for normalization. IGSGLOESS removes most, but not all, global and local spatial effect (a strip of red spots on the right side of "5850" and on the bottom of the PT groups in the first row of "5938" remain). IGSGLOESS may not be as effective at removing dye effect as expected because, as the developers indicate, lowess curve construction uses the standardized spatial variables (rloc, cloc), which may not be appropriate for location variables [21].
Figure 2 Spatial plots of microarrays 5850 and 5938 in the Lymphoma data set. Spatial plots of microarrays 5850 and 5938 in the LYMPHOMA data set. The plots show the results before and after location normalization designed to remove spatial effect. The spatial plot is a spatial representation of spots on the microarray color-coded by their M values (marrayPlots/maImage(x="maM", subset = T)). Spots in white are spots flagged in the original microarray data (missing values). Rows depict non-normalized (NONRM), and normalized Ml values (SLLOESS, SLFILTERW3, SLFILTERW7, IGSGLOESS).
Figure 3 shows intensity-dependent MA plots for one specific LYMPHOMA microarray overlaid with one lowess curve (left) or one lowess curve per print tip group (right) using six methods designed to correct intensity effect (IGLOESS, ILLOESS, ISTSPLINE, QSPLINEG, QSPLINER, IGSGLOESS). For non-normalized M values (NONRM), the curvature in the MA plot indicates the presence of intensity effect at the array (left) and PT (right) level. All six methods remove global intensity effect completely (flat lowess curves, left), but only ILLOESS and IGSGLOESS remove local intensity effect thoroughly (right).
Figure 3 MA plots of microarray 5812 in the LYMPHOMA data set. The plots show the results before and after location normalization designed to remove intensity effect. The MA plot is a scatter plot of log ratio M = log2(Rf / Gf) (abscissa) versus average log intensity (ordinate). Columns depict non-normalized (NONRM), and normalized Ml values (IGLOESS, ILLOESS, ISTSPLINE, QSPLINEG, QSPLINER, IGSGLOESS). Plots in the same row represent same data except that each plot in the left panel shows one lowess curve for all the spots (marrayPlots/maPlot(data, z = NULL)); while that in the right panel shows one lowess curve per PT group (marrayPlots/maPlot(x="maA", y="maM", z="maPrintTip")). Different colors and line types are used to represent different groups from different rows ("ngr", Figure 1) and columns ("ngc") respectively.
Visual inspection of the diagnostic plots in Figures 2 and 3 suggest that SLFILTERW3 is an effective method for removing both global and local spatial effect, whereas ILLOESS is good at removing intensity effect.
k-NN LOOCV Classification error
For a functional, quantitative evaluation of location normalization methods, we first computed k-NN LOOCV classification error rates for data normalized using these methods individually and/or in combination. Then for each data set, we ranked the normalization methods based on their LOOCV classification error rates. The smaller the LOOCV classification error rate, the lower the rank of the normalization strategy. In order to assess whether normalization is beneficial (or not), we also computed the following quantity for a normalization method in each data set:
IMPROVEMENT = (ErrorRate(NONRM) - ErrorRate(Method)) / ErrorRate(NONRM) × 100%
where ErrorRate(NONRM) is the error rate of NONRM, and ErrorRate(Method) is the error rate of the method. Tables 5 and 6 give results for five data sets (Table 4) and 23 location methods designed to remove spatial- and/or intensity effect (Tables 1 and 2). Figures 4 and 5 are alternative, visual representations of the classification "Error Rate" and "Rank" in Table 5.
Table 5 Leave-one-out cross-validation k-NN error rates for location normalized data. For each data set, the normalization methods were ranked based on their LOOCV classification error rates ("Rank"). The smaller the LOOCV classification error rate, the lower the rank. The methods are arranged in the following order: single-bias-removal methods (block 1), double-bias-removal methods (block 2) and the qspline-related methods (block 3). For a given data set, the smallest error rate(s) and rank(s) are shown in bold. The methods and data sets are described in Tables 1, 2 and 4, respectively.
Location Normalization method LIVER CANCER LYMPHOMA RENAL CELL CARCINOMA GASTRIC CARCINOMA LUNG CANCER
Error Rate Rank Error Rate Rank Error Rate Rank Error Rate Rank Error Rate Rank
NONRM 0.202 24 0.266 23 0.237 24 0.0347 24 0.359 23.5
GMEDIAN 0.163 21 0.247 21 0.158 22 0.0270 23 0.342 20.5
SLLOESS 0.136 9.5 0.272 24 0.132 16.5 0.0154 12 0.350 22
SLFILTERW3 0.155 16 0.216 20 0.132 16.5 0.0190 14 0.359 23.5
SLFILTERW7 0.144 12.5 0.253 22 0.132 16.5 0.0228 16 0.325 17.5
IGLOESS 0.133 8 0.186 15.5 0.132 16.5 0.0231 20 0.342 20.5
ILLOESS 0.110 2 0.154 13 0.132 16.5 0.0231 20 0.275 12.5
ISTSPLINE 0.129 7 0.177 14 0.114 7 0.0153 10 0.334 19
IGSGLOESS 0.136 9.5 0.130 10 0.132 16.5 0 2 0.283 15
IGLOESS-SLLOESS 0.113 3.5 0.117 6.5 0.119 10 0.0154 12 0.242 8.5
ILLOESS-SLLOESS 0.105 1 0.111 4 0.132 16.5 0.0193 15 0.267 10
IGLOESS-LFILTERW3 0.158 19.5 0.136 11 0.092 1.5 0.0231 20 0.242 8.5
IGLOESS-SLFILTERW7 0.113 3.5 0.111 4 0.119 10 0.0154 12 0.217 4
ISTSPLINE-SLLOESS 0.121 6 0.102 1 0.119 10 0.0233 22 0.192 1
ISTSPLINE-SLFILTERW3 0.157 18 0.139 12 0.092 1.5 0.0229 17.5 0.209 2.5
IISTSPLINE-SLFILTERW7 0.118 5 0.127 9 0.132 16.5 0.0229 17.5 0.209 2.5
QSPLINEG 0.158 19.5 0.192 17.5 0.096 3.5 0 2 0.275 12.5
QSPLINER 0.166 22 0.123 8 0.096 3.5 0 2 0.275 12.5
QSPLINEG-SLLOESS 0.138 11 0.198 19 0.119 10 0.00769 7.5 0.225 6
QSPLINEG-SLFILTERW3 0.144 12.5 0.186 15.5 0.172 23 0.00758 5 0.317 16
QSPLINEG-SLFILTERW7 0.149 14 0.192 17.5 0.106 6 0.00758 5 0.225 6
QSPLINER-SLLOESS 0.155 16 0.105 2 0.105 5 0.00769 7.5 0.225 6
QSPLINER-SLFILTERW3 0.155 16 0.111 4 0.145 21 0.00758 5 0.325 17.5
QSPLINER-SLFILTERW7 0.169 23 0.117 6.5 0.119 10 0.0114 9 0.275 12.5
Table 6 IMPROVEMENT of location normalization methods. IMPROVEMENT is defined (in the Results) based on improvement of LOOCV classification error rate of a given normalization method over that of NONRM. The methods are arranged in the same order as those in Table 5. For a given data set, the biggest IMPROVEMENT(s) is shown in bold. The methods and data sets are described in Tables 1, 2 and 4, respectively.
Location Normalization method IMPROVEMENT (%, LIVER CANCER) IMPROVEMENT (%, LYMPHOMA) IMPROVEMENT (%, RENAL CELL CARCINOMA) IMPROVEMENT (%, GASTRIC CARCINOMA) IMPROVEMENT (%, LUNG CANCER) IMPROVEMENT RANGE (%)
NONRM 0 0 0 0 0 0 - 0
GMEDIAN 19 7 33 22 5 5 – 33
SLLOESS 33 -2 44 56 3 -2 – 56
SLFILTERW3 23 19 44 45 0 0 – 45
SLFILTERW7 29 5 44 34 9 5 – 44
IGLOESS 34 30 44 33 5 5 – 44
ILLOESS 46 42 44 33 23 23 – 46
ISTSPLINE 36 33 52 56 7 7 – 56
IGSGLOESS 33 51 44 100 21 21 – 100
IGLOESS-SLLOESS 44 56 50 56 33 33 – 56
ILLOESS-SLLOESS 48 58 44 44 26 26 – 58
IGLOESS-SLFILTERW3 22 49 61 33 33 22 – 61
IGLOESS-SLFILTERW7 44 58 50 56 40 40 – 58
ISTSPLINE-SLLOESS 40 62 50 33 47 33 – 62
ISTSPLINE-SLFILTERW3 22 48 61 34 42 22 – 61
IISTSPLINE-SLFILTERW7 42 52 44 34 42 34 – 52
QSPLINEG 22 28 59 100 23 22 – 100
QSPLINER 18 54 59 100 23 18 – 100
QSPLINEG-SLLOESS 32 26 50 78 37 26 – 78
QSPLINEG-SLFILTERW3 29 30 27 78 12 12 – 78
QSPLINEG-SLFILTERW7 26 28 55 78 37 26 – 78
QSPLINER-SLLOESS 23 61 56 78 37 23 – 78
QSPLINER-SLFILTERW3 23 58 39 78 9 9 – 78
QSPLINER-SLFILTERW7 16 56 50 67 23 16 – 67
Figure 4 Bar plots of leave-one-out cross-validation error rates for k-NNs in Table 5. The classifiers were estimated from five data sets (Table 4) either without normalization (NONRM) or normalized using twenty-three normalization techniques that remove spatial- and/or intensity effect to varying degrees (Tables 1 and 2). In each plot, the normalization methods are arranged in the following order: (A) Methods that remove no dye bias (GMEDIAN), or a single dye bias (SLLOESS, SLFILTERW3, SLFILTERW7, IGLOESS, ILLOESS, ISTSPLINE). (B) Methods that remove two dye biases (IGSGLOESS, IGLOESS-SLLOESS, ILLOESS-SLLOESS, IGLOESS-SLFILTERW3, IGLOESS-SLFILTERW7, ISTSPLINE-SLLOESS, ISTSPLINE-SLFILTERW3, ISTSPLINE-SLFILTERW7). (C) Qspline-related methods (QSPLINEG, QSPLINER, QSPLINEG-SLLOESS, QSPLINEG-SLFILTERW3, QSPLINEG-SLFILTERW7, QSPLINER-SLLOESS, QSPLINER-SLFILTERW3, QSPLINER-SLFILTERW7).
Figure 5 Rank summary for location normalization methods. The median and upper quantile ranks of each method are defined as the median and upper quantile values of the ranks of each method across all five data sets (see Table 5, "Ranks"). The bar plots present a visual depiction of the results in the table. (Median ranks are shown in pink; upper quantile ranks are shown in blue.)
Single-bias-removal methods
These strategies can be classified into two categories, spatial-dependent and intensity-dependent normalization methods. Three spatial-dependent normalization methods (SLLOESS, SLFILTERW3, SLFILTERW7) reduce k-NN LOOCV classification error rates to a similar extent (Tables 5 and 6) and have almost identical ranks (Figure 5), despite the fact that their abilities to remove spatial effect are quite different (Figure 2). Since both SLLOESS and SLFILTERW7 fail to remove local spatial patterns effectively (Figure 2, rows 2 and 4), SLFILTERW3 may be too aggressive in removing "dye effect" (Figure 2, row 3). However, the three intensity-dependent methods (IGLOESS, ILLOESS, ISTSPLINE) reduce k-NN LOOCV classification error rates to different degrees. The k-NN LOOCV classification error rate and rank of IGLOESS are similar to those of the three spatial-dependent methods (SLLOESS, SLFILTERW3, SLFILTERW7) (Figure 5), whereas ILLOESS, which removes intensity effect more completely than IGLOESS, has smaller k-NN LOOCV classification error rates than IGLOESS in all five data sets. ISTSPLINE, which uses a rank invariant set for normalization, is also better than IGLOESS in all five data sets (Figure 5).
In all five data sets, except for LYMPHOMA (SLLOESS), the single-bias-removal normalization methods consistently yield smaller LOOCV classification error rates than no-bias-removal methods, NONRM and GMEDIAN (which only sets the median of the distribution of M values to zero). The greatest benefit, an IMPROVEMENT of 56%, is seen with GASTRIC CARCINOMA (SLLOESS, ISTSPLINE) (Table 6).
Double-bias-removal methods
IGSGLOESS removes both spatial- and intensity effect in one step, whereas the remaining seven approaches are two-step strategies consisting of single-bias-removal methods applied sequentially (first a method to remove intensity effect, followed by a method to remove spatial effect).
In general, double-bias-removal methods have smaller k-NN LOOCV classification error rates and bigger IMPROVEMENT than single-bias-removal methods, and all perform better than NONRM and GMEDIAN (Tables 5 and 6, Figures 4 and 5). Using an arbitrary cut-off value of 10 for both median and upper quantile ranks (Figure 5), IGLOESS-SLFILTERW7, ISTSPLINE-SLLOESS and IGLOESS-SLLOESS (all of which remove intensity effect globally and then spatial effect locally) appear to be the best methods overall. These three two-step strategies not only have the lowest ranks amongst all normalization methods and across all data sets (Figure 5), they also showed most consistent and significant IMPROVEMENT over both NONRM and GMEDIAN across all five data sets (Table 6). The benefits of using IGLOESS-SLFILTERW7 over no normalization (NONRM) range from an IMPROVEMENT value of 40% in LUNG CANCER to 58% in LYMPHOMA (Table 6), whereas the IMPROVEMENT values of ISTSPLINE-SLLOESS range from 33% in GASTRIC CARCINOMA to 62% in LYMPHOMA and the IMPROVEMENT values of IGLOESS-SLLOESS range from 33% in LUNG CANCER to 56% in GASTRIC CARCINOMA.
The ranks of the SLFILTERW3-related approaches (IGLOESS-SLFILTERW3, ISTSPLINE-SLFILTERW3, QSPLINEG-SLFILTERW3, QSPLINER-SLFILTERW3) are higher than their SLFILTERW7 counterparts (Figure 5), suggesting that a window size of 7 × 7 is more preferable than that of 3 × 3. A smaller window size may over normalize the data, and thus conceal real biological variations.
Compared to the two-step approaches, the rank of the one-step method, IGSGLOESS, is higher than IGLOESS-SLFILTERW7 and ISTSPLINE-SLLOESS (yet lower than IGLOESS-SLFILTERW3 and ISTSPLINE-SLFILTERW3). This indicates that the one-step IGSGLOESS has no apparent advantage over the two-step bias-removal strategies.
Overall, the classification performances of data normalized using the double-bias-removal methods are better than that of NONRM, and the benefits (IMPROVEMENT) of doing so range from 21% in the case of LUNG CANCER (IGSGLOESS) to 100% in GASTRIC CARCINOMA (IGSGLOESS) (Table 6).
Qspline-related approaches
Unlike the location normalization methods discussed above, qspline-related approaches require a target array. QSPLINEG and QSPLINER are single-bias-removal techniques and use G and R respectively as the target array. The reduction in k-NN LOOCV classification error rates for these methods is quite significant compared to the other single-bias-removal methods. However, it is noticeable that although QSPLINEG and QSPLINER produce similar results in almost all data sets, their results are different in LYMPHOMA (Figures 4 and 5). In addition, when QSPLINEG or QSPLINER is combined with one of the three spatial-dependent methods, the rank of the resulting double-bias-removal technique is different from that of its counterpart technique (Figure 5). These results suggest that, similar to other baseline array-based normalization methods [14], the performances of the qSpline-related methods may also depend on the choice of the target array.
Overall, the classification performance of data normalized using the qspline-related methods is better than NONRM by IMPROVEMENT values of 9% in LUNG CANCER (QSPLINER-SLFILTERW3) and of 100% in GASTRIC CARCINOMA (QSPLINEG, QSPLINER). None of these qSpline-related methods, however, outperforms the IGLOESS-SLFILTERW7 (Table 6).
Scale normalization
Figure 6 shows boxplots of the distribution of non-normalized M values for microarrays in the five studies. Scale effect is more apparent between (right) rather than within (left) microarrays in a study. The LYMPHOMA data set shows considerable variations in box size and whisker length both within and between microarrays.
Figure 6 Boxplots of the distributions of non-normalized M values for microarrays in the five studies. In each boxplot, the box depicts the main body of the data and the whiskers show extreme values. The variability is indicated by the size of the box and the length of the whiskers (marray/marraymaBoxplot(y="maM")). Each panel in the left-hand column shows results for M values at the local level of a microarray chosen at random from a given data set. The bars are color-coded by PT group. Each panel in the right-hand column shows results for M values at the global level for 50 microarrays chosen at random from a given data set (the total number of microarrays in RENAL CELL CARCINOMA is 38). Each row corresponds to a particular study.
Tables 7 and 8 and Figure 7 show LOOCV classification error rates, ranks and IMPROVEMENT for the k-NN classifiers estimated using 3 scale normalization methods combined with other spatial- and/or intensity-dependent normalization methods (18 strategies in all). For data normalized first with spatial- and/or intensity-dependent methods, little or no reduction in LOOCV classification error rates was observed when within-microarray scale normalization (WSCALE) was applied later. However, when between-microarray scale normalization (BSCALE) was used alone, or when both scale normalization techniques were used sequentially (WBSCALE), there was an increase in both median and upper quantile ranks (Figure 7), suggesting that BSCALE should not be applied on the studied data sets. With regard to our running example, the LYMPHOMA data set, scale normalization has no apparent beneficial effect on classification performance.
Table 7 Leave-one-out cross-validation k-NN error rates for scale normalized data. Error rate and rank of each scale normalization method. "Rank" is described in detail in Table 5. For a given data set, the smallest error rate(s) and rank(s) are shown in bold. The methods and data sets are described in Tables 3 and 4, respectively.
Location, Scale Normalization method LIVER CANCER LYMPHOMA RENAL CELL CARCINOMA GASTRIC CARCINOMA LUNG CANCER
Error Rate Rank Error Rate Rank Error Rate Rank Error Rate Rank Error Rate Rank
NONRM 0.202 19.5 0.266 18.5 0.237 24 0.0347 14.5 0.359 19.5
NONRM, WSCALE 0.185 14.5 0.303 23.5 0.211 22.5 0.0270 12 0.350 17.5
NONRM, BSCALE 0.227 23.5 0.272 20.5 0.132 8 0.0615 22 0.425 24
NONRM, WBSCALE 0.227 23.5 0.303 23.5 0.132 8 0.0462 21 0.392 23
SLLOESS 0.136 9 0.272 20.5 0.132 8 0.0154 3 0.350 17.5
SLLOESS, WSCALE 0.127 7 0.266 18.5 0.132 8 0.0193 6 0.342 15.5
SLLOESS, BSCALE 0.202 19.5 0.260 17 0.145 14.5 0.1000 24 0.300 11.5
SLLOESS, WBSCALE 0.191 16 0.284 22 0.119 2 0.0846 23 0.283 9.5
IGLOESS 0.133 8 0.186 15 0.132 8 0.0231 9.5 0.342 15.5
IGLOESS, WSCALE 0.141 10 0.148 11 0.132 8 0.0193 6 0.317 13.5
IGLOESS, BSCALE 0.215 22 0.216 16 0.158 18.5 0.0308 13 0.375 21
IGLOESS, WBSCALE 0.202 19.5 0.179 14 0.158 18.5 0.0231 9.5 0.383 22
ILLOESS 0.110 3.5 0.154 12 0.132 8 0.0231 9.5 0.275 7
ILLOESS, WSCALE 0.116 6 0.161 13 0.145 14.5 0.0231 9.5 0.300 11.5
ILLOESS, BSCALE 0.193 17 0.111 3 0.145 14.5 0.0385 17.5 0.359 19.5
ILLOESS, WBSCALE 0.202 19.5 0.105 1 0.145 14.5 0.0424 20 0.317 13.5
ILLOESS-SLLOESS 0.105 1.5 0.111 3 0.132 8 0.0193 6 0.267 4.5
ILLOESS-SLLOESS, WSCALE 0.105 1.5 0.136 9.5 0.132 8 0.00769 1 0.267 4.5
ILLOESS-SLLOESS, BSCALE 0.160 12 0.130 8 0.211 22.5 0.0385 17.5 0.275 7
ILLOESS-SLLOESS, WBSCALE 0.152 11 0.123 6 0.158 18.5 0.0347 14.5 0.259 3
IGLOESS-SLFILTERW7 0.113 5 0.111 3 0.119 2 0.0154 3 0.217 2
IGLOESS-SLFILTERW7, WSCALE 0.110 3.5 0.124 7 0.119 2 0.0154 3 0.209 1
IGLOESS-SLFILTERW7, BSCALE 0.180 13 0.117 5 0.171 21 0.0385 17.5 0.275 7
IGLOESS-SLFILTERW7, WBSCALE 0.185 14.5 0.136 9.5 0.158 18.5 0.0385 17.5 0.283 9.5
Table 8 IMPROVEMENT of the scale normalization methods. IMPROVEMENT is described in detail in Table 6. For a given data set, the biggest IMPROVEMENT(s) is shown in bold. The methods and data sets are described in Tables 3 and 4, respectively.
Location, Scale Normalization method IMPROVEMENT (%, LIVER CANCER) IMPROVEMENT (%, LYMPHOMA) IMPROVEMENT (%, RENAL CELL CARCINOMA) IMPROVEMENT (%, GASTRIC CARCINOMA) IMPROVEMENT (%, LUNG CANCER)
NONRM 0 0 0 0 0
NONRM, WSCALE 8 -13 11 22 3
NONRM, BSCALE -12 -2 44 -77 -18
NONRM, WBSCALE -12 -13 44 -33 -9
SLLOESS 33 -2 44 56 3
SLLOESS, WSCALE 37 0 44 44 5
SLLOESS, BSCALE 0 2 39 -188 16
SLLOESS, WBSCALE 5 -7 50 -144 21
IGLOESS 34 30 44 33 5
IGLOESS, WSCALE 30 44 44 44 12
IGLOESS, BSCALE -6 19 33 11 -4
IGLOESS, WBSCALE 0 33 33 33 -7
ILLOESS 45 42 44 33 23
ILLOESS, WSCALE 43 39 39 33 16
ILLOESS, BSCALE 4 58 39 -11 0
ILLOESS, WBSCALE 0 61 39 -22 12
ILLOESS-SLLOESS 48 58 44 44 26
ILLOESS-SLLOESS, WSCALE 48 49 44 78 26
ILLOESS-SLLOESS, BSCALE 21 51 11 -11 23
ILLOESS-SLLOESS, WBSCALE 25 54 33 0 28
IGLOESS-SLFILTERW7 44 58 50 56 40
IGLOESS-SLFILTERW7, WSCALE 46 53 50 56 42
IGLOESS-SLFILTERW7, BSCALE 11 56 28 -11 23
IGLOESS-SLFILTERW7, WBSCALE 8 49 33 -11 21
Figure 7 Rank summary for scale normalization methods. The median ranks and upper quantile ranks are defined as described in Figure 5. The bar plots present a visual depiction of the results in the table. (Mean ranks are shown in pink; median ranks are shown in blue.) In each plot, normalization strategies are arranged in the following order: a location normalization method, a location normalization method followed by WSCALE (+WSCALE), a location normalization method followed by BSCALE (+BSCALE), a location normalization method followed by WBSCALE (+WBSCALE).
Discussion
This computational investigation employed two types of visual diagnostic plots and k-NN LOOCV classification error rates to evaluate a broad suite of known normalization strategies. These analyses were applied to cDNA microarray data from five published cancer studies. Since all these data sets were acquired using GenePix image analysis software and a recent study showed that background adjustment using GenePix can increase variability of microarray data and compromise downstream data analyses [3], we used foreground intensity values of the probes without background adjustment in this work. The normalization approaches examined are based on a variety of different techniques and implementations that are readily available and accessible.
Our results show that the LOOCV classification error of k-NN classifiers depends on how much of spatial- and intensity effect can be removed by a normalization strategy. Overall, the single-bias-removal location approaches perform better than GMEDIAN and NONRM, while the double-bias-removal location strategies perform better than the single-bias-removal location approaches. Of the twenty-three location normalization techniques investigated, three two-step processes (IGLOESS-SLFILTERW7, ISTSPLINE-SLLOESS and IGLOESS-SLLOESS), all of which removes intensity effect at the global level and spatial effect at the local level, appear to be the most effective at reducing LOOCV classification error. However, removing spatial- or intensity effect alone is not sufficient for reducing LOOCV classification error (see below).
A recent review of normalization methods [26] raised the concern that removing spatial effect (SLLOESS and the related methods) may add additional noise to normalized data, and suggested that a safe alternative was removing only intensity effect at the local level (ILLOESS) [26]. Our results show that, although the classification performance of data normalized with SLLOESS alone can be worse than non-normalized data as in the case of the LYMPHOMA data set, when SLLOESS is combined with another intensity-dependent approach (IGLOESS, ILLOESS, ISTSPLINE, QSPLINEG, or QSPLINER), there is considerable improvement over NONRM, with IMPROVEMENT ranging from 23% in LIVER CANCER (QSPLINER-SLLOESS) to 78% in GASTRIC CARCINOMA (QSPLINER-SLLOESS, QSPLINEG-SLLOESS). Thus, removing both spatial- and intensity effect is beneficial for the downstream analytical task of classification. Another study compared various lowess-based single-bias-removal intensity normalization approaches, and found that ILLOESS may not significantly improve the results compared to IGLOESS [27]. Our results show that the benefits (IMPROVEMENT) of IGLOESS over NONRM range from 5% in LUNG CANCER to 44% in RENAL CELL CARCINOMA; while that the benefits (IMPROVEMENT) of ILLOESS over NONRM range from 23% in RENAL CELL CARCINOMA to 46% in LIVER CANCER. Therefore, ILLOESS performs better than IGLOESS in our study. However, as a single-bias-removal approach, ILLOESS still fail to outperform IGLOESS-SLFILTERW7, ISTSPLINE-SLLOESS and IGLOESS-SLLOESS, which are the best overall methods and whose IMPROVEMENT values over NONRM range from 40% in LUNG CANCER to 58% in LYMPHOMA for IGLOESS-SLFILTERW7, from 33% in GASTRIC CARCINOMA to 62% in LYMPHOMA for ISTSPLINE-SLLOESS and from 33% in LUNG CANCER to 56% in GASTRIC CARCINOMA for IGLOESS-SLLOESS (Table 6).
A previous study employed k-NN classification of diluted samples to assess a small number of global linear methods for normalization [28]. The study presented here is more comprehensive, both in terms of the range of data sets and the diversity of normalization techniques. Our results indicate that the k-NN LOOCV classification error of real biological samples provides an informative functional quantitative measure that can be used to evaluate normalization approaches.
Differences in scale between microarrays can arise from both unwanted technical factors (differences in experimental reagents, equipment, personnel, and so on), as well as from genuine biological variations. The scale normalization techniques applied here aim to remove unwanted technical factors, and assume the existence of little biological variations between samples. For the five studied data sets, scale normalization of non- or location-normalized data do not result in an overall reduction in LOOCV classification error. Indeed, two between-microarray normalization methods (BSCALE, WBSCALE) result in an overall increase in LOOCV classification error (poorer performance, Figure 7). These results suggest that in the examined cancer-related data sets, there can be considerable genuine biological variations (which is plausible because genomic aberrations found in cancer cells [29,30] may alter the number and nature of expressed genes compared to normal cells), and that these variations are masked by the applied scale normalization. The data sets considered here do not contain replicated data, so it is difficult to ascertain how much of the scale effect result from unwanted technical factors. Scale normalization may be warranted in situations where technical differences can be discerned by examination of the replicated data and genuine biological variations are known or believed to exist. In such cases, scale normalization using external control samples may be more useful than the total gene approaches.
While our empirical analyses are thoroughgoing in terms of both normalization procedures and test data sets, we acknowledge that there are two caveats in this study that deserve attention and further investigation. First, we employed the LOOCV classification error as a functional measure to assess normalization methods. In principle, LOOCV provides an almost unbiased estimate of the generalization ability of a classifier [31], especially when the number of the available training samples is severely limited (as in the case of LYMPHOMA and RENAL CELL CARCINOMA in Table 4), and is thus highly desirable for model selection or other relevant algorithm evaluation [32,33]. However, it is also known that the LOOCV error estimator may have high variance in some situations [34,35], which could in turn affect the accuracy of the rankings of the normalization methods. Empirically, however, we found that the LOOCV errors we obtained from various round of classification are quite stable, therefore we believe that our estimation is in practice reliable and suitable for ranking. Nevertheless, error estimators that have shown to have low variance (e.g., bootstrapping and k-fold cross-validation [34,35]) are worth further investigation in the future.
The second caveat of this work is that normalization methods were evaluated using k-NN classification without the aid of auxiliary techniques, such as feature selection. The reasons we did not employ feature selection, but rather used all the probes that are present in the majority of the microarrays for classification are as follow: i) We believe that the influence of the dye effect (which usually affect a large number of the probes) on the downstream data analysis can be better and more consistently reflected when a large number of the probes are examined. As such, using all valid probes for training a classifier can best reflect the effectiveness of the normalization methods, whereas using subsets of the probes may generate inconsistent results due to the heterogeneous nature of the dye effect across microarrays. ii) We also included low intensity probes in the analyses. Although this may add noise and therefore could compromise the absolute classification performance of the examined normalization methods, we nevertheless think that these probes should not be excluded because reducing variability in low intensity probes is by itself an important objective of normalization methods. That is, a good normalization approach should be able to reduce variability in both low intensity- and high intensity probes effectively. And iii) we are aware that k-NNs without feature selection may add variability to the classification results, however, k-NN classification is also appealing in that it is simple and requires no data pre-processing or assumption on data distribution. In addition, k-NN classifiers have been widely used in many classification tasks including high-dimensional problems arising from image and text data [36].
Due to the above two caveats, the relative rankings of the investigated normalization strategies can hardly be obtained accurately in this work. For example, our results show that IGLOESS-SLFILTERW7, ISTSPLINE-SLLOESS and IGLOESS-SLLOESS reduce LOOCV classification errors most consistently and effectively across all five data sets. It is difficult, however, to determine further which of these three strategies is the best, because small differences in classification results can either arise from inherent differences in these approaches and/or from the variability introduced by the LOOCV error estimator and less optimal k-NN classifiers. Moreover, our results should not be taken as a warrant of directly using baseline methods, such as k-NNs without feature selection, for high-dimensional classification tasks. More investigations are needed to understand the interplay between normalization (which improves data quality) and feature selection (which improves the classifier by throwing away non-informative data) to ascertain normalization strategies to produce an optimal classifier.
Conclusion
Using LOOCV error of k-NNs as the evaluation criterion, we assessed a variety of normalization methods that remove spatial effect, intensity effect and scale differences from cDNA microarray data. Overall, the single-bias-removal location approaches (which remove either spatial- or intensity effect from the data) perform better than GMEDIAN and NONRM, while the double-bias-removal location strategies (which remove both spatial- and intensity effect) perform better than the single-bias-removal location approaches. Of the 41 different strategies examined, IGLOESS-SLFILTERW7, ISTSPLINE-SLLOESS and IGLOESS-SLBSCALE, all of which are two-step approaches and remove both intensity effect at the global level and spatial effect at the local level, appear to be the most effective at reducing LOOCV classification error. The investigated scale normalization methods do not have beneficial effect on classification performance. These results also indicate that spatial- and intensity effect do have profound impact on downstream data analyses, such as classification, and that removing these effects can improve the quality of such analyses.
Methods
Extant data sets and software
Table 4 summarizes relevant information on the cDNA microarray data sets from the Stanford Microarray Database (SMD) reexamined here. These data sets were selected because the published studies assayed samples from distinct cancers, the profiling experiments were performed at different times, four out of the five data sets were produced by different investigators, and the microarrays used were printed with different probes on different occasions. The LYMPHOMA study has been used as the illustrative, running example.
A variety of computational tools for manipulating, analyzing, and visualizing microarray data are available. These include open source implementations based on the R language for statistical computing [37] such as the Bioconductor , MAANOVA , tRMA , and braju packages. Standard R and Bioconductor packages and functions were used apart from one normalization method found in MAANOVA (joint removal of both spatial- and intensity effect at the global microarray level, IGSGBSCALE) and two normalization methods found in tRMA (removal of spatial effect at the local PT level, SLFILTERW3 and SLFILTERW7).
Pre-normalization data processing
For each spot, the foreground red (Rf) and green (Gf) quantitated fluorescent intensities (acquired using GenePix image analysis software) of the arrayed DNA sequences were used to compute the non-normalized log ratio, M = log2(Rf / Gf), and average log intensity, . Because of the concern that local background values estimated by GenePix may add additional noise to the data [3], these values were not subtracted from their corresponding foreground values. For a given microarray, the log ratios were normalized using location- and/or scale-normalization techniques and its particular configuration (the LYMPHOMA and GASTRIC CARCINOMA studies employed microarrays with two and three distinct configurations respectively).
Normalization methods
Tables 1 and 2 summarize the 23 location normalization methods that remove none, one, or both of spatial- and intensity effect (detailed descriptions of how they adjust M values can be found elsewhere [4,6,12,13,17,21,38,39]). In particular, Table 1 includes two methods that remove no spatial- or intensity-dependent dye bias: (i) NONRM neither removes any effect nor alters the distribution of M values; and (ii) GMEDIAN does not remove any effect but acts as a baseline normalization method because it sets the mean or median of M to zero. There are eight methods that remove either spatial- or intensity effect: (i) SLBSCALE removes spatial effect at the PT level using lowess; (ii) SLFILTERW3 and SLFILTERW7 remove spatial effect using median filters of the block sizes 3 × 3 and 7 × 7, respectively [17]; (iii) IGBSCALE removes intensity effect at the global level; (iv) ILBSCALE removes intensity effect at the local level and as a byproduct removes spatial effect partially; (v) ISTSPLINE removes intensity effect at the global level using rank invariant set and a spline smoothing technique [38]; and (vi) QSPLINER (QSPLINEG) removes intensity effect at the global level using spline smoothing applied to quantiles obtained from R (G) and using the geometric mean of the R (G) channels of all arrays as the target array [13]. In Table 2, IGSGLOESS is a one-step process that removes global intensity effect and global spatial effect, while the remaining thirteen strategies are two-step processes that remove both dye effect by combining methods in Table 1.
Table 3 summarizes the three scale normalization methods used (detailed descriptions of how these methods adjust the scale of M values can be found elsewhere [4]). WSCALE adjusts the scale of M values at the PT level. BSCALE adjusts the scale of M values globally across all microarrays in a data set. WBSCALE adjusts the scale locally followed by globally, in two steps. These scale normalization methods were applied to non-normalized data (NONRM) and to data that had been normalized using the five location methods SLLOESS, IGLOESS, ILLOESS, ILLOESS-SLLOESS, or IGLOESS-SLFILTERW7. These methods were selected to represent methods that remove spatial and/or intensity effect at different levels.
Post-normalization data processing
For the five cancer-biology studies, examination of the published data indicated that probes printed on different microarrays (even those with the same configuration) were not necessarily identical. For the N microarrays associated with a given study (N can be equated with the value given for "Microarrays" in Table 4), the 41 data sets used to estimate k-NN classifiers and to determine their LOOCV classification errors were created as follows. Each microarray was handled as described in "Pre-normalization data processing" and the ensuing M values were normalized using the 41 distinct location and/or scale techniques discussed above. A probe was retained for further processing only if it was printed and present (i.e., successfully measured and computed) in 95% of the N microarrays. If a probe met these criteria, missing M values were imputed using the k-NNimpute algorithm [40] as implemented in the Bioconductor package/function pamr/pamr.knnimpute(k = 10) [41]. Given the 41 data sets, the M values for a probe in all N microarrays were centered and rescaled to a unit norm. For LYMPHOMA, the final dimensionality (number of probes after post-normalization data processing) of each of the N = 81 data points was 6,850 ("Probes"). The 41 post-normalized data sets for the five examined studies are available at .
Classification error
Given D data points, each of which is assigned to one of K categories (e.g., "normal", "DLBCL", "FL"), a LOOCV procedure for this K-class data set is as follows. The data set is partitioned into a test set of one data point and a learning set of D-1 data points. The learning set is used to train a classifier and the ensuing model is employed to predict the class label of the test data point. This process is repeated so that the class of each data point is predicted using a classifier estimated from all other data points in the data set. Classification error is the number of the instances in which the predicted class of a data point differs from its known class. The error rate is this value divided by the number of data points, D.
k-NN classifier
Given a K-class data set, the k-NN algorithm predicts the class label of a test data point by first finding which of the data points in the data set are its k closest neighbors. The classes of these k nearest neighbors are examined and the class of the test data point decided by a majority vote, with ties being broken at random. If there are ties for the kth nearest data point, all candidates are included in the vote. Classification using k-NNs does not require any special handling of multi-class data sets. A widely employed measure of the proximity of two data points and the one utilized here is the standardized Euclidean distance [42,43]. Since all probes are treated with equal weight, the classification results are affected by all the probes rather than just a subset, as would have been the case if feature selection had been employed. Euclidean distance has been shown to be effective and accurate on a variety of data sets [43,44].
The optimal number of nearest neighbors, k*, was determined via leave-one-out cross-validation. An original data set of D data points was partitioned into a test set of one data point and a learning set of D - 1 data points. Given a specific k, the k-NN algorithm was used to predict the class of each data point in the learning set using the D - 2 remaining data points. The classification error, εk, of the learning set was determined. This procedure was performed using k = {3,...,10} and k* taken to be the k producing the smallest classification error, i.e., mink(εk). The class of the test data point for the original data set was predicted using k* and the k-NN algorithm. This entire process was repeated such that each of the data points in the original data set was used as a test set. The classification error of the original data was calculated. The k-NN step was performed using the R class/package [45]class/knn.cv(k = "number of neighbors") where "number of neighbors" was set to 3,...,10. The prediction step was performed using class/knn(k = "optimal k") where "optimal k" was k*.
List of abbreviations
AC: adenocarcinoma;
CCC: clear cell carcinoma;
DLBCL: diffuse large B-cell lymphoma;
FL: follicular lymphoma;
GCC: granular cell carcinoma;
k-NN: k-nearest neighbor;
LCLC: large cell lung cancer;
LOOCV: Leave-one-out cross-validation;
lowess: local regression estimation;
PC: papillary carcinoma;
PT: print-tip group;
SCC: squamous cell carcinoma;
SCLC: small cell lung cancer;
SMD: Stanford Microarray Database.
Authors' contributions
WW designed and performed computational experiments, and drafted the manuscript. EPX participated in experimental design and in drafting the manuscript. ISM participated in experimental design and edited the manuscript. CM and MJB read and edited the manuscript. All authors contributed to, read and approved the final manuscript.
Acknowledgements
We thank three reviewers for their critical yet helpful comments. We thank Professor David Banks in the Institute of Statistics & Decision Sciences of Duke University for helpful discussion and suggestions. We also thank Aylin Rizki and Paraic A. Kenny for critical reading of the manuscript, and Kevin Peet and Penelope Siig for proofreading the manuscript. This work was supported by the Department of Energy (OBER DEAC0376SF00098), by NIH (grant CA64786 8744), and by an Innovator Award from the Department of Defense (DAMD17-02-1-0438) to M.J.B. Work by I.S.M. was supported by the California Breast Cancer Research Program, National Institute on Aging, National Institute of Environmental Health Sciences, and U.S. Department of Energy under contract No. DEAC0376SF00098.
==== Refs
Chen Y Dougherty ER Bittner ML Ratio-Based Decisions and the Quantitative Analysis of cDNA Microarray Images Journal of Biomedical Optics 1997 2 364 374
Yang YH Buckley MJ Speed TP Analysis of cDNA microarray images Brief Bioinform 2001 2 341 349 11808746
Qin LX Kerr KF Empirical evaluation of data transformations and ranking statistics for microarray analysis Nucleic Acids Res 2004 32 5471 5479 15479783
Yang YH Dudoit S Luu P Speed TP Bittner ML, Chen Y, Dorsel AN and Dougherty ER Normalization for cDNA microarray data: ; San Jose, California. Proceedings of SPIE 2001 4266 SPIE-International Society for Optical Engineering 141 152
Quackenbush J Microarray data normalization and transformation Nat Genet 2002 32 Suppl 2 496 501 12454644
Smyth GK Yang YH Speed TP Brownstein MJ and Khodursky AB Statistical issues in microarray data analysis Functional Genomics: Methods and Protocols Methods in Molecular Biology 2003 224 Totowa, NJ, Humana Press 111 136
Eisen MB Spellman PT Brown PO Botstein D Cluster analysis and display of genome-wide expression patterns Proc Natl Acad Sci U S A 1998 95 14863 14868 9843981
Zien A Aigner T Zimmer R Lengauer T Centralization: a new method for the normalization of gene expression data Bioinformatics 2001 17 323S 331
Kroll TC Wolfl S Ranking: a closer look on globalisation methods for normalisation of gene expression arrays Nucleic Acids Res 2002 30 e50 12034851
Cleveland WS Devlin SJ Locally Weighted Regression: An Approach to Regression Analysis by Local Fitting Journal of the American Statistical Association 1988 83 596 610
Dudoit S Yang YH Callow MJ Speed TP Statistical methods for identifying differentially expressed genes in replicated cDNA microarray experiments. 2000 , University of California, Berkeley
Dudoit S Yang YH Parmigiani G, Garrett ES, Irizarry RA and Zeger SL Bioconductor R packages for exploratory analysis and normalization of cDNA microarray data. The Analysis of Gene Expression Data: Methods and Software 2003 New York, Springer 73 101
Workman C Jensen LJ Jarmer H Berka R Gautier L Nielser HB Saxild HH Nielsen C Brunak S Knudsen S A new non-linear normalization method for reducing variability in DNA microarray experiments Genome Biol 2002 3 research0048 12225587
Bolstad BM Irizarry RA Astrand M Speed TP A comparison of normalization methods for high density oligonucleotide array data based on variance and bias Bioinformatics 2003 19 185 193 12538238
Yang YH Dudoit S Luu P Lin DM Peng V Ngai J Speed TP Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation Nucleic Acids Res 2002 30 e15 11842121
Tseng GC Oh MK Rohlin L Liao JC Wong WH Issues in cDNA microarray analysis: quality filtering, channel normalization, models of variations and assessment of gene effects Nucleic Acids Res 2001 29 2549 2557 11410663
Wilson DL Buckley MJ Helliwell CA Wilson IW New normalization methods for cDNA microarray data Bioinformatics 2003 19 1325 1332 12874043
Kepler TB Crosby L Morgan KT Normalization and analysis of DNA microarray data by self-consistency and local regression Genome Biol 2002 3 RESEARCH0037 12184811
Huang J Kuo HC Koroleva I Zhang CH Soares MB A Semi-linear Model for Normalization and Analysis of cDNA Microarray Data 2003 Iowa City, Statistics and Actuarial Science, The University of Iowa
Fan J Tam P Woude GV Ren Y Normalization and analysis of cDNA microarrays using within-array replications applied to neuroblastoma cell response to a cytokine Proc Natl Acad Sci U S A 2004 101 1135 1140 14739336
Cui X Kerr MK Churchill GA Transformations for cDNA Microarray Data Statistical Applications in Genetics and Molecular Biology 2003 2 Article 4 16646782
Fang Y Brass A Hoyle DC Hayes A Bashein A Oliver SG Waddington D Rattray M A model-based analysis of microarray experimental error and normalisation Nucl Acids Res 2003 31 e96 12907748
Wang X Hessner MJ Wu Y Pati N Ghosh S Quantitative quality control in microarray experiments and the application in data filtering, normalization and false positive rate prediction Bioinformatics 2003 19 1341 1347 12874045
Cope LM Irizarry RA Jaffee HA Wu Z Speed TP A benchmark for Affymetrix GeneChip expression measures Bioinformatics 2004 20 323 331 14960458
Irizarry R cope L Bioconductor Expression Assessment Tool for Affymetrix Oligonucleotide Arrays (affycomp) 2004
Smyth GK Speed TP Carter D Normalization of cDNA microarray data METHODS: Selecting Candidate Genes from DNA Array Screens: Application to Neuroscience
Park T Yi SG Kang SH Lee S Lee YS Simon R Evaluation of normalization methods for microarray data BMC Bioinformatics 2003 4 33 12950995
Schuchhardt J Beule D Malik A Wolski E Eickhoff H Lehrach H Herzel H Normalization strategies for cDNA microarrays Nucleic Acids Res 2000 28 E47 10773095
Gray JW Collins C Genome changes and gene expression in human solid tumors Carcinogenesis 2000 21 443 452 10688864
Albertson DG Collins C McCormick F Gray JW Chromosome aberrations in solid tumors Nat Genet 2003 34 369 376 12923544
Cawley GC Talbot NLC Efficient leave-one-out cross-validation of kernel fisher discriminant classifiers. Pattern Recognition 2003 36 2585 2592
Vapnik V Chapelle O Smola AJ, Bartlett PJ, Scholkopf B and Schuurmans D Bounds on Error Expectation for SVM Advances in Large Margin Classifiers (Neural Information Processing Series) 2000 , MIT Press 261 280
Chapelle O Vapnik V Bousquet O Mukherjee S Choosing multiple parameters for support vector machines Machine Learning 2002 46 131 1159
Braga-Neto U Dougherty ER Bolstered error estimation. Pattern Recognition 2004 37 1267 1281
McLachlan GJ Do K Ambroise C Analyzing Microarray Gene Expression Data 2004 , Wiley-Interscience
Liu T Moore A Gray A Efficient Exact k-NN and Nonparametric Classification in High Dimensions 2003
Team RDC R: A language and environment for statistical computing 2004 , R Foundation for Statistical Computing
Li C Hung Wong W Model-based analysis of oligonucleotide arrays: model validation, design issues and standard error application Genome Biol 2001 2 RESEARCH0032 11532216
Wu H Kerr MK Cui X Churchill GA Parmigiani G, Garrett ES, Irizarry RA and Zeger SL MAANOVA: a software package for the analysis of spotted cDNA microarray experiments The Analysis of Gene Expression Data: Methods and Software 2003 New York, Springer 313 341
Troyanskaya O Cantor M Sherlock G Brown P Hastie T Tibshirani R Botstein D Altman RB Missing value estimation methods for DNA microarrays Bioinformatics 2001 17 520 525 11395428
Hastie T Tibshirani R Narasimhan B Chu G Pam: prediction analysis for microarrays 2003
Olshen AB Jain AN Deriving quantitative conclusions from microarray expression data Bioinformatics 2002 18 961 970 12117794
Holmes CC Adams NM A probabilistic nearest neighbour method for statistical pattern recognition J Royal Statistical Soc B 2002 64 295 306
Mitchie D Spiegelhalter DJ Taylor CC Machine Learning, Neural and Statistical Classification 1994 , Ellis Horwood
Ripley B The class package in the VR bundle. 2003
Chen X Cheung ST So S Fan ST Barry C Higgins J Lai KM Ji J Dudoit S Ng IO Van De Rijn M Botstein D Brown PO Gene expression patterns in human liver cancers Mol Biol Cell 2002 13 1929 1939 12058060
Alizadeh AA Eisen MB Davis RE Ma C Lossos IS Rosenwald A Boldrick JC Sabet H Tran T Yu X Powell JI Yang L Marti GE Moore T Hudson JJ Lu L Lewis DB Tibshirani R Sherlock G Chan WC Greiner TC Weisenburger DD Armitage JO Warnke R Levy R Wilson W Grever MR Byrd JC Botstein D Brown PO Staudt LM Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling Nature 2000 403 503 511 10676951
Higgins JP Shinghal R Gill H Reese JH Terris M Cohen RJ Fero M Pollack JR van de Rijn M Brooks JD Gene expression patterns in renal cell carcinoma assessed by complementary DNA microarray Am J Pathol 2003 162 925 932 12598325
Chen X Leung SY Yuen ST Chu KM Ji J Li R Chan AS Law S Troyanskaya OG Wong J So S Botstein D Brown PO Variation in gene expression patterns in human gastric cancers Mol Biol Cell 2003 14 3208 3215 12925757
Garber ME Troyanskaya OG Schluens K Petersen S Thaesler Z Pacyna-Gengelbach M van de Rijn M Rosen GD Perou CM Whyte RI Altman RB Brown PO Botstein D Petersen I Diversity of gene expression in adenocarcinoma of the lung Proc Natl Acad Sci U S A 2001 98 13784 13789 11707590
|
16045803
|
PMC1201132
|
CC BY
|
2021-01-04 16:27:25
|
no
|
BMC Bioinformatics. 2005 Jul 26; 6:191
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-191
|
oa_comm
|
==== Front
BMC Cell BiolBMC Cell Biology1471-2121BioMed Central London 1471-2121-6-311611149110.1186/1471-2121-6-31Research ArticleGlycosaminoglycans modulate C6 glioma cell adhesion to extracellular matrix components and alter cell proliferation and cell migration Mendes de Aguiar Claudia Beatriz Nedel [email protected]ão-Soares Bruno [email protected] Marcio [email protected] Andréa Gonçalves [email protected] Departamento de Biologia Celular, Embriologia e Genética, Centro de Ciências Biológicas, Universidade Federal de Santa Catarina, Florianópolis, Brazil2 Faculdade de Medicina de Ribeirão Preto, Hospital das Clinicas de Ribeirão Preto, Universidade de São Paulo, Ribeirão Preto, Brazil2005 19 8 2005 6 31 31 13 5 2005 19 8 2005 Copyright © 2005 de Aguiar et al; licensee BioMed Central Ltd.2005de Aguiar et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Adhesion to extracellular matrix (ECM) components has been implicated in the proliferative and invasive properties of tumor cells. We investigated the ability of C6 glioma cells to attach to ECM components in vitro and described the regulatory role of glycosaminoglycans (GAGs) on their adhesion to the substrate, proliferation and migration.
Results
ECM proteins (type IV collagen, laminin and fibronectin) stimulate rat C6 glioma cell line adhesion in vitro, in a dose-dependent manner. The higher adhesion values were achieved with type IV collagen. Exogenous heparin or chondroitin sulfate impaired, in a dose-dependent manner the attachment of C6 glioma cell line to laminin and fibronectin, but not to type IV collagen. Dextran sulfate did not affect C6 adhesion to any ECM protein analyzed, indicating a specific role of GAGs in mediating glioma adhesion to laminin and fibronectin. GAGs and dextran sulfate did not induce C6 glioma detachment from any tested substrate suggesting specific effect in the initial step of cell adhesion. Furthermore, heparin and chondroitin sulfate impaired C6 cells proliferation on fibronectin, but not on type IV collagen or laminin. In contrast, both GAGs stimulate the glioma migration on laminin without effect on type IV collagen or fibronectin.
Conclusion
The results suggest that GAGs and proteoglycans regulate glioma cell adhesion to ECM proteins in specific manner leading to cell proliferation or cell migration, according to the ECM composition, thus modulating tumor cell properties.
==== Body
Background
Glioblastoma multiforme (GBM) is the most common malignant form of glioma, and resistant to therapeutic interventions, causing most patients to die within 1 year after diagnosis[1]. The rat C6 glioma cell line, originally produced by Wistar-Furth rats exposed to N,N'-nitroso-methylurea, is morphologically similar to GBM when injected into brain of rats and has been used as both in vivo and in vitro model for the study of this kind of tumor[2].
Extracellular matrix (ECM) components control many aspects of cell behavior, such as differentiation, proliferation, cell morphology and attachment[3]. It has been suggested that adhesion to ECM in vivo and in vitro is involved in the earlier steps of tumor progression, resulting in genomic instability which allows for the accumulation of multiple mutations[4]. Transformation of cells is frequently accompanied by a change in adhesive properties, often resulting in a decreased adhesion of these cells[4]. Moreover, glioma proliferation and migration can be modulated by adhesion to different substrates [5-8] and they have been suggested as mutually exclusive behavior in astrocytoma cell lines[9]. In particular, C6 glioma cells have been shown to secrete laminin-2[10], and upon nerve growth factor stimulation, tenascin and fibronectin[11]. Integrins, the receptors to ECM proteins, participate in and regulate cell-substrate adhesion[4]. Adhesion to ECM proteins could direct integrin signaling with subsequent activation of cytoplasmic tyrosine kinases and induction of downstream effector pathways[12]. It has been shown that C6 glioma cells express the well-characterized fibronectin (α5β1) and the multi-specific laminin, collagens and fibronectin (α3β1) receptors[13]. Nevertheless, the function of individual ECM components, their involvement in cell adhesion as well as the molecular processes mediating growth and invasion of malignant gliomas are not properly characterized.
Several experimental evidences suggest that cell surface proteoglycans are also involved in tumor growth and progression. Highly sulfated glycosaminoglycans (GAGs) and the sulfated polysaccharide fucoidan, stimulate melanoma cell invasion in vitro[14]. On the other hand, inhibition of GAGs' sulfation promoted by sodium chlorate treatment reduced the proliferation and adhesion of the leukemic cell line WeHi-3B[15] and C6 glioma cells[16]. Moreover, heparan sulfate proteoglycans have been suggested to play a role in cell adhesion as positive modulators of cell proliferation in Chinese hamster ovary cells[17].
In the present study we have evaluated the role of ECM components in the adhesion, of glioma cells in vitro. We report that the addition of the GAGs, heparin and chondroitin sulfate affected C6 cells adhesion to laminin and fibronectin in a specific way in contrast to type IV collagen. In contrast, C6 glioma detachment from these ECM proteins was not altered by GAGs. We also demonstrate an involvement of heparin and chondroitin sulfate in regulating glioma proliferation and migration on ECM proteins. The results suggest a role of GAGs and proteoglycans in regulating glioma adhesion to ECM components as well as proliferation and migration according to the ECM composition.
Results
The three ECM proteins analyzed, type IV collagen, laminin and fibronectin, induced C6 cell adhesion in a dose-dependent manner (Figure 1A). At the lower ECM concentration (2 μg/ml), C6 cells presented greater adhesion to type IV collagen that was significantly higher (160%) than control (absence of ECM). In contrast, 2 μg/ml of fibronectin or laminin did not display considerable effect on cell adhesion. However, at the concentration of 4 μg/ml, both type IV collagen and fibronectin significantly stimulated C6 cell adhesion (+171% and +133%, respectively) whereas laminin did not affect it. All three ECM proteins presented the maximum adhesion effect at 10 μg/ml, after which a plateau was reached. At this concentration, type IV collagen improved C6 cell adhesion in 203%, fibronectin in 193%, and laminin in 105%. Laminin was the less adhesive substrate to C6 cells in all concentrations analyzed. Similar adherence activity of C6 cells were observed in the absence of ECM molecule and with BSA only (data not shown).
Figure 1 Glycosaminoglycans (GAGs) Effects on C6 Glioma Adhesion to extracellular matrix (ECM) proteins. (A) Dose dependency of C6 glioma adhesion to ECM-coated wells. Wells were coated with different concentrations of type IV collagen (Col IV), Laminin (Ln) or Fibronectin (Fn). Adhesion assays were performed as described in Methods. Results, expressed in Absorbance at 540 nm, were compared with the adhesion in the absence of ECM. Each point represents the mean of four independent experiments performed in triplicate ± SD. Effects of (B) Heparin, (C) chondroitin sulfate and (D) dextran sulfate on C6 Cells adhesion to ECM proteins. Results were compared with cells seeded on ECM substrate only (controls: 100%). Stars in B and C are equivalent for adhesion on laminin or fibronectin. Each point represents the mean of four independent experiments performed in triplicate ± SD. **p < 0.001, ***p < 0.0001 when compared to controls.
In order to analyze if GAGs could control C6 attachment to ECM molecules, we performed adhesion assays in the presence of heparin or chondroitin sulfate (Figure 1B, C). The addition of heparin significantly impaired cellular attachment to both laminin and fibronectin in a dose-dependent manner (Figure 1B). The higher heparin concentration (5 mg/ml) promoted 42% of inhibition in C6 cell adhesion to laminin and 42% to fibronectin. No inhibitory effect was observed in the attachment to type IV collagen. We also analyzed the attachment of C6 cells to ECM proteins in presence of commercial chondroitin sulfate to verify if other GAG could have similar effects as observed for heparin (Figure 1C). Interestingly, the addition of chondroitin sulfate inhibited C6 adhesion to laminin and fibronectin in a dose dependent manner, similarly to heparin. The higher chondroitin sulfate concentration (5 mg/ml) produced inhibition of 33.2% in C6 cell adhesion to laminin and 31.4% to fibronectin. Chondroitin sulfate did not affect C6 cell attachment to type IV collagen. In addition, both heparin and chondroitin sulfate did not significantly alter C6 adhesion in the absence of ECM molecules, to culture dish plastic (100% ± 10 and 105% ± 8 of adhesion at 5 mg/ml, respectively) or BSA-coated well (92% ± 7 and 95% ± 9 of adhesion at 5 mg/ml, respectively). After that, we investigated the influence of dextran sulfate on C6 adhesion, as a control polyanion (Figure 1D). Dextran sulfate did not alter C6 cells attachment to any ECM protein analyzed. C6 cells adhesion to the culture dish plastic or BSA-coated well were also not affected by dextran sulfate (95% ± 10 and 99% ± 11 at 5 mg/ml, respectively). The results could suggest that heparin and chondroitin sulfate are involved in C6 glioma adhesion to fibronectin and laminin in a specific way. Nevertheless, these GAGs seems do not affect glioma adhesion to collagen IV. These findings could be explained by the presence of proteoglycan binding sites in fibronectin and laminin molecules but not to collagen[18].
To further verify the involvement of GAGs in the modulation of glioma interaction with substrate, we analyzed cellular detachment (Table 1). It is interesting to note that the addition of GAGs and dextran sulfate (alone or in combination) to C6 cells attached to ECM molecules did not induce cell detachment. GAGs and dextran sulfate also have not effect on glioma detachment from culture dish plastic or BSA coated wells. Taken together, the results could suggest that heparin and chondroitin sulfate specifically regulate the initial phase of C6 cells adhesion to the substrate composed of laminin and fibronectin, perhaps in the organization or stabilization of focal contacts. However, these GAGs seem do not display significant effect after the complete cellular attachment.
Table 1 GAGs effects on C6 Glioma detachment
Substrate GAG % Detachment
Type IV Collagen Heparin 0% ± 0.14
Chondroitin Sulfate 7.5 % ± 0.7
Dextran Sulfate 0% ± 0.2
Fibronectin Heparin 5% ± 0.9
Chondroitin Sulfate 4% ± 0.6
Dextran Sulfate 0% ± 0.2
Laminin Heparin 0% ± 0.9
Chondroitin Sulfate 0% ± 0.12
Dextran Sulfate 0% ± 0.5
Culture dish plastic Heparin 7% ± 0,2
Chondroitin Sulfate 5% ± 0,1
Dextran Sulfate 2% ± 0,3
BSA Heparin 0% ± 0.1
Chondroitin Sulfate 5% ± 0,7
Dextran Sulfate 4% ± 0,4
Wells were coated with 10 μg/ml of type IV collagen, Fibronectin or Laminin. Detachment assays were performed as described. Results were compared with cells seeded on ECM substrate only. Coating with BSA or the culture dish plate was used as controls. Each point represents the mean of three independent experiments performed in triplicate ± SD.
We hypothesized if heparin and chondroitin sulfate treatment of C6 glioma cells adhesion could somewhat be related to modifications in cell proliferation. To test this we analyzed C6 cells proliferation when cultured on ECM proteins in the presence of these GAGs (Figure 2). We observed that heparin significantly reduced C6 cell proliferation on fibronectin as substrate at about 24% C6 in all concentrations analyzed (Figure 2A). In contrast, this GAG did not alter glioma proliferation when cultured on type IV collagen or laminin coated wells. The effects of chondroitin sulfate treatment on C6 cells proliferation were similar to that of heparin (Figure 2B). Chondroitin sulfate significantly impaired about 20% C6 cell proliferation on fibronectin in a similar proportion at the three concentrations tested. On the other hand, cell proliferation on laminin and type IV collagen was not altered. In addition, both heparin and chondroitin sulfate did not affect C6 cell proliferation in the absence of ECM molecules (data not shown).
Figure 2 Effects of (A) Heparin and (B) chondroitin sulfate on C6 Cells proliferation. Wells were coated with 10 μg/ml of type IV collagen (Col IV), Laminin (Ln) or Fibronectin (Fn). Proliferation assay was performed by MTT test as described in Methods. Each point represents the mean of three independent experiments performed in triplicate ± SD. **p < 0.001 when compared to cells seeded on ECM substrate only (controls: 100%).
Afterward, we investigated if GAGs also affect C6 glioma migration on ECM (Figure 3). To eliminate possible influence of proliferation, we established as nine hours as time limit in migration assay. The C6 glioma cells migration on different substrates increased over time. Amongst the ECM proteins, type IV collagen was the most effective in inducing C6 cells migration, followed by fibronectin, and subsequently by laminin. The addition of heparin or chondroitin sulfate at tested concentrations (50 or 100 mg/ml) did not alter cell migration on both type IV collagen (Figure 3A) and fibronectin (Figure 3B). In contrast, these GAGs significantly increased cell migration on laminin substrate different proportions according to the dose (Figure 3C). The most prominent effect was observed at time point of six hour when heparin increased cell migration in 222% and 567% 50 and 100 mg/ml, respectively, when compared to the control (without GAG). At this time point, chondroitin sulfate also presented expressive effect in stimulating cell migration (+166% and +373% at 50 and 100 mg/ml, respectively). C6 cell migration without ECM proteins was not affected by both GAGs (data not shown).
Figure 3 Migration of C6 cells on (A) Type IV Collagen, (B) Fibronectin or (C) Laminin. Wells were coated with 10 μg/ml of each ECM molecule. Cell migration was analyzed at 0, 6 or 9 hour. Quantitative migration measurements were calculated as the increase in the ratio beyond the initial radius of the cell population. CS (chondroitin sulfate), Hep (heparin). Each point represents the mean of three independent experiments performed in triplicate ± SD. ***p < 0.0001 when compared to cells seeded on ECM substrate only (without GAG).
Discussion
Tumor-stroma interactions play a pivotal role in regulation tumor progression and malignance[19]. We had shown previously that undersulfation of GAGs reduces the proliferation of the leukemic cell line Wehi 3B[15] and C6 glioma cells[16] suggesting a regulatory role of proteoglycans on cellular processes involved in tumor progression. We report here that GAGs are also involved in the control of glioma cell adhesion, proliferation and migration on ECM proteins.
It has long been appreciated that some of the earliest steps of neoplastic transformation involve changes in tissue interaction, including alterations in cell adhesion. Loss of cell adhesion was postulated as involved in genomic instability and related to the invasive behavior and motility of malignant cells[4]. Among ECM proteins, laminin was shown as the most potent stimulator of glioma migration and invasion[10,20]. In particular, laminin mediate C6 glioma cells invasion both in vitro and in vivo [10]. Mahesparan and co-workers[20] reported that laminin was the most efficient ECM protein in inducing cell migration of three different human glioma cell lines (U-373 MG, A-172 MG and HF-66), followed by fibronectin, while type IV collagen did not promote cell migration. In our experiments, however, we observed an inverse correlation with rat C6 glioma regarding both adhesion and migration: type IV collagen was the most potent substrate, followed by fibronectin and somewhat weaker attachment and migration on laminin. It is possible that the different adhesion ability of C6 cells to ECM proteins may be related to its migratory capacity.
For cell adhesion and migration, specific cell surface receptors that interact with ECM components are needed. Cell-surface heparan sulfate[21] or chondroitin sulfate[18] proteoglycans have been implicated in cell attachment via interactions with ECM proteins, such as laminin and fibronectin, and integrins. The syndecan family of transmembrane proteoglycans can act as co-receptors to modulate integrin-mediated cell-matrix adhesion. Together with integrins, they form a dual receptor system active in cell-matrix adhesion[22]. Cell attachment and spreading can be promoted through integrin interactions with the integrin-binding domain of fibronectin. Additional binding of a heparin-binding domain of fibronectin to a cell surface heparan sulfate proteoglycans (HSPG) promotes focal adhesion formation by activating PKC[23]. The HSPG involved appears to be exclusively syndecan-4[23] that binds to fibronectin via its GAG side chains[24].
Several studies have suggested that cell surface proteoglycans are involved in the control of glioma cell growth, adhesion and invasion. The chondroitin sulfate lectican BEHAB is up-regulated in malignant gliomas and derived cell lines, and was proposed to play a role in brain tumor invasion[19]. C6 glioma cells transfected with appican display dramatic change in their phenotypic appearance as well as increased cell adhesion [25]. In addition, we have previously demonstrated that sodium chlorate treatment, an inhibitor of GAG's sulfation, reduces C6 cells proliferation and adhesion to ECM proteins further suggesting that cell surface highly sulfated components act as co-receptor for glioma cell attachment to ECM[16]. In the present paper, we observed that the addition of heparin or chondroitin sulfate to the culture dish together within cells impaired C6 adhesion affecting the earliest steps of C6 cell adhesion, possibly due to the competition with GAG chains of cell surface proteoglycans to fibronectin and laminin binding sites. We could suggest that the specific interactions of GAGs with fibronectin and laminin affect the organization of focal adhesions necessary to promote cell adhesion. In contrast, addition of heparin or chondroitin sulfate after cells completely attached to substrate did not promote cell detachment possibly because focal adhesion points were already established and the addition of GAGs were not sufficient to disturb them. In fact, it has been reported that ECM ligands bind to heparan sulfate proteoglycans and produce actin filament reorganization even in the absence of integrin occupancy[26]. The interactions of heparan sulfate chains with extracellular ligands may help the formation of proteoglycan clusters, which appears to be critical for focal adhesion formation[26].
Moreover, it was reported that cell growth is anchorage dependent[27]. Thus, the reduction in C6 cell proliferation on fibronectin promoted by heparin and chondroitin sulfate observed in our experiments could be consequence of the altered cell adhesion. In addition, highly sulfated GAGs, heparin and heparan sulfate, and sulfated polysaccharide, fucoidan, have been reported stimulating tumor cell invasion in vitro, due to a stimulation of the proteolytic cascade of plasminogen activation[14]. This suggests that sulfated GAGs liberated by tumor cells, mediate ECM degradation amplifying pericellular plasminogen activation and locally enhancing tumor cell invasion in a positive feedback. In our experiments we observed that both chondroitin sulfate and heparin induced C6 cell migration on laminin substrate in different proportions according to the dose, with the most prominent effect at the fist 6 hours. These results are consistent with the influence of the GAGs on the earliest steps of cell adhesion to ECM proteins (Figure 1 and Table 1). It is possible that GAGs induce the organization of adhesion contacts in C6 cells that specifically lead to proliferation or migration when maintained on fibronectin or laminin, respectively.
Conclusion
One of the fundamental properties of pericellular proteoglycans is to interact with other molecules, thereby modulating cell adhesion, proliferation, and differentiation or the structural characteristics of tissues[22]. Indeed, we suggest that cell surface proteoglycans may function as coordinators of glioma cell interactions to microenvironment possible acting in the organization focal adhesions. Specific interactions of cell surface proteoglycans with ECM proteins may differently control glioma cell adhesion leading to alterations in proliferation or migration.
Methods
C6 glioma cell culture
The rat glioblastoma C6 cell line were a gift from Dr. Vivaldo Moura Neto (Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil) and cultured as previously described[16]. Cells were grown in Dulbecco's modified Eagle's medium (DMEM, Sigma, St. Louis) supplemented with 5% fetal calf serum (Fazenda Pigue, Brazil), 100 mg/ml streptomycin and 100 U/ml penicillin (all from Sigma) in a humidified incubator at 37°C in a 5% CO2, 95% air atmosphere (standard conditions). Cells were harvested with 0.125% trypsin/0,78 mM EDTA, when they reached confluence.
ECM coating and adhesion assay
Culture plates (Corning, NY, USA; 96 well) were coated with mouse type IV collagen, laminin or fibronectin (Gibco BRL, Grand Island, NY; 2 to 20 μg/ml) dissolved in phosphate buffered saline (PBS) pH 7.4 (12 hours, 4°C). Wells were washed with PBS, blocked with 2% BSA in PBS (1 hour, 37°C). Cellular adhesion was performed as previously described[16]. Briefly, cells were detached by trypsinization, washed in PBS, suspended in DMEM and seeded into ECM-coated 96-well plates (105 cells/well). After 2 hours of incubation under standard culture conditions, the non-adherent cells were removed by PBS washing. Attached cells were fixed (4% paraformaldehyde) and stained with 0.5% crystal violet dissolved in 20% methanol. The plates were then washed (PBS) and stain was eluted from cells with 0.1 M sodium citrate, pH 4.2 in 50% ethanol. The amount of stain was analyzed by optical density (OD) in ELISA wells microplate reader (Model ELX800, Bio-Tek Instruments) at 540 nm. In some experiments heparin, chondroitin sulfate or dextran sulfate (all from Sigma, 1 to 5 mg/ml) were added to the attachment medium at the same time as C6 cells were inoculated to the plates previously coated with 10 μg/ml of type IV collagen, laminin or fibronectin. For the detachment assays, cells were prepared as for the adhesion experiment, seeded on ECM-coated wells (5 × 104 cells/well) and incubated for 3 h in standard culture condition when most of them were attached. After PBS washing, 5 mg/ml of heparin, chondroitin sulfate or dextran sulfate were added and the cultures were incubated for additional 2 hours in standard conditions. Wells were them washed, and the attached cells fixed, stained and processed as described for adhesion assays. Cell detachment was calculated by the difference from the percentage of adherent cells in relation to controls (ECM-coated well without GAG).
MTT assay
This method has been extensively used to determine glial cell proliferation[16,28,29]. C6 cells were plated (104 cell/well) in 96-well culture plates previously coated with 10 μg/ml ECM proteins as described above in the presence of heparin or chondroitin sulfate (1 to 3 mg/ml) and cultured for 2 days. Viable cells were quantified by MTT colorimetric assay for mitochondrial dehydrogenase[30].
Migration assay
A monolayer migration assay was used to quantify locomotion of C6 cells on ECM. Cells (1.5 × 105) were seeded into 13 mm coverslips. After overnight incubation in standard culture conditions, coverslips containing confluent monolayers were transferred to the center of 35 mm wells (with cells up) previously coated with 10 μg/ml of ECM proteins as described above. Fresh DMEM containing heparin or chondroitin sulfate (1 to 3 mg/ml) was added. Cells were maintained in standard culture conditions except during brief microscopic examinations. Migration was calculated by serial measurements of the ratio beyond the initial cell population (cells that migrate out of the coverslip to ECM substrate). The analysis were performed at 0, 6, and 9 hours, using a micrometric ocular (Olympus) attached to an Olympus CK 40 microscope.
Statistical analysis
The significance of differences was evaluated by means of One-way ANOVA with Newman-Keuls or Dunnett as post-tests using the GraphPad InStat program (San Diego California USA).
List of abbreviations used
BSA – Bovine serum albumin
DMEM – Dulbecco's modified Eagle's medium
ECM – Extracellular matrix
GAG – Glycosaminoglycans
GBM – Glioblastoma multiforme
HSPG – heparin sulfate proteoglycans
PBS – Phosphate buffered saline
Authors' contributions
CBNMA carried out cell culture, detachment, adhesion, proliferation and migration assays. BLS carried out cell culture and adhesion assays. AGT and MAS conceived the study and participate in the design and coordination. All authors participate in the analysis of data and discussion of the results. All authors read and approved the current version of the manuscript.
Acknowledgements
We thank Professor Vivaldo Moura Neto, Universidade Federal do Rio de Janeiro, Brazil, for providing C6 cells. Research supported by CAPES and CNPq.
==== Refs
Holland EC Glioblastoma multiforme: The terminator Proceedings of the National Academy of Sciences of the United States of America 2000 97 6242 6244 10841526 10.1073/pnas.97.12.6242
Grobben B De Deyn PP Slegers H Rat C6 glioma as experimental model system for the study of glioblastoma growth and invasion Cell Tissue Res 2002 310 257 270 12457224 10.1007/s00441-002-0651-7
Lin CQ Bissell MJ Multifaceted Regulation of Cell-Differentiation by Extracellular-Matrix Faseb Journal 1993 7 737 743 8330681
Tlsty TD Cell-adhesion-dependent influences on genomic instability and carcinogenesis Current Opinion in Cell Biology 1998 10 647 653 9818176 10.1016/S0955-0674(98)80041-0
Giese A Westphal M Glioma invasion in the central nervous system Neurosurgery 1996 39 235 250 8832660 10.1097/00006123-199608000-00001
Haugland HK Tysnes BB Tysnes OB Adhesion and migration of human glioma cells are differently dependent on extracellular matrix molecules Anticancer Research 1997 17 1035 1042 9137446
Rempel SA Golembieski WA Fisher JL Maile M Nakeff A SPARC modulates cell growth, attachment and migration of U87 glioma cells on brain extracellular matrix proteins Journal of Neuro-Oncology 2001 53 149 160 11716067 10.1023/A:1012201300188
Koochekpour S Merzak A Pilkington GJ Extracellular-Matrix Proteins Inhibit Proliferation, Up-Regulate Migration and Induce Morphological-Changes in Human Glioma Cell-Lines European Journal of Cancer 1995 31A 375 380 7540403 10.1016/0959-8049(94)00476-L
Giese A Loo MA Tran N Haskett D Coons SW Berens ME Dichotomy of astrocytoma migration and proliferation Int J Cancer 1996 67 275 282 8760599 10.1002/(SICI)1097-0215(19960717)67:2<275::AID-IJC20>3.0.CO;2-9
Muir D Johnson J Rojiani M Inglis BA Rojiani A Maria BL Assessment of laminin-mediated glioma invasion in vitro and by glioma tumors engrafted within rat spinal cord Journal of Neuro-Oncology 1996 30 199 211 8943095 10.1007/BF00177271
Yavin E Gabai A Gil S Nerve growth-factor mediates monosialoganglioside-induced release of fibronectin and J1/tenascin from C6 glioma-cells Journal of Neurochemistry 1991 56 105 112 1702827
Gunther W Skaftnesmo KO Arnold H Terzis AJ Molecular approaches to brain tumour invasion Acta Neurochir (Wien ) 2003 145 1029 1036 14663559 10.1007/s00701-003-0099-x
Malekhedayat S Rome LH Expression of multiple integrins and extracellular-matrix components by C6 glioma-cells Journal of Neuroscience Research 1992 31 470 478 1640499 10.1002/jnr.490310309
Brunner G Reimbold K Meissauer A Schirrmacher V Erkell LJ Sulfated glycosaminoglycans enhance tumor cell invasion in vitro by stimulating plasminogen activation Experimental Cell Research 1998 239 301 310 9521847 10.1006/excr.1997.3877
Alvarez-Silva M Trentin AG Undersulfation of glycosaminoglycans reduces the proliferation of a leukemia cell line in vitro Braz J Med Biol Res 1996 29 1239 1242 9181069
Mendes de Aguiar CBN Garcez RC Alvarez-Silva M Trentin AG Undersulfation of proteoglycans and proteins alter C6 glioma cells proliferation, adhesion and extracellular matrix organization International Journal of Developmental Neuroscience 2002 20 563 571 12485624 10.1016/S0736-5748(02)00081-3
Franco CR Rocha HA Trindade ES Santos IA Leite EL Veiga SS Nader HB Dietrich CP Heparan sulfate and control of cell division: adhesion and proliferation of mutant CHO-745 cells lacking xylosyl transferase Braz J Med Biol Res 2001 34 971 975 11471034 10.1590/S0100-879X2001000800001
Moyano JV Carnemolla B Albar JP Leprini A Gaggero B Zardi L Garcia-Pardo A Cooperative role for activated alpha 4 beta 1 integrin and chondroitin sulfate proteoglycans in cell adhesion to the heparin III domain of fibronectin - Identification of a novel heparin and cell binding sequence in repeat III5 Journal of Biological Chemistry 1999 274 135 142 9867821 10.1074/jbc.274.1.135
Bellail AC Hunter SB Brat DJ Tan C Van Meir EG Microregional extracellular matrix heterogeneity in brain modulates glioma cell invasion Int J Biochem Cell Biol 2004 36 1046 1069 15094120 10.1016/j.biocel.2004.01.013
Mahesparan R Tysnes BB Edvardsen K Haugeland HK Cabrera IG Lund-Johansen M Engebraaten O Bjerkvig R Role of high molecular weight extracellular matrix proteins in glioma cell migration Neuropathol Appl Neurobiol 1997 23 102 112 9160895 10.1046/j.1365-2990.1997.8398083.x
Liu W Litwack ED Stanley MJ Langford JK Lander AD Sanderson RD Heparan sulfate proteoglycans as adhesive and anti-invasive molecules - Syndecans and glypican have distinct functions Journal of Biological Chemistry 1998 273 22825 22832 9712917 10.1074/jbc.273.35.22825
Woods A Couchman JR Syndecans: synergistic activators of cell adhesion Trends in Cell Biology 1998 8 189 192 9695837 10.1016/S0962-8924(98)01244-6
Tumova S Woods A Couchman JR Heparan sulfate chains from glypican and syndecans bind the Hep II domain of fibronectin similarly despite minor structural differences Journal of Biological Chemistry 2000 275 9410 9417 10734086 10.1074/jbc.275.13.9410
Woods A Longley RL Tumova S Couchman JR Syndecan-4 binding to the high affinity heparin-binding domain of fibronectin drives focal adhesion formation in fibroblasts Arch Biochem Biophys 2000 374 66 72 10640397 10.1006/abbi.1999.1607
Wu AF Pangalos MN Efthimiopoulos S Shioi J Robakis NK Appican expression induces morphological changes in C6 glioma cells and promotes adhesion of neural cells to the extracellular matrix Journal of Neuroscience 1997 17 4987 4993 9185536
Bernfield M Gotte M Park PW Reizes O Fitzgerald ML Lincecum J Zako M Functions of cell surface heparan sulfate proteoglycans Annual Review of Biochemistry 1999 68 729 777 10872465 10.1146/annurev.biochem.68.1.729
Wary KK Mariotti A Zurzolo C Giancotti FG A requirement for caveolin-1 and associated kinase Fyn in integrin signaling and anchorage-dependent cell growth Cell 1998 94 625 634 9741627 10.1016/S0092-8674(00)81604-9
Trentin AG Neto VM T-3 Affects Cerebellar Astrocyte Proliferation, Gfap and Fibronectin Organization Neuroreport 1995 6 293 296 7756613
Trentin AG Alvarez-Silva M Moura Neto V Thyroid hormone induces cerebellar astrocytes and C6 glioma cells to secrete mitogenic growth factors Am J Physiol Endocrinol Metab 2001 281 E1088 E1094 11595667
Mosmann T Rapid colorimetric assay for cellular growth and survival: application to proliferation and cytotoxicity assays J Immunol Methods 1983 65 55 63 6606682 10.1016/0022-1759(83)90303-4
|
16111491
|
PMC1201133
|
CC BY
|
2021-01-04 16:31:30
|
no
|
BMC Cell Biol. 2005 Aug 19; 6:31
|
utf-8
|
BMC Cell Biol
| 2,005 |
10.1186/1471-2121-6-31
|
oa_comm
|
==== Front
BMC Clin PatholBMC Clinical Pathology1472-6890BioMed Central London 1472-6890-5-71609554310.1186/1472-6890-5-7Technical AdvanceDNA index determination with Automated Cellular Imaging System (ACIS) in Barrett's esophagus: Comparison with CAS 200 Huang Qin [email protected] Chenggong [email protected] Michael [email protected] James [email protected] Raj K [email protected] Center for Swallowing and Motility Disorders, Departments of Medicine & Pathology and Laboratory Medicine, VA Boston Healthcare System and Harvard Medical School, 1400 VFW Parkway, West Roxbury, MA 02132, USA2 Department of Pathology and Laboratory Medicine, Providence VA Medical Center and Brown Medical School, Providence, RI 02908, USA2005 12 8 2005 5 7 7 8 2 2005 12 8 2005 Copyright © 2005 Huang et al; licensee BioMed Central Ltd.2005Huang et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
For solid tumors, image cytometry has been shown to be more sensitive for diagnosing DNA content abnormalities (aneuploidy) than flow cytometry. Image cytometry has often been performed using the semi-automated CAS 200 system. Recently, an Automated Cellular Imaging System (ACIS) was introduced to determine DNA content (DNA index), but it has not been validated.
Methods
Using the CAS 200 system and ACIS, we compared the DNA index (DI) obtained from the same archived formalin-fixed and paraffin embedded tissue samples from Barrett's esophagus related lesions, including samples with specialized intestinal metaplasia without dysplasia, low-grade dysplasia, high-grade dysplasia and adenocarcinoma.
Results
Although there was a very good correlation between the DI values determined by ACIS and CAS 200, the former was 25% more sensitive in detecting aneuploidy. ACIS yielded a mean DI value 18% higher than that obtained by CAS 200 (p < 0.001; paired t test). In addition, the average time required to perform a DNA ploidy analysis was shorter with the ACIS (30–40 min) than with the CAS 200 (40–70 min). Results obtained by ACIS gave excellent inter-and intra-observer variability (coefficient of correlation >0.9 for both, p < 0.0001).
Conclusion
Compared with the CAS 200, the ACIS is a more sensitive and less time consuming technique for determining DNA ploidy. Results obtained by ACIS are also highly reproducible.
==== Body
Background
By molecular pathology, Barrett's adenocarcinoma has been shown to be an acquired malignancy with multi-genetic alterations involving almost all chromosomes [1,2]. These observations have indicated that numerical and structural chromosomal abnormalities may be important events in the carcinogenic process in Barrett's adenocarcinoma. This is especially important inasmuch as there is accumulating evidence that DNA aneuploidy is a key early event in tumorigenesis and may be a cause rather than a consequence of malignancy [3,4]. Chromosomal abnormalities occur predominantly in aneuploid malignant cells, leading to progressive deterioration of aneuploidy [5]. In addition to changes in chromosome copy numbers, cancer cells may also experience a change in the size of individual chromosomes, due to chromosomal translocations, deletions, and duplications. In flow or image cytometry of cancer cells, DNA content is represented by the DNA index (DI) and described in reference to the DI of normal diploid cells. DI values reflect integrated optical density (IOD), which takes into account chromosome copy number, extra chromosomal fragments and the DNA content of individual chromosomes, as well as morphometric features of the nuclei. Thus, DI values reflect more than the copy number of the chromosomes in a nucleus, and the term aneuploidy, which was originally introduced to indicate only changes in chromosome copy number, is now used to indicate any changes in the DI of cancer and pre-cancerous cells.
DNA aneuploidy has been used as a marker of cancers that exhibit more aggressive behavior than diploid cancers, and as a marker of pre-malignant lesions that are at high risk for malignant progression to cancer. Although DNA aneuploidy has usually been measured by flow cytometry, the sensitivity of this method is limited in diagnosing DNA content abnormalities because it includes both affected cells as well as benign epithelial and stromal cells in the test sample. Image cytometry has recently been used to estimate the DI of cells in disaggregated cytospin preparations or in microscopically identified epithelial cells [4]. Image cytometry has been shown to be more sensitive than flow cytometry for analyzing DNA content [6,7]. Image cytometry has been performed using a number of systems, the most common being the semi-automated CAS 200 system [8-10]. Recently, a more automated system, the Automated Cellular Imaging System (ACIS) (ChromaVision Medical System, Inc., San Juan Capistrano, CA) was introduced for the analysis of DNA ploidy, but results using this system have not been validated against the conventional CAS 200 system. Using these two systems, in the present work we compared the DNA index (DI) obtained from the same archived formalin-fixed and paraffin embedded tissue samples from Barrett's esophagus related lesions, including those with dysplasia and adenocarcinoma. We found that the two systems gave correlated DI values and that the ACIS system gave higher DI values, diagnosed aneuploidy more frequently, and may be less labor intensive than the CAS 200 system. Furthermore, results obtained with ACIS were highly reproducible.
Methods
Tissue samples
A total of 34 distal esophageal biopsies or distal esophageal resection specimens containing Barrett's esophagus related lesions were collected from 13 patients. The DI of these tissue samples had been examined using the CAS 200 (Becton Dickenson, San Jose, CA) [8] and were subsequently assayed using the ACIS instrument. The study protocol was approved by the Institutional Review Boards of the Boston and Providence Veterans Affairs Medical Centers.
Histology
The tissue blocks were sectioned to obtain two adjacent sections, one stained with H&E and the other with Feulgen dye. The H&E stained sections were examined by two experienced pathologists (Q.H. and M.K.) for consensus diagnoses of Barrett's esophagus with specialized intestinal metaplasia (SIM), without dysplasia (BE), and for low-grade dysplasia (LGD), high-grade dysplasia (HGD), and esophageal adenocarcinoma (EAC), according to published criteria [11,12]. Briefly, SIM was diagnosed if the esophageal epithelium was columnar and contained characteristic goblet cells [11]. A diagnosis of LGD depended primarily on prominent cytological changes such as enlarged nuclei with hyperchromasia and increased number of mitoses as well as mild architectural changes. The crypt architecture in LGD was slightly distorted but generally preserved. The epithelial nuclei were enlarged, crowded, and hyperchromatic with increased mitotic figures, and the changes were present in the upper portion of the crypt. The nuclei remained polarized but were slightly stratified, and the stratification did not reach the apical surface of the glands. Mucus in the goblet and columnar cells was diminished or absent. In contrast, samples with high-grade dysplasia (HGD) showed marked cytological and architectural changes, including abnormal glandular proliferations with villiform and cribriform growth patterns in the upper portion of the mucosa, and marked distortion of crypt architecture with branching, lateral budding of crypts, and "back-to-back" patterns [12]. Nuclear stratification reached the crypt luminal surface, with loss of nuclear polarity, and nuclei varied markedly in size, shape and staining characteristics. The nuclear abnormalities extended to the mucosal surface. Mucin production in the goblet and columnar cells was usually absent. Esophageal adenocarcinoma (EAC) was diagnosed when dysplastic columnar epithelial cells invaded through the basement membrane into the muscularis mucosa, submucosa, and beyond.
DNA index (DI) determination by CAS 2000
For analysis on the CAS 200 System, fixed 5 μm tissue sections were Feulgen-stained using the Quantitative DNA Staining Kit (Cell Analysis Systems, Inc., Elmhurst, IL) as recommended by the manufacturer. This staining process involves hydrolysis with concentrated hydrochloric acid, that removes non-nuclear substances and hydrolyses chromatin into its constituent nucleic acids, followed by stoichiometric binding of the dye to the nucleic acids, imparting a blue color. The intensity of the blue color is directly proportional to the amount of DNA. Areas on the Feulgen-stained sections that contained pathological lesions, as defined by adjacent H&E stained slides, were marked for DNA content analysis. The corresponding areas on the Feulgen-stained sections were scanned, digitized, and stored as individual files using the CAS 200 quantitative DNA software program [13]. Artifacts, including overlapping nuclei, were edited out. The digitized images of microscopically selected nuclei were converted into a series of pixels, which were quantified on the basis of parameters such as the integrated optical density (IOD), reflecting the DNA content of the selected nucleus. CAS 200 also uses a tissue correction software, which corrects for underestimation of DI due to cutting off of the nuclei. Both the Feulgen staining and ploidy analysis on the CAS 200 were performed in the Molecular Pathology Laboratory of Columbia University Medical Center in New York City, according to standard protocols and institutional criteria [8,13].
DNA index determination by ACIS
DNA content analysis on ACIS was performed according to the procedures suggested by the manufacturer. Briefly, the sections were cut at 7 μm and stained with Feulgen stain (ACIS kit), which includes hydrolysis with hydrochloric acid that removes non-nuclear substances and purines from the deoxyribose backbone of DNA molecules, followed by reaction with Schiff's reagent, an aqueous solution of cresyl-violet and sulfurous acid. Thus, the amount of Feulgen stain is directly proportional to the amount of DNA present in the nucleus. Feulgen-stained slides were automatically scanned with an Olympus microscope equipped with 3 Sony digital CCD chips at a speed of 30-frames/second. At a 40× magnification, the final estimated resolution was 250,000 per pixel (770,000 per airy disk), each individual Field of View (FOV) measured 160 × 120 μm, and there were157 × 209 (32,813) FOVs on a slide. Only the properly and uniformly stained slides were included. The final display was a 24-bit resolution (1280 × 1024), with the area of interest captured at 40× magnification. A daily quality control run with a standard calibration kit from the manufacturer was performed to ensure proper focus, black and white level balance for each microscope objective, and linear camera output.
The ACIS uses an inbuilt system that sets the filter threshold controlling the inclusion of stained nuclei at predefined (1–5) levels. The operator selects the level that helps the user separate/cut in between adjacent cells and cannot be identified by the software as individual cells, ranging from the least (level 1) to the most (level 5) aggressive. The system 'remembers' analysis regions to prevent the user from re-collecting nuclei from the same region. Unlike the CAS 200, no tissue correction factor is applied in the ACIS system to correct for underestimation of DI. To avoid inclusion of touching and overlapping nuclei, ACIS uses a set of image processing algorithms known as Watershed Segmentation. In these algorithms, nuclei that touch or overlap other nuclei are recognized through their size and other morphometric parameters and are separated by insertion of a single pixel-wide boundary at the point of contact. The analysis software provides five distinct cell separation profiles to allow optimal separation across a range of specimen morphologies, a process that works well for nuclei that touch or overlap to a modest degree. To avoid inclusion of excessively overlapped nuclei, each of the cell separation profiles was designed to recognize and exclude nuclei that exhibit excessive overlapping based on signature combinations of size, shape, color and morphometric filter descriptors.
Overlapping nuclei, nuclear debris and other artifacts that escaped auto-detection and removal by the system were deleted by the operator. Qualified nuclei of approximately 30 control stromal cells, such as endothelial cells, macrophages, and fibroblasts, in the same tissue and about 200-targeted epithelial cells are obtained. DNA content histograms are automatically plotted in another window using the ACIS DNA ploidy software (Figure 1). The user can easily navigate between individual nuclei and their exact position on the DI histogram.
Figure 1 ACIS screen showing montage of microscopically selected and operator edited Feulgen-stained nuclei and the DNA content histogram generated on the basis of integrated optical density.
The mean integrated optical density (IOD) of control cells is assigned a DI of 1, which serves as an internal diploid (2N) standard and reference for DI calculation of the targeted cells. The IOD of control cells has a coefficient of variation (CV) of less than 10%. The histograms of the target cells showed a primary G0/G1 peak, and additional cells with or without well defined peaks with DI outside the G0/G1 peak (e.g., in S- or G2 phase of the cell cycle).
Reproducibility of the ACIS
We randomly selected 30 samples of Barrett's lesions, including 5 cases each of gastric cardiac mucosa, SIM, ID (Indefinite for Dysplasia), LGD, HGD and EAC. To determine inter-observer variability, two trained, independent observers who were blinded to the diagnosis separately performed DI analysis on the same slide. To determine intra-observer reproducibility, the same observer performed DI analyses on the same slide at an interval of more than 15 days.
Aneuploidy determination
Aneuploidy was diagnosed when the targeted epithelial nuclei showed no diploid G0/G1 peak and the DI of the G0/G1 peak was clearly positioned outside the diploid range.
Statistical analysis
Statistical analysis was performed largely with Microsoft Excel. For linear regression analysis, the software SPSS (Chicago) was used. A p value of < 0.05 was considered statistically significant.
Results
DI values determined by the CAS 200 and ACIS systems
Figure 2 shows examples of DI histograms obtained with the CAS 200 and ACIS instruments on distal esophageal tissue samples with various pathologic lesions. In general, we found that the ACIS yielded G0/G1 peaks with higher DI values than the CAS 200 system. When we assessed individual DI values of G0/G1 peaks obtained with the two systems, we again found that, in most cases and in all histopathological types of Barrett related lesions, the ACIS yielded higher DI values (p < 0.001, paired t test; Figure 3). On average, the DI values obtained with the ACIS were 18% higher than those obtained with the CAS 200. Both linear and logistical correlation analyses showed a statistically significant correlation between DI values generated by the two instruments (p < 0.01, two tailed; Figure 4).
Figure 2 DI histograms generated by the CAS 200 (the middle column) and ACIS (the right column). The left column shows typical histology of the Barrett's related lesions (H&E stained, 400×) including specialized intestinal metaplasia without dysplasia (row A), low-grade dysplasia (row B), high-grade dysplasia (row C), and adenocarcinoma (row D). The corresponding DI histograms showed progressive increases in DNA content as disease progressed. On each DI histogram, the Y axis represents the number of total cells included and the X axis shows changes in DNA content in arbitrary units.
Figure 3 Comparison of DI values as determined by the CAS 200 and ACIS methods in various Barrett's lesions (p < 0.001, paired t test). Dots represent individual sections examined by both instruments.
Figure 4 Linear regression analysis of DI values determined on the CAS 200 and ACIS instruments (n = 37). Note the excellent correlation obtained by the two systems. R = 0.89, p < 0.01 (two tailed).
Ploidy analysis
Diagnoses of aneuploidy have been based on the upper limit of normal G0/G1 peak DI values, which vary from 1.1 to 1.29. Using 1.1 as the upper limit of the normal euploid range, CAS 200 diagnosed aneuploidy in 2/4 (50%) cases of SIM with no dysplasia, 3/7 (43%) cases of LGD, 11/12 (92%) cases of HGD, and 7/8 (88%) cases of EAC, whereas the ACIS diagnosed aneuploidy in all of these tissue samples (Table 1). Using 1.29 as the upper limit of the normal DI range, CAS 200 detected aneuploidy in 1/4 (25%) cases of SIM with no dysplasia, 0/7 (0%) cases of LGD, 10/12 (83%) cases of HGD, and 6/8 (75%) cases of EAC. In contrast, the ACIS diagnosed aneuploidy in 3/4 (75%) cases of SIM with no dysplasia, 4/7 (57%) cases of LGD, 11/12 (92%) cases of HGD, and 8/8 (100%) cases of EAC. Together, these findings indicate that the ACIS is more sensitive than the CAS 200 in diagnosing aneuploidy based on DI values of the G0/G1 peak.
Table 1 Prevalence of aneuploidy in various Barrett's lesions on the basis of 2 different DI cut-off upper limits
DI ≥1.1 ≥1.3
CAS 200 ACIS CAS 200 ACIS
C-GI 0/3 (0%) 0/3 (0%) 0/3 (0%) 0/3 (0%)
SIM 2/4(50%) 4/4(100%) 1/4(25%) 3/4(75%)
LGD 3/7(43%) 7/7(100%) 0/7(0%) 4/7(57%)
HGD 11/12(92%) 12/12(100%) 10/12(83%) 11/12(92%)
EAC 7/8(88%) 8/8(100%) 6/8(75%) 8/8(100%)
Total 23/34(68%) 31/34(91%) 17/34(50%) 26/34(76%)
Abbreviations: C-GI, normal gastrointestinal mucosa; SIM, specialized intestinal metaplasia; LGD, low grade dysplasia; HGD, high grade dysplasia; EAC, esophageal adenocarcinoma.
Reproducibility of ACIS DI values
The DI values of G0/G1 peaks from an assortment of 30 Barrett's associated lesions were analyzed for inter- and intra-observer variability. For inter-observer variability, the Pearson coefficient of correlation was 0.917 (p < 0.0001, 2-tailed, n = 30), whereas, for intra-observer variability, it was 0.920 (p < 0.0001, 2-tailed, n = 30), thus showing that ACIS gives highly reproducible DI values.
Discussion
Results have indicated that analysis of DNA ploidy may be a better method for the detection of pre-malignant lesions at high risk of malignant progression than histopathological dysplasia [4,8-10,14]. In addition, DNA aneuploidy may indicate the aggressiveness of malignant neoplasms in a variety of organs, including the prostate, breast, urothelial tract, cervix, ovary, lung, skin, and oral mucosa [3,4,15-20]. Moreover, very early DNA alterations, such as single chromosome or locus specific chromosomal abnormalities, may occur prior to expression of DNA abnormalities.
DNA ploidy analysis of solid tumors can be performed on cell suspensions using flow or image cytometry. Image cytometry performed on histological sections allows selective examination of the targeted epithelial cells. In addition, flow cytometry is less sensitive than image cytometry, and many lesions diagnosed as euploid on flow cytometry have been found to be aneuploid on image cytometry [6,7]. In Barrett's related esophageal lesions, however, only flow cytometry has been used in the diagnosis of large series of patients (14), indicating the need for large-scale studies of image cytometry in these lesions [8].
Diagnosis of aneuploidy by image cytometry has not been standardized. Different laboratories use different image cytometry systems, DNA staining techniques, section thicknesses, and types of control cells. Among the systems used are the CAS 100 [21] and CAS 200-Cell Analysis System Image Analyzers (Becton-Dickinson, San Jose, CA) [8,14,22,23], the QPATH (LEICA, Cambridge, England) [24], the MPV3 (Leitz, Wetzlar, Germany) equipped with a DNA cytometry system (ACAS, Ahrens, Bargteheide, Germany) [25], the Cyto-Savant Image Analyzer (Oncometrics Inc, British Columbia, Canada) [26,27]; the SAMBA 4000 Image Analyzer (Imaging Products International, Chantilly, VA) [28], and the Firefield ploidy system (Firefield Imaging Ltd, Nottingham, UK) [4,16]. The performance of these instruments has not been compared systematically and there is little published data comparing them. We therefore compared the results of DI analysis obtained by the ACIS and CAS 200 systems. We found that DI determination obtained by these two systems were well correlated, but that the DI values obtained with the ACIS were consistently higher than those acquired on the CAS 200. We also found that the ACIS was more sensitive in diagnosing aneuploidy, and that the protocol using the ACIS took less time than the CAS 200. Finally, the DI values obtained by the ACIS were highly reproducible.
We were somewhat surprised that the DI values obtained using the ACIS were consistently higher than those obtained using the CAS 200, inasmuch as similar image analysis systems should yield similar results. There are several possible explanations for this difference. First, the ChromaVision Blue Feulgen Stain Kit we used with the ACIS is composed of a Schiff reagent, which is truly stoichiometric, enabling a complete and thorough penetration of the stain and permitting uniform staining of the DNA molecules. Second, DI values are also dependent upon nuclear morphological features, including nuclear area, shape and density, and the ACIS system may more thoroughly account for these morphological features. Third, the tissue sections used for the ACIS (7 μm) were thicker than those used for the CAS 200 (5 μm). The latter system uses thinner sections to decrease the incidence of nuclear overlap, but it also yields an increased frequency of cut nuclei. The CAS 200 addressed this issue by using a tissue correction factor, which in essence is a right shift applied to the DI histogram and bringing the values into an acceptable numerical range. This correction may permit the detection of small aneuploid and tetraploid peaks, thereby preventing a 10% to 15% underestimate of aneuploid cases as diploid. The software, however, may not fully correct for nuclear truncation. In contrast, the thicker sections used in the ACIS markedly decrease the frequency of cut nuclei, thus avoiding the need for an artificial mathematical shift in the DI histograms. The ACIS uses a set of image processing algorithms (known as Watershed Segmentation) to consistently separate touching or moderately overlapping nuclei. In these algorithms, nuclei that touch or overlap are recognized through their size and other morphometric parameters, and are separated through the segmentation process by insertion of a single pixel-wide boundary between them at the point of contact. The analysis software provides five distinct cell separation profiles to allow optimal separation across a range of specimen morphologies. Each of these cell separation profiles is designed to recognize and exclude nuclei that exhibit excessive overlap based on signature combinations of size, shape, color and morphometric filter descriptors. In addition, the system allows the user to further delete any overlapping nuclear images and artifacts that may evade aneuploidy detection by ACIS. In this fashion, sufficient numbers of nuclei can be obtained from thicker sections without the complications of cut cells and the requirement for tissue correction factors.
Another difference between these two systems is in their control cells. The ACIS uses benign stromal cells, such as endothelial cells, macrophages, fibroblasts, and large lymphocytes, whereas the CAS 200 uses inactive small lymphocytes. Inactive small lymphocytes have more condensed nuclear chromatin structure than benign stromal cells and may not be stained optimally with the Feulgen dye. However, the lower IODs of lymphocytes would yield higher DI values for the epithelial cells. Some users of CAS 200 have also used external reference cells, such as rat hepatocytes or normal urothelial cells or human cerebellar cells. In general, selection of control cells may not make a dramatic difference in DI determination unless the controls are actively dividing cells such as gastrointestinal epithelial cells. It is therefore unlikely that the selection of control cells can explain the significant differences in the DI values generated by these two systems.
One of the main limitations of this study is that the two imaging systems were operated by different individuals, and operator related differences may yield different results. For example, differences in determining the threshold of digitized nuclear images may yield different values for DNA content, thereby changing the appearance of the DNA histograms. Moreover, the exact same cells were not analyzed by the two systems, adding to the variability in the results, although this difference was unlikely to produce consistently lower valves for the CAS 200.
The diagnosis of an aneuploid G0/G1 peak depends on its distinct separation from the normal diploid peak. Several performance measures have been proposed to allow image cytometry systems to identify aneuploidy if the DI peak deviates more than 10% from the diploid peak [29], but these measures are rarely reported in published studies [8,22-27]. Similarly, these studies rarely state the criteria by which the DI peak is determined. Euploidy (diploidy) may be diagnosed when cells in the G0/G1 peak have a mean DI value of less than 1.1 [22,23,26,27] or less than 1.3 [8,24,25]. In formalin-fixed gastrointestinal tissues, where the normal mucosa may have a mean DI value in the G0/G1 peak up to 1.29, a DI cut-off value of 1.3 seems more appropriate. In this study, we analyzed the prevalence of aneuploidy with the two cytometers using DI cut-off values of 1.1 and 1.30. We found that the ACIS was more sensitive than the CAS 200 in diagnosing aneuploidy using either cut-off value.
Aneuploidy has also been diagnosed in the presence of a normal euploid G0/G1 peak when additional aneuploid peaks, as evidenced by peaks of cells with DI values greater than those of the cells in the G0/G1 peak, are identified. However, what constitutes the aneuploid peaks is not clear. The number of nuclei in the G2 phase that are considered abnormal (tetraploid aneuploidy) varies from >10% [4] to >15% [8,27] to >25% [23]. Similarly, aneuploidy has been diagnosed when there are additional well-defined aneuploid peaks or when the number of nuclei with DI values in the aneuploid region constitute >1% of the total cells [4]. In this study, we did not diagnose aneuploidy when a normal diploid G0/G1 peak was present. Rather, we diagnosed aneuploidy only when the G0/G1 peak was clearly outside the diploid range. Since ACIS is more sensitive and yields higher DI values throughout the entire range of the DI, it will detect aneuploidy more frequently using similar diagnostic criteria.
To compare the two systems, we used an assortment of tissues with different histological types, including specialized intestinal metaplasia, various grades of dysplasia and adenocarcinoma, so that the full spectrum of pathological lesions may be covered. We found that the type of histology did not affect the results, in that the ACIS yielded higher DI values than the CAS 200 regardless of the type of histological lesions.
Conclusion
ACIS provides a sensitive, efficient and highly reproducible method of DNA ploidy determination, and results obtained using this system correlated well with those obtained by the CAS 200. The ACIS system yielded DI values that were consistently higher than those obtained by the CAS 200. Variables such as the DNA stain used, the thickness of tissue sections, the tissue correction factor and the capture of morphological features of the nuclei may account for differences in results obtained with these two image cytometers. Moreover, automated image capture, semi-automated collection of cell nuclei, easy on-screen navigation and overall improved workflow may contribute to the shorter time needed for ploidy determination using the ACIS. Our findings indicate the importance of standardizing DNA content and ploidy determination methods.
List of abbreviations used
ACIS, automated cellular imaging system
DI, DNA index
IOD, integrated optical density
CV, coefficient of variation
BE, Barrett's esophagus
SIM, specialized intestinal metaplasia
LGD, low-grade dysplasia
HGD, high-grade dysplasia
EAC, esophageal adenocarcinoma
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
QH designed the project, participated in the data collection, and prepared the manuscript.
CY carried out the image analysis on ACIS, performed statistical analyses, and helped prepare the manuscript.
MF participated in the DNA ploidy analysis on CAS200.
MK participated in the histopathologic analysis of the specimens and edited the manuscript.
RG participated in the study design, statistical analysis, and manuscript preparation
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This work was supported in part by NIH grants DK031092 and DK62867, a Merit Review Award from the Research Service, VHA, and an award from the New England Medical Research Institute, Boston. The authors thank Dr. Errol Baker at the Veterans Affairs Boston Healthcare System for help with the statistical analyses.
==== Refs
Beuzen F Dubois S Flejou JF Chromosomal numerical aberrations are frequent in oesophageal and gastric adenocarcinomas: a study using in-situ hybridization Histopathology 2000 37 241 249 10971700 10.1046/j.1365-2559.2000.00887.x
Walch A Bink K Gais P Stangl S Hutzler P Aubele M Mueller J Hofler H Werner M Evaluation of c-erbB-2 overexpression and Her-2/neu gene copy number heterogeneity in Barrett's adenocarcinoma Anal Cell Pathol 2000 20 25 32 11007435
Haroske G Baak JP Danielsen H Giroud F Gschwendtner A Oberholzer M Reith A Spieler P Bocking A Fourth updated ESACP consensus report on diagnostic DNA image cytometry Anal Cell Pathol 2001 23 89 95 11904464
Sudbo J Kildal W Risberg B Koppang HS Danielsen HE Reith A DNA content as a prognostic marker in patients with oral leukoplakia N Eng J Med 2001 344 1270 1278 10.1056/NEJM200104263441702
Ghadimi RM Sackett DI Difilippantonio MJ Schrock E Neumann T Jauho A Auer G Ried T Centrosome amplification and instability occurs exclusively in aneuploid, but not in diploid colorectal cancer cell lines, and correlates with numerical chromosomal aberrations Genes Chromosomes Cancer 2000 27 183 190 10612807 10.1002/(SICI)1098-2264(200002)27:2<183::AID-GCC10>3.0.CO;2-P
Pindur A Chakraborty S Welch DG Wheeler TM DNA ploidy measurements in prostate cancer: differences between image analysis and flow cytometry and clinical implications Prostate 1994 25 189 198 8084836
Alanen KA Lintu M Joensuu H Image cytometry of breast carcinomas that are DNA diploid by flow cytometry: time to revise the concept of DNA diploidy? Anal Quant Cytol Histol 1998 20 178 186 9642444
Fang M Lew E Klein M Sebo T Su Y Goyal R DNA abnormalities as marker of risk for progression of Barrett's esophagus to adenocarcinoma: image cytometric DNA analysis in formalin-fixed tissues Am J Gastroenterol 2004 99 1887 1894 15447746 10.1111/j.1572-0241.2004.30886.x
James PD Atkinson M Value of DNA image cytometry in the prediction of malignant change in Barrett's oesophagus Gut 1989 30 899 905 2759487
Khan M Bui HX del Rosario A Abdulla M Ballouk F Sim YJ Ross JS Role of DNA content determination by image analysis in confirmation of dysplasia in Barrett's esophagus Mod Pathol 1994 7 169 174 8008738
Haggitt RC Barrett's esophagus, dysplasia, and adenocarcinoma Hum Pathol 1994 25 982 993 7927321 10.1016/0046-8177(94)90057-4
Goldblum JR Barrett's esophagus and Barrett's-related dysplasia Mod Pathol 2003 16 316 324 12692197 10.1097/01.MP.0000062996.66432.12
Bacus JW Boone CW Bacus JV Follen M Kelloff GJ Kagan V Lippman SM Image morphmetric nuclear grading of intraepithelial neoplastic lesions with applications to cancer chemoprevention trials Cancer Epidemiol Biomarkers Prev 1999 8 1087 1094 10613341
Reid BJ Levine DS Longton G Blount PL Rabinovitch PS Predictors of progression to cancer in Barrett's esophagus: baseline histology and flow cytometry identify low- and high-risk patient subsets Am J Gastroenterol 2000 95 1669 1676 10925966
Kay PA Riehle DL Cheville JC Lohse CM Pankratz VS Hill EM Sebo TJ Comparison of quantitative histomorphometry and DNA ploidy in tissue sections of prostate carcinoma Anal Quant Cytol Histol 2002 24 7 14 11865951
Kristensen GB Kildal W Abeler VM Kaern J Vergote I Trope CG Danielsen HE Large-scale genomic instability predicts long-term outcome for women with invasive stage I ovarian cancer Ann Oncol 2003 14 1494 1500 14504048 10.1093/annonc/mdg403
Mainguene C Choquenet C Deplano C Gavelli A Clement N Vanzo E Hofman P DNA ploidy by image cytometry in urothelial carcinomas. Comparison of touch imprints and paraffin-embedded biopsies from 31 patients Anal Quant Cytol Histol 1997 19 437 442 9349904
Berchuck A Boente MP Kerns BJ Kinney RB Soper JT Clarke-Pearson DL Bast RC JrBacus SS Ploidy analysis of epithelial ovarian cancers using image cytometry Gynecol Oncol 1992 44 61 65 1370427 10.1016/0090-8258(92)90013-9
Yamamoto T Horiguchi H Kamma H Noro M Ogata T Inage Y Akaogi E Mitsui K Hori M Isobe M Comparative DNA analysis by image cytometry and flow cytometry in non-small cell lung cancer Jpn J Cancer Res 1994 85 1171 1177 7829404
Talve LA Collan YU Ekfors TO Primary malignant melanoma of the skin. Relationships of nuclear DNA content, nuclear morphometric variables, Clark level and tumor thickness Anal Quant Cytol Histol 1997 19 62 74 9051188
Dreinhofer KE Baldetorp B Akerman M Ferno M Rydholm A Gustafson P DNA ploidy in soft tissue sarcoma: comparison of flow and image cytometry with clinical follow-up in 93 patients Cytometry 2002 50 19 24 11857594 10.1002/cyto.10034
Reles AE Gee C Schellschmidt I Schmider A Unger M Friedmann W Lichtenegger W Press MF Prognostic significance of DNA content and S-phase fraction in epithelial ovarian carcinomas analyzed by image cytometry Gynecol Oncol 1998 71 3 13 9784312 10.1006/gyno.1998.5156
Davey DD Zaleski S Sattich M Gallion H Prognostic significance of DNA cytometry of postirradiation cervicovaginal smears Cancer 1998 84 11 16 9500647 10.1002/(SICI)1097-0142(19980225)84:1<11::AID-CNCR3>3.0.CO;2-I
Baak JP Bol MG van Diermen B Janssen EA Buhr-Wildhagen SB Mestad O Ogreid P Kjellevold KH DNA cytometric features in biopsies of TaT1 urothelial cell cancer predict recurrence and stage progression more accurately than stage, grade, or treatment modality Urology 2003 61 1266 1272 12809927 10.1016/S0090-4295(03)00024-4
Keller R Foerster EC Kohler A Floer B Winde G Terpe HJ Domschke W Diagnostic value of DNA image cytometry in ulcerative colitis Dig Dis Sci 2001 46 870 878 11330427 10.1023/A:1010777123250
Mairinger T Gschwendtner A Ploidy determination in prostatic adenocarcinoma using thin histological sections: a more sensitive way of detecting aneuploid tumor clones Prostate 1998 37 30 35 9721066 10.1002/(SICI)1097-0045(19980915)37:1<30::AID-PROS5>3.0.CO;2-C
Gschwendtner A Fend F Hoffmann Y Krugmann J Klingler PJ Mairinger T DNA-ploidy analysis correlates with the histogenetic classification of thymic epithelial tumours J Pathol 1999 189 576 580 10629561 10.1002/(SICI)1096-9896(199912)189:4<576::AID-PATH455>3.0.CO;2-L
Jiang F Caraway NP Sabichi AL Zhang HZ Ruitrok A Grossman HB Gu J Lerner SP Lippman S Katz RL Centrosomal abnormality is common in and a potential biomarker for bladder cancer Int J Cancer 2003 106 661 665 12866024 10.1002/ijc.11251
Haroske G Giroud F Reith A Bocking A 1997 ESACP consensus report on diagnostic DNA image cytometry. Part I: basic considerations and recommendations for preparation, measurement and interpretation. European Society for Analytical Cellular Pathology Anal Cell Pathol 1998 17 189 200 10391371
|
16095543
|
PMC1201134
|
CC BY
|
2021-01-04 16:27:55
|
no
|
BMC Clin Pathol. 2005 Aug 12; 5:7
|
utf-8
|
BMC Clin Pathol
| 2,005 |
10.1186/1472-6890-5-7
|
oa_comm
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.